sha
stringlengths 40
40
| text
stringlengths 1
13.4M
| id
stringlengths 2
117
| tags
listlengths 1
7.91k
| created_at
stringlengths 25
25
| metadata
stringlengths 2
875k
| last_modified
stringlengths 25
25
| arxiv
listlengths 0
25
| languages
listlengths 0
7.91k
| tags_str
stringlengths 17
159k
| text_str
stringlengths 1
447k
| text_lists
listlengths 0
352
| processed_texts
listlengths 1
353
| tokens_length
listlengths 1
353
| input_texts
listlengths 1
40
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
4214dd721309dbcf891e27c4238e9e6f0bf604be
|
# Dataset Card for "coig-pc-zhtw"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
erhwenkuo/coig-pc-zhtw
|
[
"region:us"
] |
2023-10-26T04:27:34+00:00
|
{"dataset_info": {"features": [{"name": "instruction", "dtype": "string"}, {"name": "input", "dtype": "string"}, {"name": "output", "dtype": "string"}, {"name": "split", "dtype": "string"}, {"name": "task_name_in_eng", "dtype": "string"}, {"name": "task_type", "struct": [{"name": "major", "sequence": "string"}, {"name": "minor", "sequence": "string"}]}, {"name": "domain", "sequence": "string"}, {"name": "other", "dtype": "string"}, {"name": "filename", "dtype": "string"}], "splits": [{"name": "valid", "num_bytes": 2240096565, "num_examples": 2087767}], "download_size": 1068284005, "dataset_size": 2240096565}, "configs": [{"config_name": "default", "data_files": [{"split": "valid", "path": "data/valid-*"}]}]}
|
2023-10-26T04:33:26+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "coig-pc-zhtw"
More Information needed
|
[
"# Dataset Card for \"coig-pc-zhtw\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"coig-pc-zhtw\"\n\nMore Information needed"
] |
[
6,
17
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"coig-pc-zhtw\"\n\nMore Information needed"
] |
551dc4f499eddda7981e2e12088aaed7d1c66300
|
# Dataset Card for "mmlu_1pc"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
imdatta0/mmlu_sample
|
[
"region:us"
] |
2023-10-26T04:32:56+00:00
|
{"dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train_1pc", "num_bytes": 76328814, "num_examples": 56886}, {"name": "train_5pc", "num_bytes": 585203496, "num_examples": 284544}], "download_size": 201927295, "dataset_size": 661532310}, "configs": [{"config_name": "default", "data_files": [{"split": "train_1pc", "path": "data/train_1pc-*"}, {"split": "train_5pc", "path": "data/train_5pc-*"}]}]}
|
2023-10-26T04:47:36+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "mmlu_1pc"
More Information needed
|
[
"# Dataset Card for \"mmlu_1pc\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"mmlu_1pc\"\n\nMore Information needed"
] |
[
6,
14
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"mmlu_1pc\"\n\nMore Information needed"
] |
c0c4234e2bf77e5e77bec868a21c645184950805
|
# Dataset Card for "Alpaca_Instruct_Processed_train_ready"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
PiyushLavaniya/Alpaca_Instruct_Processed_train_ready
|
[
"region:us"
] |
2023-10-26T04:35:51+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "input", "dtype": "string"}, {"name": "output", "dtype": "string"}, {"name": "input_ids", "sequence": "int32"}, {"name": "attention_mask", "sequence": "int8"}, {"name": "labels", "sequence": "int64"}], "splits": [{"name": "train", "num_bytes": 93680964.0, "num_examples": 46800}, {"name": "test", "num_bytes": 10408996.0, "num_examples": 5200}], "download_size": 32202704, "dataset_size": 104089960.0}}
|
2023-10-26T04:36:05+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "Alpaca_Instruct_Processed_train_ready"
More Information needed
|
[
"# Dataset Card for \"Alpaca_Instruct_Processed_train_ready\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"Alpaca_Instruct_Processed_train_ready\"\n\nMore Information needed"
] |
[
6,
26
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"Alpaca_Instruct_Processed_train_ready\"\n\nMore Information needed"
] |
e151b7738cdac24b1c38c55276647b5ec49daeca
|
# Dataset Card for "oaast_seed"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
ycchen/oaast_seed
|
[
"region:us"
] |
2023-10-26T04:54:53+00:00
|
{"dataset_info": {"features": [{"name": "messages", "list": [{"name": "content", "dtype": "string"}, {"name": "role", "dtype": "string"}]}], "splits": [{"name": "train", "num_bytes": 4258491, "num_examples": 3359}], "download_size": 2403423, "dataset_size": 4258491}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T10:09:13+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "oaast_seed"
More Information needed
|
[
"# Dataset Card for \"oaast_seed\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"oaast_seed\"\n\nMore Information needed"
] |
[
6,
16
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"oaast_seed\"\n\nMore Information needed"
] |
b8df14ed826f331722418bd2037059427bb68a3c
|
---
# Information
This is a clean version of the union of BigVul and CVE-Fixes used
[here](https://huggingface.co/datasets/MickyMike/cvefixes_bigvul).
We have
four splits:
- `train`, which has the de-duplicated training data;
- `cleantest`, which has de-duplicated testing data that is completely disjoint from the
training set;
- `test`, which has the deduplicated training data with a
significant intersection with the training data (as seen in the original
repository);
- `output`, which is [VulRepair](https://github.com/awsm-research/VulRepair/tree/main)'s output on the data found in the `test` split.
Our preprocessing is available in `preprocessing.ipynb`.
|
nus-yam/vulrepair
|
[
"region:us"
] |
2023-10-26T04:55:10+00:00
|
{"pretty_name": "BigFixes", "description": "A clean union of BigVul and CVE-Fixes.", "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "train.csv"}, {"split": "cleantest", "path": "clean_test.csv"}, {"split": "test", "path": "test.csv"}, {"split": "output", "path": "output.csv"}]}]}
|
2023-11-16T08:44:09+00:00
|
[] |
[] |
TAGS
#region-us
|
---
# Information
This is a clean version of the union of BigVul and CVE-Fixes used
here.
We have
four splits:
- 'train', which has the de-duplicated training data;
- 'cleantest', which has de-duplicated testing data that is completely disjoint from the
training set;
- 'test', which has the deduplicated training data with a
significant intersection with the training data (as seen in the original
repository);
- 'output', which is VulRepair's output on the data found in the 'test' split.
Our preprocessing is available in 'URL'.
|
[
"# Information\n\nThis is a clean version of the union of BigVul and CVE-Fixes used\n here.\n We have\n four splits: \n - 'train', which has the de-duplicated training data;\n - 'cleantest', which has de-duplicated testing data that is completely disjoint from the\n training set;\n - 'test', which has the deduplicated training data with a\n significant intersection with the training data (as seen in the original\n repository);\n - 'output', which is VulRepair's output on the data found in the 'test' split.\nOur preprocessing is available in 'URL'."
] |
[
"TAGS\n#region-us \n",
"# Information\n\nThis is a clean version of the union of BigVul and CVE-Fixes used\n here.\n We have\n four splits: \n - 'train', which has the de-duplicated training data;\n - 'cleantest', which has de-duplicated testing data that is completely disjoint from the\n training set;\n - 'test', which has the deduplicated training data with a\n significant intersection with the training data (as seen in the original\n repository);\n - 'output', which is VulRepair's output on the data found in the 'test' split.\nOur preprocessing is available in 'URL'."
] |
[
6,
143
] |
[
"passage: TAGS\n#region-us \n# Information\n\nThis is a clean version of the union of BigVul and CVE-Fixes used\n here.\n We have\n four splits: \n - 'train', which has the de-duplicated training data;\n - 'cleantest', which has de-duplicated testing data that is completely disjoint from the\n training set;\n - 'test', which has the deduplicated training data with a\n significant intersection with the training data (as seen in the original\n repository);\n - 'output', which is VulRepair's output on the data found in the 'test' split.\nOur preprocessing is available in 'URL'."
] |
2c46c16a1e0d59fdf827a30a4abb79ef94e0a3a5
|
# Dataset Card for "rsna_5k_ii"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Phaedrus/rsna_5k_ii
|
[
"region:us"
] |
2023-10-26T04:55:57+00:00
|
{"dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "label1", "dtype": "image"}, {"name": "label2", "dtype": "image"}, {"name": "label3", "dtype": "image"}, {"name": "label4", "dtype": "image"}], "splits": [{"name": "train", "num_bytes": 5400048146.0, "num_examples": 5000}], "download_size": 366351514, "dataset_size": 5400048146.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T04:59:24+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "rsna_5k_ii"
More Information needed
|
[
"# Dataset Card for \"rsna_5k_ii\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"rsna_5k_ii\"\n\nMore Information needed"
] |
[
6,
17
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"rsna_5k_ii\"\n\nMore Information needed"
] |
9b1a57a80834a006b50262b1f04d161f06ad0ac2
|
# Dataset Card for Dataset Name
<!-- Provide a quick summary of the dataset. -->
This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1).
## Dataset Details
### Dataset Description
<!-- Provide a longer summary of what this dataset is. -->
- **Curated by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
### Dataset Sources [optional]
<!-- Provide the basic links for the dataset. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the dataset is intended to be used. -->
### Direct Use
<!-- This section describes suitable use cases for the dataset. -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
[More Information Needed]
## Dataset Structure
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
[More Information Needed]
## Dataset Creation
### Curation Rationale
<!-- Motivation for the creation of this dataset. -->
[More Information Needed]
### Source Data
<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
#### Data Collection and Processing
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
[More Information Needed]
#### Who are the source data producers?
<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
[More Information Needed]
### Annotations [optional]
<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
#### Annotation process
<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
[More Information Needed]
#### Who are the annotators?
<!-- This section describes the people or systems who created the annotations. -->
[More Information Needed]
#### Personal and Sensitive Information
<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
## Citation [optional]
<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Dataset Card Authors [optional]
[More Information Needed]
## Dataset Card Contact
[More Information Needed]
|
zhaospei/cmg-codellama
|
[
"task_categories:text2text-generation",
"size_categories:10K<n<100K",
"language:en",
"license:mit",
"region:us"
] |
2023-10-26T05:20:29+00:00
|
{"language": ["en"], "license": "mit", "size_categories": ["10K<n<100K"], "task_categories": ["text2text-generation"]}
|
2023-10-26T05:24:04+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-text2text-generation #size_categories-10K<n<100K #language-English #license-mit #region-us
|
# Dataset Card for Dataset Name
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
## Dataset Details
### Dataset Description
- Curated by:
- Funded by [optional]:
- Shared by [optional]:
- Language(s) (NLP):
- License:
### Dataset Sources [optional]
- Repository:
- Paper [optional]:
- Demo [optional]:
## Uses
### Direct Use
### Out-of-Scope Use
## Dataset Structure
## Dataset Creation
### Curation Rationale
### Source Data
#### Data Collection and Processing
#### Who are the source data producers?
### Annotations [optional]
#### Annotation process
#### Who are the annotators?
#### Personal and Sensitive Information
## Bias, Risks, and Limitations
### Recommendations
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
[optional]
BibTeX:
APA:
## Glossary [optional]
## More Information [optional]
## Dataset Card Authors [optional]
## Dataset Card Contact
|
[
"# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.",
"## Dataset Details",
"### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:",
"### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:",
"## Uses",
"### Direct Use",
"### Out-of-Scope Use",
"## Dataset Structure",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Data Collection and Processing",
"#### Who are the source data producers?",
"### Annotations [optional]",
"#### Annotation process",
"#### Who are the annotators?",
"#### Personal and Sensitive Information",
"## Bias, Risks, and Limitations",
"### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:",
"## Glossary [optional]",
"## More Information [optional]",
"## Dataset Card Authors [optional]",
"## Dataset Card Contact"
] |
[
"TAGS\n#task_categories-text2text-generation #size_categories-10K<n<100K #language-English #license-mit #region-us \n",
"# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.",
"## Dataset Details",
"### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:",
"### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:",
"## Uses",
"### Direct Use",
"### Out-of-Scope Use",
"## Dataset Structure",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Data Collection and Processing",
"#### Who are the source data producers?",
"### Annotations [optional]",
"#### Annotation process",
"#### Who are the annotators?",
"#### Personal and Sensitive Information",
"## Bias, Risks, and Limitations",
"### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:",
"## Glossary [optional]",
"## More Information [optional]",
"## Dataset Card Authors [optional]",
"## Dataset Card Contact"
] |
[
40,
34,
4,
40,
29,
3,
4,
9,
6,
5,
7,
4,
7,
10,
9,
5,
9,
8,
10,
46,
8,
7,
10,
5
] |
[
"passage: TAGS\n#task_categories-text2text-generation #size_categories-10K<n<100K #language-English #license-mit #region-us \n# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.## Dataset Details### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:## Uses### Direct Use### Out-of-Scope Use## Dataset Structure## Dataset Creation### Curation Rationale### Source Data#### Data Collection and Processing#### Who are the source data producers?### Annotations [optional]#### Annotation process#### Who are the annotators?#### Personal and Sensitive Information## Bias, Risks, and Limitations### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:## Glossary [optional]## More Information [optional]## Dataset Card Authors [optional]## Dataset Card Contact"
] |
24e8cc039899260da809b7c41b83d0faaef1d0a8
|
# Campus dataset for dynamic novel view synthesis
From Paper [NeurIPS 2023 spotlight] [MSTH: Masked Space-Time Hash Encoding for Efficient Dynamic Scene Reconstruction]()
[Paper]() | [Project Page](https://masked-spacetime-hashing.github.io) | [Code](https://github.com/masked-spacetime-hashing/msth)
## Dataset Description
Each scene contains:
- `<scene_name>_<camera_id>.mp4`: The multi-view videos collected for this scene.
- `transforms_<split>.json`: The camera poses and intrinsics estimated using COLMAP. The format is the same as [nerfstudio](https://docs.nerf.studio/quickstart/data_conventions.html).
- `colmap/`: The sparse point cloud and camera infos obtained during COLMAP pose estimation, could be used for methods starting with a sparse point cloud (like 3D Gaussian Splatting).
## Demos
|
masked-spacetime-hashing/Campus
|
[
"license:apache-2.0",
"novel view synthesis",
"dynamic",
"multi-view video",
"region:us"
] |
2023-10-26T05:25:12+00:00
|
{"license": "apache-2.0", "tags": ["novel view synthesis", "dynamic", "multi-view video"]}
|
2023-10-26T09:16:06+00:00
|
[] |
[] |
TAGS
#license-apache-2.0 #novel view synthesis #dynamic #multi-view video #region-us
|
# Campus dataset for dynamic novel view synthesis
From Paper [NeurIPS 2023 spotlight] [MSTH: Masked Space-Time Hash Encoding for Efficient Dynamic Scene Reconstruction]()
[Paper]() | Project Page | Code
## Dataset Description
Each scene contains:
- '<scene_name>_<camera_id>.mp4': The multi-view videos collected for this scene.
- 'transforms_<split>.json': The camera poses and intrinsics estimated using COLMAP. The format is the same as nerfstudio.
- 'colmap/': The sparse point cloud and camera infos obtained during COLMAP pose estimation, could be used for methods starting with a sparse point cloud (like 3D Gaussian Splatting).
## Demos
|
[
"# Campus dataset for dynamic novel view synthesis \nFrom Paper [NeurIPS 2023 spotlight] [MSTH: Masked Space-Time Hash Encoding for Efficient Dynamic Scene Reconstruction]()\n\n[Paper]() | Project Page | Code",
"## Dataset Description\nEach scene contains:\n- '<scene_name>_<camera_id>.mp4': The multi-view videos collected for this scene.\n- 'transforms_<split>.json': The camera poses and intrinsics estimated using COLMAP. The format is the same as nerfstudio.\n- 'colmap/': The sparse point cloud and camera infos obtained during COLMAP pose estimation, could be used for methods starting with a sparse point cloud (like 3D Gaussian Splatting).",
"## Demos"
] |
[
"TAGS\n#license-apache-2.0 #novel view synthesis #dynamic #multi-view video #region-us \n",
"# Campus dataset for dynamic novel view synthesis \nFrom Paper [NeurIPS 2023 spotlight] [MSTH: Masked Space-Time Hash Encoding for Efficient Dynamic Scene Reconstruction]()\n\n[Paper]() | Project Page | Code",
"## Dataset Description\nEach scene contains:\n- '<scene_name>_<camera_id>.mp4': The multi-view videos collected for this scene.\n- 'transforms_<split>.json': The camera poses and intrinsics estimated using COLMAP. The format is the same as nerfstudio.\n- 'colmap/': The sparse point cloud and camera infos obtained during COLMAP pose estimation, could be used for methods starting with a sparse point cloud (like 3D Gaussian Splatting).",
"## Demos"
] |
[
28,
59,
124,
3
] |
[
"passage: TAGS\n#license-apache-2.0 #novel view synthesis #dynamic #multi-view video #region-us \n# Campus dataset for dynamic novel view synthesis \nFrom Paper [NeurIPS 2023 spotlight] [MSTH: Masked Space-Time Hash Encoding for Efficient Dynamic Scene Reconstruction]()\n\n[Paper]() | Project Page | Code## Dataset Description\nEach scene contains:\n- '<scene_name>_<camera_id>.mp4': The multi-view videos collected for this scene.\n- 'transforms_<split>.json': The camera poses and intrinsics estimated using COLMAP. The format is the same as nerfstudio.\n- 'colmap/': The sparse point cloud and camera infos obtained during COLMAP pose estimation, could be used for methods starting with a sparse point cloud (like 3D Gaussian Splatting).## Demos"
] |
fcfed7a1bae17efd1e961cfe2e696753dad71957
|
# Bangkok Metropolitan Urban Issue Image (Traffy Fondue Issue Buckets)
BMA_Fondue_Image Dataset is an original raw dataset without frame labels.
|
AeraX-Valley/BMA_Fondue_Images
|
[
"language:th",
"language:en",
"license:apache-2.0",
"region:us"
] |
2023-10-26T05:40:00+00:00
|
{"language": ["th", "en"], "license": "apache-2.0", "pretty_name": "s"}
|
2023-12-07T11:04:28+00:00
|
[] |
[
"th",
"en"
] |
TAGS
#language-Thai #language-English #license-apache-2.0 #region-us
|
# Bangkok Metropolitan Urban Issue Image (Traffy Fondue Issue Buckets)
BMA_Fondue_Image Dataset is an original raw dataset without frame labels.
|
[
"# Bangkok Metropolitan Urban Issue Image (Traffy Fondue Issue Buckets)\nBMA_Fondue_Image Dataset is an original raw dataset without frame labels."
] |
[
"TAGS\n#language-Thai #language-English #license-apache-2.0 #region-us \n",
"# Bangkok Metropolitan Urban Issue Image (Traffy Fondue Issue Buckets)\nBMA_Fondue_Image Dataset is an original raw dataset without frame labels."
] |
[
23,
36
] |
[
"passage: TAGS\n#language-Thai #language-English #license-apache-2.0 #region-us \n# Bangkok Metropolitan Urban Issue Image (Traffy Fondue Issue Buckets)\nBMA_Fondue_Image Dataset is an original raw dataset without frame labels."
] |
edf14cbf60b67ab5089f6217704a2ca1b91cbb85
|
# Dataset Card for "banel_wit_postag_v1"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
fia24/banel_wit_postag_v1
|
[
"region:us"
] |
2023-10-26T05:57:33+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}, {"split": "val", "path": "data/val-*"}]}], "dataset_info": {"features": [{"name": "Inflected_Word", "dtype": "string"}, {"name": "Lemma", "dtype": "string"}, {"name": "POS", "dtype": "string"}, {"name": "pos_label", "dtype": {"class_label": {"names": {"0": "\u0985\u09a8\u09cd\u09af\u09be\u09a8\u09cd\u09af", "1": "\u0985\u09ac\u09cd\u09af\u09df", "2": "\u0985\u09ac\u09cd\u09af\u09df\u09c7\u09b0\u09ac\u09bf\u09b6\u09c7\u09b7\u09a3", "3": "\u0995\u09cd\u09b0\u09bf\u09df\u09be", "4": "\u0995\u09cd\u09b0\u09bf\u09df\u09be\u09ac\u09bf\u09b6\u09c7\u09b7\u09a3", "5": "\u09ac\u09bf\u09b6\u09c7\u09b7\u09a3", "6": "\u09ac\u09bf\u09b6\u09c7\u09b7\u09a3\u09c7\u09b0\u09ac\u09bf\u09b6\u09c7\u09b7\u09a3", "7": "\u09ac\u09bf\u09b6\u09c7\u09b7\u09cd\u09af", "8": "\u09b8\u09b0\u09cd\u09ac\u09a8\u09be\u09ae"}}}}], "splits": [{"name": "train", "num_bytes": 1380534.719008634, "num_examples": 17882}, {"name": "test", "num_bytes": 172624.74173489018, "num_examples": 2236}, {"name": "val", "num_bytes": 172547.53925647563, "num_examples": 2235}], "download_size": 0, "dataset_size": 1725707.0}}
|
2023-10-26T05:58:15+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "banel_wit_postag_v1"
More Information needed
|
[
"# Dataset Card for \"banel_wit_postag_v1\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"banel_wit_postag_v1\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"banel_wit_postag_v1\"\n\nMore Information needed"
] |
910dbafe97e4e481fec144cf3ea254e8b44984a3
|
# Dataset Card for Dataset Name
<!-- Provide a quick summary of the dataset. -->
This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1).
## Dataset Details
### Dataset Description
<!-- Provide a longer summary of what this dataset is. -->
- **Curated by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
### Dataset Sources [optional]
<!-- Provide the basic links for the dataset. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the dataset is intended to be used. -->
### Direct Use
<!-- This section describes suitable use cases for the dataset. -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the dataset will not work well for. -->
[More Information Needed]
## Dataset Structure
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
[More Information Needed]
## Dataset Creation
### Curation Rationale
<!-- Motivation for the creation of this dataset. -->
[More Information Needed]
### Source Data
<!-- This section describes the source data (e.g. news text and headlines, social media posts, translated sentences, ...). -->
#### Data Collection and Processing
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, tools and libraries used, etc. -->
[More Information Needed]
#### Who are the source data producers?
<!-- This section describes the people or systems who originally created the data. It should also include self-reported demographic or identity information for the source data creators if this information is available. -->
[More Information Needed]
### Annotations [optional]
<!-- If the dataset contains annotations which are not part of the initial data collection, use this section to describe them. -->
#### Annotation process
<!-- This section describes the annotation process such as annotation tools used in the process, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
[More Information Needed]
#### Who are the annotators?
<!-- This section describes the people or systems who created the annotations. -->
[More Information Needed]
#### Personal and Sensitive Information
<!-- State whether the dataset contains data that might be considered personal, sensitive, or private (e.g., data that reveals addresses, uniquely identifiable names or aliases, racial or ethnic origins, sexual orientations, religious beliefs, political opinions, financial or health data, etc.). If efforts were made to anonymize the data, describe the anonymization process. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
## Citation [optional]
<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the dataset or dataset card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Dataset Card Authors [optional]
[More Information Needed]
## Dataset Card Contact
[More Information Needed]
|
KevinTao511/pets
|
[
"task_categories:image-classification",
"size_categories:n<1K",
"language:en",
"license:mit",
"pets",
"region:us"
] |
2023-10-26T06:08:59+00:00
|
{"language": ["en"], "license": "mit", "size_categories": ["n<1K"], "task_categories": ["image-classification"], "pretty_name": "tao-ai-pets", "dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "label", "dtype": {"class_label": {"names": {"0": "abyssinian", "1": "basset", "2": "beagle"}}}}], "splits": [{"name": "train", "num_bytes": 32892148, "num_examples": 289}], "download_size": 32848292, "dataset_size": 32892148}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "tags": ["pets"]}
|
2023-10-26T06:32:40+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-image-classification #size_categories-n<1K #language-English #license-mit #pets #region-us
|
# Dataset Card for Dataset Name
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
## Dataset Details
### Dataset Description
- Curated by:
- Funded by [optional]:
- Shared by [optional]:
- Language(s) (NLP):
- License:
### Dataset Sources [optional]
- Repository:
- Paper [optional]:
- Demo [optional]:
## Uses
### Direct Use
### Out-of-Scope Use
## Dataset Structure
## Dataset Creation
### Curation Rationale
### Source Data
#### Data Collection and Processing
#### Who are the source data producers?
### Annotations [optional]
#### Annotation process
#### Who are the annotators?
#### Personal and Sensitive Information
## Bias, Risks, and Limitations
### Recommendations
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
[optional]
BibTeX:
APA:
## Glossary [optional]
## More Information [optional]
## Dataset Card Authors [optional]
## Dataset Card Contact
|
[
"# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.",
"## Dataset Details",
"### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:",
"### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:",
"## Uses",
"### Direct Use",
"### Out-of-Scope Use",
"## Dataset Structure",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Data Collection and Processing",
"#### Who are the source data producers?",
"### Annotations [optional]",
"#### Annotation process",
"#### Who are the annotators?",
"#### Personal and Sensitive Information",
"## Bias, Risks, and Limitations",
"### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:",
"## Glossary [optional]",
"## More Information [optional]",
"## Dataset Card Authors [optional]",
"## Dataset Card Contact"
] |
[
"TAGS\n#task_categories-image-classification #size_categories-n<1K #language-English #license-mit #pets #region-us \n",
"# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.",
"## Dataset Details",
"### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:",
"### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:",
"## Uses",
"### Direct Use",
"### Out-of-Scope Use",
"## Dataset Structure",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Data Collection and Processing",
"#### Who are the source data producers?",
"### Annotations [optional]",
"#### Annotation process",
"#### Who are the annotators?",
"#### Personal and Sensitive Information",
"## Bias, Risks, and Limitations",
"### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:",
"## Glossary [optional]",
"## More Information [optional]",
"## Dataset Card Authors [optional]",
"## Dataset Card Contact"
] |
[
39,
34,
4,
40,
29,
3,
4,
9,
6,
5,
7,
4,
7,
10,
9,
5,
9,
8,
10,
46,
8,
7,
10,
5
] |
[
"passage: TAGS\n#task_categories-image-classification #size_categories-n<1K #language-English #license-mit #pets #region-us \n# Dataset Card for Dataset Name\n\n\n\nThis dataset card aims to be a base template for new datasets. It has been generated using this raw template.## Dataset Details### Dataset Description\n\n\n\n\n\n- Curated by: \n- Funded by [optional]: \n- Shared by [optional]: \n- Language(s) (NLP): \n- License:### Dataset Sources [optional]\n\n\n\n- Repository: \n- Paper [optional]: \n- Demo [optional]:## Uses### Direct Use### Out-of-Scope Use## Dataset Structure## Dataset Creation### Curation Rationale### Source Data#### Data Collection and Processing#### Who are the source data producers?### Annotations [optional]#### Annotation process#### Who are the annotators?#### Personal and Sensitive Information## Bias, Risks, and Limitations### Recommendations\n\n\n\nUsers should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.\n\n[optional]\n\n\n\nBibTeX:\n\n\n\nAPA:## Glossary [optional]## More Information [optional]## Dataset Card Authors [optional]## Dataset Card Contact"
] |
3ba0299353f5a58eaec8fb5b180060eed5555f70
|
# Dataset Card for "SOPHIAE"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
GHOFRANEE/SOPHIAE
|
[
"region:us"
] |
2023-10-26T06:13:58+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "validation", "path": "data/validation-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "ground_truth", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 6853714.0, "num_examples": 94}, {"name": "validation", "num_bytes": 6853714.0, "num_examples": 94}, {"name": "test", "num_bytes": 6853714.0, "num_examples": 94}], "download_size": 5009808, "dataset_size": 20561142.0}}
|
2023-10-26T08:04:38+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "SOPHIAE"
More Information needed
|
[
"# Dataset Card for \"SOPHIAE\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"SOPHIAE\"\n\nMore Information needed"
] |
[
6,
13
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"SOPHIAE\"\n\nMore Information needed"
] |
c912832617c591827daf42e5cf897f3a209532da
|
# Dataset Card for "lj_speech_DifferentStructure_removedVocabs"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
HamdanXI/lj_speech_DifferentStructure_removedVocabs
|
[
"region:us"
] |
2023-10-26T06:22:30+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "audio", "dtype": {"audio": {"sampling_rate": 22050}}}, {"name": "file", "dtype": "string"}, {"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 1347808036.0, "num_examples": 4620}, {"name": "test", "num_bytes": 487719584.0, "num_examples": 1680}], "download_size": 1828316030, "dataset_size": 1835527620.0}}
|
2023-10-26T06:24:35+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "lj_speech_DifferentStructure_removedVocabs"
More Information needed
|
[
"# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs\"\n\nMore Information needed"
] |
[
6,
27
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs\"\n\nMore Information needed"
] |
bc74b337350d22ba19eb719338fd0bb8ad7170af
|
# Dataset Card for Evaluation run of TigerResearch/tigerbot-7b-base
## Dataset Description
- **Homepage:**
- **Repository:** https://huggingface.co/TigerResearch/tigerbot-7b-base
- **Paper:**
- **Leaderboard:** https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- **Point of Contact:** [email protected]
### Dataset Summary
Dataset automatically created during the evaluation run of model [TigerResearch/tigerbot-7b-base](https://huggingface.co/TigerResearch/tigerbot-7b-base) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).
To load the details from a run, you can for instance do the following:
```python
from datasets import load_dataset
data = load_dataset("open-llm-leaderboard/details_TigerResearch__tigerbot-7b-base",
"harness_winogrande_5",
split="train")
```
## Latest results
These are the [latest results from run 2023-10-26T07:32:19.507473](https://huggingface.co/datasets/open-llm-leaderboard/details_TigerResearch__tigerbot-7b-base/blob/main/results_2023-10-26T07-32-19.507473.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
```python
{
"all": {
"em": 0.008284395973154363,
"em_stderr": 0.0009282472025612859,
"f1": 0.0630316694630872,
"f1_stderr": 0.0015568932363703221,
"acc": 0.4022740314737014,
"acc_stderr": 0.010745030991055528
},
"harness|drop|3": {
"em": 0.008284395973154363,
"em_stderr": 0.0009282472025612859,
"f1": 0.0630316694630872,
"f1_stderr": 0.0015568932363703221
},
"harness|gsm8k|5": {
"acc": 0.10841546626231995,
"acc_stderr": 0.008563852506627483
},
"harness|winogrande|5": {
"acc": 0.6961325966850829,
"acc_stderr": 0.012926209475483574
}
}
```
### Supported Tasks and Leaderboards
[More Information Needed]
### Languages
[More Information Needed]
## Dataset Structure
### Data Instances
[More Information Needed]
### Data Fields
[More Information Needed]
### Data Splits
[More Information Needed]
## Dataset Creation
### Curation Rationale
[More Information Needed]
### Source Data
#### Initial Data Collection and Normalization
[More Information Needed]
#### Who are the source language producers?
[More Information Needed]
### Annotations
#### Annotation process
[More Information Needed]
#### Who are the annotators?
[More Information Needed]
### Personal and Sensitive Information
[More Information Needed]
## Considerations for Using the Data
### Social Impact of Dataset
[More Information Needed]
### Discussion of Biases
[More Information Needed]
### Other Known Limitations
[More Information Needed]
## Additional Information
### Dataset Curators
[More Information Needed]
### Licensing Information
[More Information Needed]
### Citation Information
[More Information Needed]
### Contributions
[More Information Needed]
|
open-llm-leaderboard/details_TigerResearch__tigerbot-7b-base
|
[
"region:us"
] |
2023-10-26T06:32:24+00:00
|
{"pretty_name": "Evaluation run of TigerResearch/tigerbot-7b-base", "dataset_summary": "Dataset automatically created during the evaluation run of model [TigerResearch/tigerbot-7b-base](https://huggingface.co/TigerResearch/tigerbot-7b-base) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).\n\nTo load the details from a run, you can for instance do the following:\n```python\nfrom datasets import load_dataset\ndata = load_dataset(\"open-llm-leaderboard/details_TigerResearch__tigerbot-7b-base\",\n\t\"harness_winogrande_5\",\n\tsplit=\"train\")\n```\n\n## Latest results\n\nThese are the [latest results from run 2023-10-26T07:32:19.507473](https://huggingface.co/datasets/open-llm-leaderboard/details_TigerResearch__tigerbot-7b-base/blob/main/results_2023-10-26T07-32-19.507473.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):\n\n```python\n{\n \"all\": {\n \"em\": 0.008284395973154363,\n \"em_stderr\": 0.0009282472025612859,\n \"f1\": 0.0630316694630872,\n \"f1_stderr\": 0.0015568932363703221,\n \"acc\": 0.4022740314737014,\n \"acc_stderr\": 0.010745030991055528\n },\n \"harness|drop|3\": {\n \"em\": 0.008284395973154363,\n \"em_stderr\": 0.0009282472025612859,\n \"f1\": 0.0630316694630872,\n \"f1_stderr\": 0.0015568932363703221\n },\n \"harness|gsm8k|5\": {\n \"acc\": 0.10841546626231995,\n \"acc_stderr\": 0.008563852506627483\n },\n \"harness|winogrande|5\": {\n \"acc\": 0.6961325966850829,\n \"acc_stderr\": 0.012926209475483574\n }\n}\n```", "repo_url": "https://huggingface.co/TigerResearch/tigerbot-7b-base", "leaderboard_url": "https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard", "point_of_contact": "[email protected]", "configs": [{"config_name": "harness_drop_3", "data_files": [{"split": "2023_10_26T07_32_19.507473", "path": ["**/details_harness|drop|3_2023-10-26T07-32-19.507473.parquet"]}, {"split": "latest", "path": ["**/details_harness|drop|3_2023-10-26T07-32-19.507473.parquet"]}]}, {"config_name": "harness_gsm8k_5", "data_files": [{"split": "2023_10_26T07_32_19.507473", "path": ["**/details_harness|gsm8k|5_2023-10-26T07-32-19.507473.parquet"]}, {"split": "latest", "path": ["**/details_harness|gsm8k|5_2023-10-26T07-32-19.507473.parquet"]}]}, {"config_name": "harness_winogrande_5", "data_files": [{"split": "2023_10_26T07_32_19.507473", "path": ["**/details_harness|winogrande|5_2023-10-26T07-32-19.507473.parquet"]}, {"split": "latest", "path": ["**/details_harness|winogrande|5_2023-10-26T07-32-19.507473.parquet"]}]}, {"config_name": "results", "data_files": [{"split": "2023_10_26T07_32_19.507473", "path": ["results_2023-10-26T07-32-19.507473.parquet"]}, {"split": "latest", "path": ["results_2023-10-26T07-32-19.507473.parquet"]}]}]}
|
2023-10-26T06:32:31+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for Evaluation run of TigerResearch/tigerbot-7b-base
## Dataset Description
- Homepage:
- Repository: URL
- Paper:
- Leaderboard: URL
- Point of Contact: clementine@URL
### Dataset Summary
Dataset automatically created during the evaluation run of model TigerResearch/tigerbot-7b-base on the Open LLM Leaderboard.
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).
To load the details from a run, you can for instance do the following:
## Latest results
These are the latest results from run 2023-10-26T07:32:19.507473(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
### Supported Tasks and Leaderboards
### Languages
## Dataset Structure
### Data Instances
### Data Fields
### Data Splits
## Dataset Creation
### Curation Rationale
### Source Data
#### Initial Data Collection and Normalization
#### Who are the source language producers?
### Annotations
#### Annotation process
#### Who are the annotators?
### Personal and Sensitive Information
## Considerations for Using the Data
### Social Impact of Dataset
### Discussion of Biases
### Other Known Limitations
## Additional Information
### Dataset Curators
### Licensing Information
### Contributions
|
[
"# Dataset Card for Evaluation run of TigerResearch/tigerbot-7b-base",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model TigerResearch/tigerbot-7b-base on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T07:32:19.507473(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for Evaluation run of TigerResearch/tigerbot-7b-base",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model TigerResearch/tigerbot-7b-base on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T07:32:19.507473(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
6,
19,
31,
167,
67,
10,
4,
6,
6,
5,
5,
5,
7,
4,
10,
10,
5,
5,
9,
8,
8,
7,
8,
7,
5,
6,
6,
5
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for Evaluation run of TigerResearch/tigerbot-7b-base## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL### Dataset Summary\n\nDataset automatically created during the evaluation run of model TigerResearch/tigerbot-7b-base on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:## Latest results\n\nThese are the latest results from run 2023-10-26T07:32:19.507473(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):### Supported Tasks and Leaderboards### Languages## Dataset Structure### Data Instances### Data Fields### Data Splits## Dataset Creation### Curation Rationale### Source Data#### Initial Data Collection and Normalization#### Who are the source language producers?### Annotations#### Annotation process#### Who are the annotators?### Personal and Sensitive Information## Considerations for Using the Data### Social Impact of Dataset### Discussion of Biases### Other Known Limitations## Additional Information### Dataset Curators### Licensing Information### Contributions"
] |
bc5f3b6a09b2a4b711ba3be04d71287a86192db0
|
# main
This is a benchmark for data fere testing.
## structure
a seed with some examples.
## others
version 1.0
|
mujif/VRP-test
|
[
"task_categories:question-answering",
"size_categories:n<1K",
"language:en",
"license:cc-by-4.0",
"region:us"
] |
2023-10-26T06:38:03+00:00
|
{"language": ["en"], "license": "cc-by-4.0", "size_categories": ["n<1K"], "task_categories": ["question-answering"]}
|
2023-10-26T11:46:38+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-question-answering #size_categories-n<1K #language-English #license-cc-by-4.0 #region-us
|
# main
This is a benchmark for data fere testing.
## structure
a seed with some examples.
## others
version 1.0
|
[
"# main\n\n\nThis is a benchmark for data fere testing.",
"## structure\n\n\na seed with some examples.",
"## others\n\n\nversion 1.0"
] |
[
"TAGS\n#task_categories-question-answering #size_categories-n<1K #language-English #license-cc-by-4.0 #region-us \n",
"# main\n\n\nThis is a benchmark for data fere testing.",
"## structure\n\n\na seed with some examples.",
"## others\n\n\nversion 1.0"
] |
[
41,
12,
10,
4
] |
[
"passage: TAGS\n#task_categories-question-answering #size_categories-n<1K #language-English #license-cc-by-4.0 #region-us \n# main\n\n\nThis is a benchmark for data fere testing.## structure\n\n\na seed with some examples.## others\n\n\nversion 1.0"
] |
81557922456ed1c135ce6196ad4fe56f6edbbd96
|
## Table of Contents
- [Dataset Summary](#dataset-summary)
- [Dataset Attribution](#dataset-attribution)
- [Languages](#languages)
- [Dataset Structure](#dataset-structure)
- [Data Fields](#data-fields)
- [Dataset Creation](#dataset-creation)
- [Curation Rationale](#curation-rationale)
- [Source Data](#source-data)
- [Dataset Use](#dataset-use)
- [Use Cases](#use-cases)
- [Getting Started](#getting-started)
**Disclaimer: this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.**
<a name="dataset-summary"></a>
# Dataset Summary
We curated this dataset to finetune open source base models as part of [NeurIPS 2023 LLM Efficiency Challenge](https://llm-efficiency-challenge.github.io/) (1 LLM + 1 GPU + 1 Day). This challenge requires participants to use open source models and datasets with permissible licenses to encourage wider adoption, use and dissemination of open source contributions in generative AI space. Additionally, LLM generated datasets such as Alpaca and Orca datasets are not allowed.
**Open-Otter** combines the non-LLM generated subset of [Open-Platypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus) datasets with other datasets, and is used for finetuning Llama-2-7b, Llama-2-13b and Mistral-7b-v0.1 base models to perform reasonably well in a suit of reasoning tasks selected by the organizers. Please visit the challenge website for more detailed information on the rules.
<a name="dataset-attribution"></a>
# Dataset Attribution
<a name="languages"></a>
# Languages
Evaluation for the challenge includes only English text. Therefore, Open-Otter includes data sources only in English.
_Note: we are not aware of any compelling literature demonstrating the value of finetuning on multilingual datasets (over datasets in target language). Please leave a comment if you come across any relevant work addressing this question._
<a name="dataset-structure"></a>
# Dataset Structure
<a name="data-fields"></a>
## Data Fields
Data fields follow Alpaca style formatting.
The fields are:
1) 'input', an optional field for providing additional context for response type
2) 'output', response, answer or solution to the corresponding instruction (e.g., a multiple choice question)
3) 'instruction', required field including the question and multiple choice options (when applicable)
4) 'data_source', original dataset and split for the data instance
<a name="dataset-creation"></a>
# Dataset Creation
<a name="curation-rationale"></a>
## Curation Rationale
TODO: NeurIPS 2023 LLM Efficiency Challenge
<a name="source-data"></a>
## Source Data
We have combined the non-LLM-generated subset of Open-Platypus dataset with 4 additional datasets:
- [Open-Platypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus) dataset
- Excluding [airoboros-gpt4-1.4.1](https://huggingface.co/datasets/jondurbin/airoboros-gpt4-1.4.1) and [PRM800K](https://github.com/openai/prm800k)
- [ARC](https://allenai.org/data/arc) (Allen AI Reasoning Challenge)
- [CommonsenseQA](https://huggingface.co/datasets/commonsense_qa)
- [WinoGrande, debiased](https://huggingface.co/datasets/winogrande)
- [MedMCQA](https://huggingface.co/datasets/medmcqa)
Notably, train, validation and test splits were all included for each dataset. If answer key is not provided, test set is excluded.
<a name="dataset-use"></a>
# Dataset Use
## Getting Started
You can use Hugging Face datasets library to load this dataset.
```python
from datasets import load_dataset
dataset = load_dataset("onuralp/open-otter")
```
# Citation
If you find this dataset useful for your own work and are interested in acknowledging, please use the citation below.
```bibtex
@misc {onuralp2023,
author = { {Onuralp Soylemez} },
title = { open-otter (Revision 17db84f) },
year = 2023,
url = { https://huggingface.co/datasets/onuralp/open-otter },
doi = { 10.57967/hf/1270 },
publisher = { Hugging Face }
}
```
|
onuralp/open-otter
|
[
"task_categories:multiple-choice",
"task_categories:question-answering",
"size_categories:100K<n<1M",
"language:en",
"license:mit",
"doi:10.57967/hf/1270",
"region:us"
] |
2023-10-26T06:53:53+00:00
|
{"language": ["en"], "license": "mit", "size_categories": ["100K<n<1M"], "task_categories": ["multiple-choice", "question-answering"], "pretty_name": "open-otter", "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "input", "dtype": "string"}, {"name": "output", "dtype": "string"}, {"name": "instruction", "dtype": "string"}, {"name": "data_source", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 48901728, "num_examples": 162864}], "download_size": 22256986, "dataset_size": 48901728}}
|
2023-10-31T10:45:59+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-multiple-choice #task_categories-question-answering #size_categories-100K<n<1M #language-English #license-mit #doi-10.57967/hf/1270 #region-us
|
## Table of Contents
- Dataset Summary
- Dataset Attribution
- Languages
- Dataset Structure
- Data Fields
- Dataset Creation
- Curation Rationale
- Source Data
- Dataset Use
- Use Cases
- Getting Started
!OpenOrca Logo
Disclaimer: this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.
<a name="dataset-summary"></a>
# Dataset Summary
We curated this dataset to finetune open source base models as part of NeurIPS 2023 LLM Efficiency Challenge (1 LLM + 1 GPU + 1 Day). This challenge requires participants to use open source models and datasets with permissible licenses to encourage wider adoption, use and dissemination of open source contributions in generative AI space. Additionally, LLM generated datasets such as Alpaca and Orca datasets are not allowed.
Open-Otter combines the non-LLM generated subset of Open-Platypus datasets with other datasets, and is used for finetuning Llama-2-7b, Llama-2-13b and Mistral-7b-v0.1 base models to perform reasonably well in a suit of reasoning tasks selected by the organizers. Please visit the challenge website for more detailed information on the rules.
<a name="dataset-attribution"></a>
# Dataset Attribution
<a name="languages"></a>
# Languages
Evaluation for the challenge includes only English text. Therefore, Open-Otter includes data sources only in English.
_Note: we are not aware of any compelling literature demonstrating the value of finetuning on multilingual datasets (over datasets in target language). Please leave a comment if you come across any relevant work addressing this question._
<a name="dataset-structure"></a>
# Dataset Structure
<a name="data-fields"></a>
## Data Fields
Data fields follow Alpaca style formatting.
The fields are:
1) 'input', an optional field for providing additional context for response type
2) 'output', response, answer or solution to the corresponding instruction (e.g., a multiple choice question)
3) 'instruction', required field including the question and multiple choice options (when applicable)
4) 'data_source', original dataset and split for the data instance
<a name="dataset-creation"></a>
# Dataset Creation
<a name="curation-rationale"></a>
## Curation Rationale
TODO: NeurIPS 2023 LLM Efficiency Challenge
<a name="source-data"></a>
## Source Data
We have combined the non-LLM-generated subset of Open-Platypus dataset with 4 additional datasets:
- Open-Platypus dataset
- Excluding airoboros-gpt4-1.4.1 and PRM800K
- ARC (Allen AI Reasoning Challenge)
- CommonsenseQA
- WinoGrande, debiased
- MedMCQA
Notably, train, validation and test splits were all included for each dataset. If answer key is not provided, test set is excluded.
<a name="dataset-use"></a>
# Dataset Use
## Getting Started
You can use Hugging Face datasets library to load this dataset.
If you find this dataset useful for your own work and are interested in acknowledging, please use the citation below.
|
[
"## Table of Contents\n- Dataset Summary\n- Dataset Attribution\n- Languages\n- Dataset Structure\n - Data Fields\n- Dataset Creation\n - Curation Rationale\n - Source Data\n- Dataset Use\n - Use Cases\n - Getting Started\n\n!OpenOrca Logo\n\n\nDisclaimer: this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.\n\n<a name=\"dataset-summary\"></a>",
"# Dataset Summary\n\nWe curated this dataset to finetune open source base models as part of NeurIPS 2023 LLM Efficiency Challenge (1 LLM + 1 GPU + 1 Day). This challenge requires participants to use open source models and datasets with permissible licenses to encourage wider adoption, use and dissemination of open source contributions in generative AI space. Additionally, LLM generated datasets such as Alpaca and Orca datasets are not allowed.\n\nOpen-Otter combines the non-LLM generated subset of Open-Platypus datasets with other datasets, and is used for finetuning Llama-2-7b, Llama-2-13b and Mistral-7b-v0.1 base models to perform reasonably well in a suit of reasoning tasks selected by the organizers. Please visit the challenge website for more detailed information on the rules.\n\n<a name=\"dataset-attribution\"></a>",
"# Dataset Attribution\n\n<a name=\"languages\"></a>",
"# Languages\n\nEvaluation for the challenge includes only English text. Therefore, Open-Otter includes data sources only in English. \n\n_Note: we are not aware of any compelling literature demonstrating the value of finetuning on multilingual datasets (over datasets in target language). Please leave a comment if you come across any relevant work addressing this question._ \n\n<a name=\"dataset-structure\"></a>",
"# Dataset Structure\n\n<a name=\"data-fields\"></a>",
"## Data Fields\n\nData fields follow Alpaca style formatting. \n\nThe fields are:\n1) 'input', an optional field for providing additional context for response type\n2) 'output', response, answer or solution to the corresponding instruction (e.g., a multiple choice question)\n3) 'instruction', required field including the question and multiple choice options (when applicable)\n4) 'data_source', original dataset and split for the data instance\n\n<a name=\"dataset-creation\"></a>",
"# Dataset Creation\n\n<a name=\"curation-rationale\"></a>",
"## Curation Rationale\n\nTODO: NeurIPS 2023 LLM Efficiency Challenge\n\n<a name=\"source-data\"></a>",
"## Source Data\n\nWe have combined the non-LLM-generated subset of Open-Platypus dataset with 4 additional datasets:\n- Open-Platypus dataset\n - Excluding airoboros-gpt4-1.4.1 and PRM800K\n- ARC (Allen AI Reasoning Challenge)\n- CommonsenseQA\n- WinoGrande, debiased\n- MedMCQA\n\nNotably, train, validation and test splits were all included for each dataset. If answer key is not provided, test set is excluded.\n\n<a name=\"dataset-use\"></a>",
"# Dataset Use",
"## Getting Started\n\nYou can use Hugging Face datasets library to load this dataset.\n\n\n\nIf you find this dataset useful for your own work and are interested in acknowledging, please use the citation below."
] |
[
"TAGS\n#task_categories-multiple-choice #task_categories-question-answering #size_categories-100K<n<1M #language-English #license-mit #doi-10.57967/hf/1270 #region-us \n",
"## Table of Contents\n- Dataset Summary\n- Dataset Attribution\n- Languages\n- Dataset Structure\n - Data Fields\n- Dataset Creation\n - Curation Rationale\n - Source Data\n- Dataset Use\n - Use Cases\n - Getting Started\n\n!OpenOrca Logo\n\n\nDisclaimer: this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.\n\n<a name=\"dataset-summary\"></a>",
"# Dataset Summary\n\nWe curated this dataset to finetune open source base models as part of NeurIPS 2023 LLM Efficiency Challenge (1 LLM + 1 GPU + 1 Day). This challenge requires participants to use open source models and datasets with permissible licenses to encourage wider adoption, use and dissemination of open source contributions in generative AI space. Additionally, LLM generated datasets such as Alpaca and Orca datasets are not allowed.\n\nOpen-Otter combines the non-LLM generated subset of Open-Platypus datasets with other datasets, and is used for finetuning Llama-2-7b, Llama-2-13b and Mistral-7b-v0.1 base models to perform reasonably well in a suit of reasoning tasks selected by the organizers. Please visit the challenge website for more detailed information on the rules.\n\n<a name=\"dataset-attribution\"></a>",
"# Dataset Attribution\n\n<a name=\"languages\"></a>",
"# Languages\n\nEvaluation for the challenge includes only English text. Therefore, Open-Otter includes data sources only in English. \n\n_Note: we are not aware of any compelling literature demonstrating the value of finetuning on multilingual datasets (over datasets in target language). Please leave a comment if you come across any relevant work addressing this question._ \n\n<a name=\"dataset-structure\"></a>",
"# Dataset Structure\n\n<a name=\"data-fields\"></a>",
"## Data Fields\n\nData fields follow Alpaca style formatting. \n\nThe fields are:\n1) 'input', an optional field for providing additional context for response type\n2) 'output', response, answer or solution to the corresponding instruction (e.g., a multiple choice question)\n3) 'instruction', required field including the question and multiple choice options (when applicable)\n4) 'data_source', original dataset and split for the data instance\n\n<a name=\"dataset-creation\"></a>",
"# Dataset Creation\n\n<a name=\"curation-rationale\"></a>",
"## Curation Rationale\n\nTODO: NeurIPS 2023 LLM Efficiency Challenge\n\n<a name=\"source-data\"></a>",
"## Source Data\n\nWe have combined the non-LLM-generated subset of Open-Platypus dataset with 4 additional datasets:\n- Open-Platypus dataset\n - Excluding airoboros-gpt4-1.4.1 and PRM800K\n- ARC (Allen AI Reasoning Challenge)\n- CommonsenseQA\n- WinoGrande, debiased\n- MedMCQA\n\nNotably, train, validation and test splits were all included for each dataset. If answer key is not provided, test set is excluded.\n\n<a name=\"dataset-use\"></a>",
"# Dataset Use",
"## Getting Started\n\nYou can use Hugging Face datasets library to load this dataset.\n\n\n\nIf you find this dataset useful for your own work and are interested in acknowledging, please use the citation below."
] |
[
63,
104,
211,
14,
92,
18,
112,
18,
32,
130,
4,
48
] |
[
"passage: TAGS\n#task_categories-multiple-choice #task_categories-question-answering #size_categories-100K<n<1M #language-English #license-mit #doi-10.57967/hf/1270 #region-us \n## Table of Contents\n- Dataset Summary\n- Dataset Attribution\n- Languages\n- Dataset Structure\n - Data Fields\n- Dataset Creation\n - Curation Rationale\n - Source Data\n- Dataset Use\n - Use Cases\n - Getting Started\n\n!OpenOrca Logo\n\n\nDisclaimer: this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.\n\n<a name=\"dataset-summary\"></a># Dataset Summary\n\nWe curated this dataset to finetune open source base models as part of NeurIPS 2023 LLM Efficiency Challenge (1 LLM + 1 GPU + 1 Day). This challenge requires participants to use open source models and datasets with permissible licenses to encourage wider adoption, use and dissemination of open source contributions in generative AI space. Additionally, LLM generated datasets such as Alpaca and Orca datasets are not allowed.\n\nOpen-Otter combines the non-LLM generated subset of Open-Platypus datasets with other datasets, and is used for finetuning Llama-2-7b, Llama-2-13b and Mistral-7b-v0.1 base models to perform reasonably well in a suit of reasoning tasks selected by the organizers. Please visit the challenge website for more detailed information on the rules.\n\n<a name=\"dataset-attribution\"></a># Dataset Attribution\n\n<a name=\"languages\"></a># Languages\n\nEvaluation for the challenge includes only English text. Therefore, Open-Otter includes data sources only in English. \n\n_Note: we are not aware of any compelling literature demonstrating the value of finetuning on multilingual datasets (over datasets in target language). Please leave a comment if you come across any relevant work addressing this question._ \n\n<a name=\"dataset-structure\"></a># Dataset Structure\n\n<a name=\"data-fields\"></a>"
] |
ffbdf7ed58cf9669b66bb8a94694d04d550c6d6c
|
# AutoTrain Dataset for project: coffee-bean-quality
## Dataset Description
This dataset has been automatically processed by AutoTrain for project coffee-bean-quality.
### Languages
The BCP-47 code for the dataset's language is unk.
## Dataset Structure
### Data Instances
A sample from this dataset looks as follows:
```json
[
{
"image": "<224x224 RGB PIL image>",
"feat_width": 224,
"feat_height": 224,
"target": 1,
"feat_xmin": 42,
"feat_ymin": 84,
"feat_xmax": 174,
"feat_ymax": 202
},
{
"image": "<224x224 RGB PIL image>",
"feat_width": 224,
"feat_height": 224,
"target": 0,
"feat_xmin": 80,
"feat_ymin": 90,
"feat_xmax": 186,
"feat_ymax": 150
}
]
```
### Dataset Fields
The dataset has the following fields (also called "features"):
```json
{
"image": "Image(decode=True, id=None)",
"feat_width": "Value(dtype='int64', id=None)",
"feat_height": "Value(dtype='int64', id=None)",
"target": "ClassLabel(names=['defect', 'good'], id=None)",
"feat_xmin": "Value(dtype='int64', id=None)",
"feat_ymin": "Value(dtype='int64', id=None)",
"feat_xmax": "Value(dtype='int64', id=None)",
"feat_ymax": "Value(dtype='int64', id=None)"
}
```
### Dataset Splits
This dataset is split into a train and validation split. The split sizes are as follow:
| Split name | Num samples |
| ------------ | ------------------- |
| train | 4185 |
| valid | 400 |
|
everycoffee/autotrain-data-coffee-bean-quality
|
[
"task_categories:image-classification",
"region:us"
] |
2023-10-26T06:56:42+00:00
|
{"task_categories": ["image-classification"]}
|
2023-10-26T23:34:12+00:00
|
[] |
[] |
TAGS
#task_categories-image-classification #region-us
|
AutoTrain Dataset for project: coffee-bean-quality
==================================================
Dataset Description
-------------------
This dataset has been automatically processed by AutoTrain for project coffee-bean-quality.
### Languages
The BCP-47 code for the dataset's language is unk.
Dataset Structure
-----------------
### Data Instances
A sample from this dataset looks as follows:
### Dataset Fields
The dataset has the following fields (also called "features"):
### Dataset Splits
This dataset is split into a train and validation split. The split sizes are as follow:
|
[
"### Languages\n\n\nThe BCP-47 code for the dataset's language is unk.\n\n\nDataset Structure\n-----------------",
"### Data Instances\n\n\nA sample from this dataset looks as follows:",
"### Dataset Fields\n\n\nThe dataset has the following fields (also called \"features\"):",
"### Dataset Splits\n\n\nThis dataset is split into a train and validation split. The split sizes are as follow:"
] |
[
"TAGS\n#task_categories-image-classification #region-us \n",
"### Languages\n\n\nThe BCP-47 code for the dataset's language is unk.\n\n\nDataset Structure\n-----------------",
"### Data Instances\n\n\nA sample from this dataset looks as follows:",
"### Dataset Fields\n\n\nThe dataset has the following fields (also called \"features\"):",
"### Dataset Splits\n\n\nThis dataset is split into a train and validation split. The split sizes are as follow:"
] |
[
17,
27,
17,
23,
27
] |
[
"passage: TAGS\n#task_categories-image-classification #region-us \n### Languages\n\n\nThe BCP-47 code for the dataset's language is unk.\n\n\nDataset Structure\n-----------------### Data Instances\n\n\nA sample from this dataset looks as follows:### Dataset Fields\n\n\nThe dataset has the following fields (also called \"features\"):### Dataset Splits\n\n\nThis dataset is split into a train and validation split. The split sizes are as follow:"
] |
bf0e4f0aa5f5451266c00b883f5979cd7cb491e8
|
# Dataset Card for "flickr8k"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Naveengo/flickr8k
|
[
"task_categories:image-to-text",
"license:apache-2.0",
"region:us"
] |
2023-10-26T07:02:48+00:00
|
{"license": "apache-2.0", "task_categories": ["image-to-text"], "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 1132031887.712, "num_examples": 8091}], "download_size": 1114562282, "dataset_size": 1132031887.712}}
|
2023-10-26T07:06:49+00:00
|
[] |
[] |
TAGS
#task_categories-image-to-text #license-apache-2.0 #region-us
|
# Dataset Card for "flickr8k"
More Information needed
|
[
"# Dataset Card for \"flickr8k\"\n\nMore Information needed"
] |
[
"TAGS\n#task_categories-image-to-text #license-apache-2.0 #region-us \n",
"# Dataset Card for \"flickr8k\"\n\nMore Information needed"
] |
[
26,
15
] |
[
"passage: TAGS\n#task_categories-image-to-text #license-apache-2.0 #region-us \n# Dataset Card for \"flickr8k\"\n\nMore Information needed"
] |
bfe2f64d86d37e57b868d176b349fc0747eb079d
|
# Dataset Card for "t2i_topic_comparision_db"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
toilaluan/t2i_topic_comparision_db
|
[
"region:us"
] |
2023-10-26T07:16:26+00:00
|
{"dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "topic", "dtype": "string"}, {"name": "prompt", "dtype": "string"}, {"name": "request_id", "dtype": "int64"}, {"name": "model_type", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 251342211.0, "num_examples": 5000}], "download_size": 489446445, "dataset_size": 251342211.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T13:39:42+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "t2i_topic_comparision_db"
More Information needed
|
[
"# Dataset Card for \"t2i_topic_comparision_db\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"t2i_topic_comparision_db\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"t2i_topic_comparision_db\"\n\nMore Information needed"
] |
6e455da43c5aac0cc0cd81b57bf5dfbf1b60f876
|
Information on the dataset:
```
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: validation
path: data/validation-*
dataset_info:
features:
- name: premise
dtype: string
- name: hypothesis
dtype: string
- name: label
dtype: int64
splits:
- name: train
num_bytes: 67523459
num_examples: 538896
- name: test
num_bytes: 1285789
num_examples: 9792
- name: validation
num_bytes: 1295645
num_examples: 9792
download_size: 20806553
dataset_size: 70104893
```
# Dataset Card for "SNLI_Dutch_translated_with_Marianmt"
Translation of the **English** corpus [Stanford Natural Language Inference (SNLI)](https://nlp.stanford.edu/projects/snli/),
to **Dutch** using an [Maria NMT model](https://marian-nmt.github.io/), trained by [Helsinki NLP](https://huggingface.co/Helsinki-NLP/opus-mt-en-nl).
Note, for reference: Maria NMT is based on [BART](https://huggingface.co/docs/transformers/model_doc/bart), described [here](https://arxiv.org/abs/1910.13461).
A complete description of the dataset is given [here](https://huggingface.co/datasets/snli).
# Attribution
If you use this dataset please use the following to credit the creators of SNLI:
```citation
@inproceedings{snli:emnlp2015,
Author = {Bowman, Samuel R. and Angeli, Gabor and Potts, Christopher, and Manning, Christopher D.},
Booktitle = {Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
Publisher = {Association for Computational Linguistics},
Title = {A large annotated corpus for learning natural language inference},
Year = {2015}
}
```
The creators of the OPUS-MT models:
```
@InProceedings{TiedemannThottingal:EAMT2020,
author = {J{\"o}rg Tiedemann and Santhosh Thottingal},
title = {{OPUS-MT} — {B}uilding open translation services for the {W}orld},
booktitle = {Proceedings of the 22nd Annual Conferenec of the European Association for Machine Translation (EAMT)},
year = {2020},
address = {Lisbon, Portugal}
}
```
and
```
@misc {van_es_2023,
author = { {Bram van Es} },
title = { SNLI_Dutch_translated_with_Marianmt (Revision 9ad7971) },
year = 2023,
url = { https://huggingface.co/datasets/UMCU/SNLI_Dutch_translated_with_Marianmt },
doi = { 10.57967/hf/1268 },
publisher = { Hugging Face }
}
```
# License
For both the Maria NMT model and the original [Helsinki NLP](https://twitter.com/HelsinkiNLP) [Opus MT model](https://huggingface.co/Helsinki-NLP)
we did **not** find a license, if this was in error please let us know and we will add the appropriate licensing promptly.
We adopt the licensing of the SNLI corpus: a [Creative Commons Attribution-ShareAlike 4.0 International License](http://creativecommons.org/licenses/by-sa/4.0/).
|
UMCU/SNLI_Dutch_translated_with_Marianmt
|
[
"task_categories:sentence-similarity",
"size_categories:100K<n<1M",
"language:nl",
"license:cc-by-sa-4.0",
"generic",
"sentence similarity",
"arxiv:1910.13461",
"doi:10.57967/hf/1268",
"region:us"
] |
2023-10-26T07:19:14+00:00
|
{"language": ["nl"], "license": "cc-by-sa-4.0", "size_categories": ["100K<n<1M"], "task_categories": ["sentence-similarity"], "pretty_name": "Dutch translation of SNLI corpus with Maria NMT", "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}, {"split": "validation", "path": "data/validation-*"}]}], "dataset_info": {"features": [{"name": "premise", "dtype": "string"}, {"name": "hypothesis", "dtype": "string"}, {"name": "label", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 67523459, "num_examples": 538896}, {"name": "test", "num_bytes": 1285789, "num_examples": 9792}, {"name": "validation", "num_bytes": 1295645, "num_examples": 9792}], "download_size": 20806553, "dataset_size": 70104893}, "tags": ["generic", "sentence similarity"]}
|
2023-11-17T09:58:10+00:00
|
[
"1910.13461"
] |
[
"nl"
] |
TAGS
#task_categories-sentence-similarity #size_categories-100K<n<1M #language-Dutch #license-cc-by-sa-4.0 #generic #sentence similarity #arxiv-1910.13461 #doi-10.57967/hf/1268 #region-us
|
Information on the dataset:
# Dataset Card for "SNLI_Dutch_translated_with_Marianmt"
Translation of the English corpus Stanford Natural Language Inference (SNLI),
to Dutch using an Maria NMT model, trained by Helsinki NLP.
Note, for reference: Maria NMT is based on BART, described here.
A complete description of the dataset is given here.
# Attribution
If you use this dataset please use the following to credit the creators of SNLI:
The creators of the OPUS-MT models:
and
# License
For both the Maria NMT model and the original Helsinki NLP Opus MT model
we did not find a license, if this was in error please let us know and we will add the appropriate licensing promptly.
We adopt the licensing of the SNLI corpus: a Creative Commons Attribution-ShareAlike 4.0 International License.
|
[
"# Dataset Card for \"SNLI_Dutch_translated_with_Marianmt\"\n\nTranslation of the English corpus Stanford Natural Language Inference (SNLI),\nto Dutch using an Maria NMT model, trained by Helsinki NLP.\nNote, for reference: Maria NMT is based on BART, described here.\n\nA complete description of the dataset is given here.",
"# Attribution\n\nIf you use this dataset please use the following to credit the creators of SNLI:\n\n\n\nThe creators of the OPUS-MT models:\n\n\nand",
"# License\n\nFor both the Maria NMT model and the original Helsinki NLP Opus MT model \nwe did not find a license, if this was in error please let us know and we will add the appropriate licensing promptly.\n\nWe adopt the licensing of the SNLI corpus: a Creative Commons Attribution-ShareAlike 4.0 International License."
] |
[
"TAGS\n#task_categories-sentence-similarity #size_categories-100K<n<1M #language-Dutch #license-cc-by-sa-4.0 #generic #sentence similarity #arxiv-1910.13461 #doi-10.57967/hf/1268 #region-us \n",
"# Dataset Card for \"SNLI_Dutch_translated_with_Marianmt\"\n\nTranslation of the English corpus Stanford Natural Language Inference (SNLI),\nto Dutch using an Maria NMT model, trained by Helsinki NLP.\nNote, for reference: Maria NMT is based on BART, described here.\n\nA complete description of the dataset is given here.",
"# Attribution\n\nIf you use this dataset please use the following to credit the creators of SNLI:\n\n\n\nThe creators of the OPUS-MT models:\n\n\nand",
"# License\n\nFor both the Maria NMT model and the original Helsinki NLP Opus MT model \nwe did not find a license, if this was in error please let us know and we will add the appropriate licensing promptly.\n\nWe adopt the licensing of the SNLI corpus: a Creative Commons Attribution-ShareAlike 4.0 International License."
] |
[
77,
83,
33,
69
] |
[
"passage: TAGS\n#task_categories-sentence-similarity #size_categories-100K<n<1M #language-Dutch #license-cc-by-sa-4.0 #generic #sentence similarity #arxiv-1910.13461 #doi-10.57967/hf/1268 #region-us \n# Dataset Card for \"SNLI_Dutch_translated_with_Marianmt\"\n\nTranslation of the English corpus Stanford Natural Language Inference (SNLI),\nto Dutch using an Maria NMT model, trained by Helsinki NLP.\nNote, for reference: Maria NMT is based on BART, described here.\n\nA complete description of the dataset is given here.# Attribution\n\nIf you use this dataset please use the following to credit the creators of SNLI:\n\n\n\nThe creators of the OPUS-MT models:\n\n\nand# License\n\nFor both the Maria NMT model and the original Helsinki NLP Opus MT model \nwe did not find a license, if this was in error please let us know and we will add the appropriate licensing promptly.\n\nWe adopt the licensing of the SNLI corpus: a Creative Commons Attribution-ShareAlike 4.0 International License."
] |
c8fe8fd2cc4c7ac6bf666f54595bfd39a835d5b6
|
# Dataset Card for "model_v1_instruction_finetuning_dataset"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
sayan1101/model_v1_instruction_finetuning_dataset
|
[
"region:us"
] |
2023-10-26T07:19:48+00:00
|
{"dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 27415192.0, "num_examples": 52002}], "download_size": 12320134, "dataset_size": 27415192.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T08:09:30+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "model_v1_instruction_finetuning_dataset"
More Information needed
|
[
"# Dataset Card for \"model_v1_instruction_finetuning_dataset\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"model_v1_instruction_finetuning_dataset\"\n\nMore Information needed"
] |
[
6,
23
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"model_v1_instruction_finetuning_dataset\"\n\nMore Information needed"
] |
f6b6a8fb200aacefa7f3e4d3d4e14276e3c65258
|
# Datset Card for FLAIR land-cover semantic segmentation
## Context & Data
<hr style='margin-top:-1em; margin-bottom:0' />
The hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains).
Aerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines).
Furthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided.
More than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.
<br>
The dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.
We sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes.
In contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>
<style type="text/css">
.tg {border-collapse:collapse;border-spacing:0;}
.tg td{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;
overflow:hidden;padding:2px 5px;word-break:normal;}
.tg th{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;
font-weight:normal;overflow:hidden;padding:10px 5px;word-break:normal;}
.tg .tg-km2t{border-color:#ffffff;font-weight:bold;text-align:left;vertical-align:top}
.tg .tg-rime{background-color:#E4DF7C;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-r3rw{background-color:#a97101;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-zv4m{border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nto1{background-color:#000000;border-color:inherit;text-align:left;vertical-align:top}
.tg .tg-9efv{background-color:#938e7b;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-8jgo{border-color:#ffffff;text-align:center;vertical-align:top}
.tg .tg-b45e{background-color:#194A26;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-9xgv{background-color:#1553ae;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-3m6m{background-color:#f80c00;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2e1p{background-color:#db0e9a;border-color:#ffffff;color:#db0e9a;text-align:left;vertical-align:top}
.tg .tg-l5fa{background-color:#FFF30D;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2cns{background-color:#3DE6EB;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-jjsp{background-color:#FFF;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2w6m{background-color:#8AB3A0;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nla7{background-color:#6B714F;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-qg2z{background-color:#46E483;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nv8o{background-color:#C5DC42;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-grz5{background-color:#F3A60D;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-bja1{background-color:#99F;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-69kt{background-color:#660082;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-r1r4{background-color:#5F0;border-color:#ffffff;text-align:left;vertical-align:top}
</style>
<table class="tg">
<thead>
<tr>
<th class="tg-zv4m"></th>
<th class="tg-zv4m">Class</th>
<th class="tg-8jgo">Train/val (%)</th>
<th class="tg-8jgo">Test flair#1 (%)</th>
<th class="tg-8jgo">Test flair#2 (%)</th>
<th class="tg-zv4m"></th>
<th class="tg-zv4m">Class</th>
<th class="tg-8jgo">Train/val (%)</th>
<th class="tg-8jgo">Test flair#1 (%)</th>
<th class="tg-8jgo">Test flair#2 (%)</th>
</tr>
</thead>
<tbody>
<tr>
<td class="tg-2e1p"></td>
<td class="tg-km2t">(1) Building</td>
<td class="tg-8jgo">8.14</td>
<td class="tg-8jgo">8.6</td>
<td class="tg-8jgo">3.26</td>
<td class="tg-l5fa"></td>
<td class="tg-km2t">(11) Agricultural Land</td>
<td class="tg-8jgo">10.98</td>
<td class="tg-8jgo">6.95</td>
<td class="tg-8jgo">18.19</td>
</tr>
<tr>
<td class="tg-9efv"></td>
<td class="tg-km2t">(2) Pervious surface</td>
<td class="tg-8jgo">8.25</td>
<td class="tg-8jgo">7.34</td>
<td class="tg-8jgo">3.82</td>
<td class="tg-rime"></td>
<td class="tg-km2t">(12) Plowed land</td>
<td class="tg-8jgo">3.88</td>
<td class="tg-8jgo">2.25</td>
<td class="tg-8jgo">1.81</td>
</tr>
<tr>
<td class="tg-3m6m"></td>
<td class="tg-km2t">(3) Impervious surface</td>
<td class="tg-8jgo">13.72</td>
<td class="tg-8jgo">14.98</td>
<td class="tg-8jgo">5.87</td>
<td class="tg-2cns"></td>
<td class="tg-km2t">(13) Swimming pool</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.04</td>
<td class="tg-8jgo">0.02</td>
</tr>
<tr>
<td class="tg-r3rw"></td>
<td class="tg-km2t">(4) Bare soil</td>
<td class="tg-8jgo">3.47</td>
<td class="tg-8jgo">4.36</td>
<td class="tg-8jgo">1.6</td>
<td class="tg-jjsp"></td>
<td class="tg-km2t">(14) Snow</td>
<td class="tg-8jgo">0.15</td>
<td class="tg-8jgo">-</td>
<td class="tg-8jgo">-</td>
</tr>
<tr>
<td class="tg-9xgv"></td>
<td class="tg-km2t">(5) Water</td>
<td class="tg-8jgo">4.88</td>
<td class="tg-8jgo">5.98</td>
<td class="tg-8jgo">3.17</td>
<td class="tg-2w6m"></td>
<td class="tg-km2t">(15) Clear cut</td>
<td class="tg-8jgo">0.15</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.82</td>
</tr>
<tr>
<td class="tg-b45e"></td>
<td class="tg-km2t">(6) Coniferous</td>
<td class="tg-8jgo">2.74</td>
<td class="tg-8jgo">2.39</td>
<td class="tg-8jgo">10.24</td>
<td class="tg-nla7"></td>
<td class="tg-km2t">(16) Mixed</td>
<td class="tg-8jgo">0.05</td>
<td class="tg-8jgo">-</td>
<td class="tg-8jgo">0.12</td>
</tr>
<tr>
<td class="tg-qg2z"></td>
<td class="tg-km2t">(7) Deciduous</td>
<td class="tg-8jgo">15.38</td>
<td class="tg-8jgo">13.91</td>
<td class="tg-8jgo">24.79</td>
<td class="tg-nv8o"></td>
<td class="tg-km2t">(17) Ligneous</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.03</td>
<td class="tg-8jgo">-</td>
</tr>
<tr>
<td class="tg-grz5"></td>
<td class="tg-km2t">(8) Brushwood</td>
<td class="tg-8jgo">6.95</td>
<td class="tg-8jgo">6.91</td>
<td class="tg-8jgo">3.81</td>
<td class="tg-bja1"></td>
<td class="tg-km2t">(18) Greenhouse</td>
<td class="tg-8jgo">0.12</td>
<td class="tg-8jgo">0.2</td>
<td class="tg-8jgo">0.15</td>
</tr>
<tr>
<td class="tg-69kt"></td>
<td class="tg-km2t">(9) Vineyard</td>
<td class="tg-8jgo">3.13</td>
<td class="tg-8jgo">3.87</td>
<td class="tg-8jgo">2.55</td>
<td class="tg-nto1"></td>
<td class="tg-km2t">(19) Other</td>
<td class="tg-8jgo">0.14</td>
<td class="tg-8jgo">0.-</td>
<td class="tg-8jgo">0.04</td>
</tr>
<tr>
<td class="tg-r1r4"></td>
<td class="tg-km2t">(10) Herbaceous vegetation</td>
<td class="tg-8jgo">17.84</td>
<td class="tg-8jgo">22.17</td>
<td class="tg-8jgo">19.76</td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
</tr>
</tbody>
</table>
<br><br>
## Dataset Structure
<hr style='margin-top:-1em; margin-bottom:0' />
The FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.
Each patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m
and associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).
<p align="center"><img src="flair-patches.png" alt="" style="width:70%;max-width:600px;"/></p><br>
### Band order
<div style="display: flex;">
<div style="width: 15%;margin-right: 1;"">
Aerial
<ul>
<li>1. Red</li>
<li>2. Green</li>
<li>3. Blue</li>
<li>4. NIR</li>
<li>5. nDSM</li>
</ul>
</div>
<div style="width: 25%;">
Satellite
<ul>
<li>1. Blue (B2 490nm)</li>
<li>2. Green (B3 560nm)</li>
<li>3. Red (B4 665nm)</li>
<li>4. Red-Edge (B5 705nm)</li>
<li>5. Red-Edge2 (B6 470nm)</li>
<li>6. Red-Edge3 (B7 783nm)</li>
<li>7. NIR (B8 842nm)</li>
<li>8. NIR-Red-Edge (B8a 865nm)</li>
<li>9. SWIR (B11 1610nm)</li>
<li>10. SWIR2 (B12 2190nm)</li>
</ul>
</div>
</div>
### Annotations
Each pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN.
Movable objects like cars or boats are annotated according to their underlying cover.
### Data Splits
The dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements.
For our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test.
It can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain.
This arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set.
Consequently, every split accurately reflects the landscape diversity inherent to metropolitan France.
It is important to mention that the patches come with meta-data permitting alternative splitting schemes.
Official domain split: <br/>
<div style="display: flex; flex-wrap: nowrap; align-items: center">
<div style="flex: 40%;">
<img src="flair-splits.png" alt="flair-splits">
</div>
<div style="flex: 60%; margin: auto;"">
<table border="1">
<tr>
<th><font color="#c7254e">TRAIN:</font></th>
<td>D006, D007, D008, D009, D013, D016, D017, D021, D023, D030, D032, D033, D034, D035, D038, D041, D044, D046, D049, D051, D052, D055, D060, D063, D070, D072, D074, D078, D080, D081, D086, D091</td>
</tr>
<tr>
<th><font color="#c7254e">VALIDATION:</font></th>
<td>D004, D014, D029, D031, D058, D066, D067, D077</td>
</tr>
<tr>
<th><font color="#c7254e">TEST-flair#1:</font></th>
<td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>
</tr>
<tr>
<th><font color="#c7254e">TEST-flair#2:</font></th>
<td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>
</tr>
</table>
</div>
</div>
<br><br>
## Baseline code
<hr style='margin-top:-1em; margin-bottom:0' />
<br>
### Flair #1 (aerial only)
A U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines.
The used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques.
Flair#1 code repository 📁 : https://github.com/IGNF/FLAIR-1<br/>
Link to the paper : https://arxiv.org/pdf/2211.12979.pdf <br>
Please include a citation to the following article if you use the FLAIR#1 dataset:
```
@article{ign2022flair1,
doi = {10.13140/RG.2.2.30183.73128/1},
url = {https://arxiv.org/pdf/2211.12979.pdf},
author = {Garioud, Anatol and Peillet, Stéphane and Bookjans, Eva and Giordano, Sébastien and Wattrelos, Boris},
title = {FLAIR #1: semantic segmentation and domain adaptation dataset},
publisher = {arXiv},
year = {2022}
}
```
<br>
### Flair #2 (aerial and satellite)
We propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch,
the U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data,
applying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources,
enhancing the representation of mono-date and time series data.
U-T&T code repository 📁 : https://github.com/IGNF/FLAIR-2<br/>
Link to the paper : https://arxiv.org/abs/2310.13336 <br>
<th><font color="#c7254e"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b>
To work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided in aux-data.
You also need to add the following lines to the <font color=‘#D7881C’><em>flair-2-config.yml</em></font> file under the <em><b>data</b></em> tag: <br>
```
HF_data_path : " " # Path to unzipped FLAIR HF dataset
domains_train : ["D006_2020","D007_2020","D008_2019","D009_2019","D013_2020","D016_2020","D017_2018","D021_2020","D023_2020","D030_2021","D032_2019","D033_2021","D034_2021","D035_2020","D038_2021","D041_2021","D044_2020","D046_2019","D049_2020","D051_2019","D052_2019","D055_2018","D060_2021","D063_2019","D070_2020","D072_2019","D074_2020","D078_2021","D080_2021","D081_2020","D086_2020","D091_2021"]
domains_val : ["D004_2021","D014_2020","D029_2021","D031_2019","D058_2020","D066_2021","D067_2021","D077_2021"]
domains_test : ["D015_2020","D022_2021","D026_2020","D036_2020","D061_2020","D064_2021","D068_2021","D069_2020","D071_2020","D084_2021"]
```
<br>
Please include a citation to the following article if you use the FLAIR#2 dataset:
```
@inproceedings{garioud2023flair,
title={FLAIR: a Country-Scale Land Cover Semantic Segmentation Dataset From Multi-Source Optical Imagery},
author={Anatol Garioud and Nicolas Gonthier and Loic Landrieu and Apolline De Wit and Marion Valette and Marc Poupée and Sébastien Giordano and Boris Wattrelos},
year={2023},
booktitle={Advances in Neural Information Processing Systems (NeurIPS) 2023},
doi={https://doi.org/10.48550/arXiv.2310.13336},
}
```
<br>
## CodaLab challenges
<hr style='margin-top:-1em; margin-bottom:0' />
The FLAIR dataset was used for two challenges organized by IGN in 2023 on the CodaLab platform.<br>
Challenge FLAIR#1 : https://codalab.lisn.upsaclay.fr/competitions/8769 <br>
Challenge FLAIR#2 : https://codalab.lisn.upsaclay.fr/competitions/13447 <br>
flair#1-test | The podium:
🥇 businiao - 0.65920
🥈 Breizhchess - 0.65600
🥉 wangzhiyu918 - 0.64930
flair#2-test | The podium:
🥇 strakajk - 0.64130
🥈 Breizhchess - 0.63550
🥉 qwerty64 - 0.63510
## Acknowledgment
<hr style='margin-top:-1em; margin-bottom:0' />
This work was performed using HPC/AI resources from GENCI-IDRIS (Grant 2022-A0131013803). This work was supported by the project "Copernicus / FPCUP” of the European Union, by the French Space Agency (CNES) and by Connect by CNES.<br>
## Contact
<hr style='margin-top:-1em; margin-bottom:0' />
If you have any questions, issues or feedback, you can contact us at: [email protected]
<br>
## Dataset license
<hr style='margin-top:-1em; margin-bottom:0' />
The "OPEN LICENCE 2.0/LICENCE OUVERTE" is a license created by the French government specifically for the purpose of facilitating the dissemination of open data by public administration.<br/>
This licence is governed by French law.<br/>
This licence has been designed to be compatible with any free licence that at least requires an acknowledgement of authorship, and specifically with the previous version of this licence as well as with the following licences: United Kingdom’s “Open Government Licence” (OGL), Creative Commons’ “Creative Commons Attribution” (CC-BY) and Open Knowledge Foundation’s “Open Data Commons Attribution” (ODC-BY).
|
IGNF/FLAIR
|
[
"task_categories:image-segmentation",
"size_categories:10B<n<100B",
"license:etalab-2.0",
"IGN",
"Aerial",
"Satellite",
"Environement",
"Multimodal",
"Earth Observation",
"arxiv:2211.12979",
"arxiv:2310.13336",
"region:us"
] |
2023-10-26T07:32:37+00:00
|
{"license": "etalab-2.0", "size_categories": ["10B<n<100B"], "task_categories": ["image-segmentation"], "pretty_name": "French Land Cover from Aerospace Imagery", "tags": ["IGN", "Aerial", "Satellite", "Environement", "Multimodal", "Earth Observation"]}
|
2024-02-06T08:43:31+00:00
|
[
"2211.12979",
"2310.13336"
] |
[] |
TAGS
#task_categories-image-segmentation #size_categories-10B<n<100B #license-etalab-2.0 #IGN #Aerial #Satellite #Environement #Multimodal #Earth Observation #arxiv-2211.12979 #arxiv-2310.13336 #region-us
|
# Datset Card for FLAIR land-cover semantic segmentation
## Context & Data
<hr style='margin-top:-1em; margin-bottom:0' />
The hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains).
Aerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines).
Furthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided.
More than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.
<br>
The dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.
We sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes.
In contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>
<style type="text/css">
.tg {border-collapse:collapse;border-spacing:0;}
.tg td{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;
overflow:hidden;padding:2px 5px;word-break:normal;}
.tg th{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;
font-weight:normal;overflow:hidden;padding:10px 5px;word-break:normal;}
.tg .tg-km2t{border-color:#ffffff;font-weight:bold;text-align:left;vertical-align:top}
.tg .tg-rime{background-color:#E4DF7C;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-r3rw{background-color:#a97101;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-zv4m{border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nto1{background-color:#000000;border-color:inherit;text-align:left;vertical-align:top}
.tg .tg-9efv{background-color:#938e7b;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-8jgo{border-color:#ffffff;text-align:center;vertical-align:top}
.tg .tg-b45e{background-color:#194A26;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-9xgv{background-color:#1553ae;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-3m6m{background-color:#f80c00;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2e1p{background-color:#db0e9a;border-color:#ffffff;color:#db0e9a;text-align:left;vertical-align:top}
.tg .tg-l5fa{background-color:#FFF30D;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2cns{background-color:#3DE6EB;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-jjsp{background-color:#FFF;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-2w6m{background-color:#8AB3A0;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nla7{background-color:#6B714F;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-qg2z{background-color:#46E483;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-nv8o{background-color:#C5DC42;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-grz5{background-color:#F3A60D;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-bja1{background-color:#99F;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-69kt{background-color:#660082;border-color:#ffffff;text-align:left;vertical-align:top}
.tg .tg-r1r4{background-color:#5F0;border-color:#ffffff;text-align:left;vertical-align:top}
</style>
<table class="tg">
<thead>
<tr>
<th class="tg-zv4m"></th>
<th class="tg-zv4m">Class</th>
<th class="tg-8jgo">Train/val (%)</th>
<th class="tg-8jgo">Test flair#1 (%)</th>
<th class="tg-8jgo">Test flair#2 (%)</th>
<th class="tg-zv4m"></th>
<th class="tg-zv4m">Class</th>
<th class="tg-8jgo">Train/val (%)</th>
<th class="tg-8jgo">Test flair#1 (%)</th>
<th class="tg-8jgo">Test flair#2 (%)</th>
</tr>
</thead>
<tbody>
<tr>
<td class="tg-2e1p"></td>
<td class="tg-km2t">(1) Building</td>
<td class="tg-8jgo">8.14</td>
<td class="tg-8jgo">8.6</td>
<td class="tg-8jgo">3.26</td>
<td class="tg-l5fa"></td>
<td class="tg-km2t">(11) Agricultural Land</td>
<td class="tg-8jgo">10.98</td>
<td class="tg-8jgo">6.95</td>
<td class="tg-8jgo">18.19</td>
</tr>
<tr>
<td class="tg-9efv"></td>
<td class="tg-km2t">(2) Pervious surface</td>
<td class="tg-8jgo">8.25</td>
<td class="tg-8jgo">7.34</td>
<td class="tg-8jgo">3.82</td>
<td class="tg-rime"></td>
<td class="tg-km2t">(12) Plowed land</td>
<td class="tg-8jgo">3.88</td>
<td class="tg-8jgo">2.25</td>
<td class="tg-8jgo">1.81</td>
</tr>
<tr>
<td class="tg-3m6m"></td>
<td class="tg-km2t">(3) Impervious surface</td>
<td class="tg-8jgo">13.72</td>
<td class="tg-8jgo">14.98</td>
<td class="tg-8jgo">5.87</td>
<td class="tg-2cns"></td>
<td class="tg-km2t">(13) Swimming pool</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.04</td>
<td class="tg-8jgo">0.02</td>
</tr>
<tr>
<td class="tg-r3rw"></td>
<td class="tg-km2t">(4) Bare soil</td>
<td class="tg-8jgo">3.47</td>
<td class="tg-8jgo">4.36</td>
<td class="tg-8jgo">1.6</td>
<td class="tg-jjsp"></td>
<td class="tg-km2t">(14) Snow</td>
<td class="tg-8jgo">0.15</td>
<td class="tg-8jgo">-</td>
<td class="tg-8jgo">-</td>
</tr>
<tr>
<td class="tg-9xgv"></td>
<td class="tg-km2t">(5) Water</td>
<td class="tg-8jgo">4.88</td>
<td class="tg-8jgo">5.98</td>
<td class="tg-8jgo">3.17</td>
<td class="tg-2w6m"></td>
<td class="tg-km2t">(15) Clear cut</td>
<td class="tg-8jgo">0.15</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.82</td>
</tr>
<tr>
<td class="tg-b45e"></td>
<td class="tg-km2t">(6) Coniferous</td>
<td class="tg-8jgo">2.74</td>
<td class="tg-8jgo">2.39</td>
<td class="tg-8jgo">10.24</td>
<td class="tg-nla7"></td>
<td class="tg-km2t">(16) Mixed</td>
<td class="tg-8jgo">0.05</td>
<td class="tg-8jgo">-</td>
<td class="tg-8jgo">0.12</td>
</tr>
<tr>
<td class="tg-qg2z"></td>
<td class="tg-km2t">(7) Deciduous</td>
<td class="tg-8jgo">15.38</td>
<td class="tg-8jgo">13.91</td>
<td class="tg-8jgo">24.79</td>
<td class="tg-nv8o"></td>
<td class="tg-km2t">(17) Ligneous</td>
<td class="tg-8jgo">0.01</td>
<td class="tg-8jgo">0.03</td>
<td class="tg-8jgo">-</td>
</tr>
<tr>
<td class="tg-grz5"></td>
<td class="tg-km2t">(8) Brushwood</td>
<td class="tg-8jgo">6.95</td>
<td class="tg-8jgo">6.91</td>
<td class="tg-8jgo">3.81</td>
<td class="tg-bja1"></td>
<td class="tg-km2t">(18) Greenhouse</td>
<td class="tg-8jgo">0.12</td>
<td class="tg-8jgo">0.2</td>
<td class="tg-8jgo">0.15</td>
</tr>
<tr>
<td class="tg-69kt"></td>
<td class="tg-km2t">(9) Vineyard</td>
<td class="tg-8jgo">3.13</td>
<td class="tg-8jgo">3.87</td>
<td class="tg-8jgo">2.55</td>
<td class="tg-nto1"></td>
<td class="tg-km2t">(19) Other</td>
<td class="tg-8jgo">0.14</td>
<td class="tg-8jgo">0.-</td>
<td class="tg-8jgo">0.04</td>
</tr>
<tr>
<td class="tg-r1r4"></td>
<td class="tg-km2t">(10) Herbaceous vegetation</td>
<td class="tg-8jgo">17.84</td>
<td class="tg-8jgo">22.17</td>
<td class="tg-8jgo">19.76</td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
<td class="tg-zv4m"></td>
</tr>
</tbody>
</table>
<br><br>
## Dataset Structure
<hr style='margin-top:-1em; margin-bottom:0' />
The FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.
Each patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m
and associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).
<p align="center"><img src="URL" alt="" style="width:70%;max-width:600px;"/></p><br>
### Band order
<div style="display: flex;">
<div style="width: 15%;margin-right: 1;"">
Aerial
<ul>
<li>1. Red</li>
<li>2. Green</li>
<li>3. Blue</li>
<li>4. NIR</li>
<li>5. nDSM</li>
</ul>
</div>
<div style="width: 25%;">
Satellite
<ul>
<li>1. Blue (B2 490nm)</li>
<li>2. Green (B3 560nm)</li>
<li>3. Red (B4 665nm)</li>
<li>4. Red-Edge (B5 705nm)</li>
<li>5. Red-Edge2 (B6 470nm)</li>
<li>6. Red-Edge3 (B7 783nm)</li>
<li>7. NIR (B8 842nm)</li>
<li>8. NIR-Red-Edge (B8a 865nm)</li>
<li>9. SWIR (B11 1610nm)</li>
<li>10. SWIR2 (B12 2190nm)</li>
</ul>
</div>
</div>
### Annotations
Each pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN.
Movable objects like cars or boats are annotated according to their underlying cover.
### Data Splits
The dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements.
For our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test.
It can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain.
This arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set.
Consequently, every split accurately reflects the landscape diversity inherent to metropolitan France.
It is important to mention that the patches come with meta-data permitting alternative splitting schemes.
Official domain split: <br/>
<div style="display: flex; flex-wrap: nowrap; align-items: center">
<div style="flex: 40%;">
<img src="URL" alt="flair-splits">
</div>
<div style="flex: 60%; margin: auto;"">
<table border="1">
<tr>
<th><font color="#c7254e">TRAIN:</font></th>
<td>D006, D007, D008, D009, D013, D016, D017, D021, D023, D030, D032, D033, D034, D035, D038, D041, D044, D046, D049, D051, D052, D055, D060, D063, D070, D072, D074, D078, D080, D081, D086, D091</td>
</tr>
<tr>
<th><font color="#c7254e">VALIDATION:</font></th>
<td>D004, D014, D029, D031, D058, D066, D067, D077</td>
</tr>
<tr>
<th><font color="#c7254e">TEST-flair#1:</font></th>
<td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>
</tr>
<tr>
<th><font color="#c7254e">TEST-flair#2:</font></th>
<td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>
</tr>
</table>
</div>
</div>
<br><br>
## Baseline code
<hr style='margin-top:-1em; margin-bottom:0' />
<br>
### Flair #1 (aerial only)
A U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines.
The used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques.
Flair#1 code repository 📁 : URL
Link to the paper : URL <br>
Please include a citation to the following article if you use the FLAIR#1 dataset:
<br>
### Flair #2 (aerial and satellite)
We propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch,
the U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data,
applying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources,
enhancing the representation of mono-date and time series data.
U-T&T code repository 📁 : URL
Link to the paper : URL <br>
<th><font color="#c7254e"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b>
To work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided in aux-data.
You also need to add the following lines to the <font color=‘#D7881C’><em>URL</em></font> file under the <em><b>data</b></em> tag: <br>
<br>
Please include a citation to the following article if you use the FLAIR#2 dataset:
<br>
## CodaLab challenges
<hr style='margin-top:-1em; margin-bottom:0' />
The FLAIR dataset was used for two challenges organized by IGN in 2023 on the CodaLab platform.<br>
Challenge FLAIR#1 : URL <br>
Challenge FLAIR#2 : URL <br>
flair#1-test | The podium:
businiao - 0.65920
Breizhchess - 0.65600
wangzhiyu918 - 0.64930
flair#2-test | The podium:
strakajk - 0.64130
Breizhchess - 0.63550
qwerty64 - 0.63510
## Acknowledgment
<hr style='margin-top:-1em; margin-bottom:0' />
This work was performed using HPC/AI resources from GENCI-IDRIS (Grant 2022-A0131013803). This work was supported by the project "Copernicus / FPCUP” of the European Union, by the French Space Agency (CNES) and by Connect by CNES.<br>
## Contact
<hr style='margin-top:-1em; margin-bottom:0' />
If you have any questions, issues or feedback, you can contact us at: ai-challenge@URL
<br>
## Dataset license
<hr style='margin-top:-1em; margin-bottom:0' />
The "OPEN LICENCE 2.0/LICENCE OUVERTE" is a license created by the French government specifically for the purpose of facilitating the dissemination of open data by public administration.<br/>
This licence is governed by French law.<br/>
This licence has been designed to be compatible with any free licence that at least requires an acknowledgement of authorship, and specifically with the previous version of this licence as well as with the following licences: United Kingdom’s “Open Government Licence” (OGL), Creative Commons’ “Creative Commons Attribution” (CC-BY) and Open Knowledge Foundation’s “Open Data Commons Attribution” (ODC-BY).
|
[
"# Datset Card for FLAIR land-cover semantic segmentation",
"## Context & Data\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains). \nAerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines). \nFurthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided. \nMore than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.\n<br>\n\nThe dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.\nWe sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes. \nIn contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>\n\n<style type=\"text/css\">\n.tg {border-collapse:collapse;border-spacing:0;}\n.tg td{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n overflow:hidden;padding:2px 5px;word-break:normal;}\n.tg th{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n font-weight:normal;overflow:hidden;padding:10px 5px;word-break:normal;}\n.tg .tg-km2t{border-color:#ffffff;font-weight:bold;text-align:left;vertical-align:top}\n.tg .tg-rime{background-color:#E4DF7C;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r3rw{background-color:#a97101;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-zv4m{border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nto1{background-color:#000000;border-color:inherit;text-align:left;vertical-align:top}\n.tg .tg-9efv{background-color:#938e7b;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-8jgo{border-color:#ffffff;text-align:center;vertical-align:top}\n.tg .tg-b45e{background-color:#194A26;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-9xgv{background-color:#1553ae;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-3m6m{background-color:#f80c00;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2e1p{background-color:#db0e9a;border-color:#ffffff;color:#db0e9a;text-align:left;vertical-align:top}\n.tg .tg-l5fa{background-color:#FFF30D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2cns{background-color:#3DE6EB;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-jjsp{background-color:#FFF;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2w6m{background-color:#8AB3A0;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nla7{background-color:#6B714F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-qg2z{background-color:#46E483;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nv8o{background-color:#C5DC42;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-grz5{background-color:#F3A60D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-bja1{background-color:#99F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-69kt{background-color:#660082;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r1r4{background-color:#5F0;border-color:#ffffff;text-align:left;vertical-align:top}\n</style>\n<table class=\"tg\">\n<thead>\n <tr>\n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n </tr>\n</thead>\n<tbody>\n <tr>\n <td class=\"tg-2e1p\"></td>\n <td class=\"tg-km2t\">(1) Building</td>\n <td class=\"tg-8jgo\">8.14</td>\n <td class=\"tg-8jgo\">8.6</td> \n <td class=\"tg-8jgo\">3.26</td>\n <td class=\"tg-l5fa\"></td>\n <td class=\"tg-km2t\">(11) Agricultural Land</td>\n <td class=\"tg-8jgo\">10.98</td>\n <td class=\"tg-8jgo\">6.95</td> \n <td class=\"tg-8jgo\">18.19</td>\n </tr>\n <tr>\n <td class=\"tg-9efv\"></td>\n <td class=\"tg-km2t\">(2) Pervious surface</td>\n <td class=\"tg-8jgo\">8.25</td>\n <td class=\"tg-8jgo\">7.34</td> \n <td class=\"tg-8jgo\">3.82</td>\n <td class=\"tg-rime\"></td>\n <td class=\"tg-km2t\">(12) Plowed land</td>\n <td class=\"tg-8jgo\">3.88</td>\n <td class=\"tg-8jgo\">2.25</td> \n <td class=\"tg-8jgo\">1.81</td>\n </tr>\n <tr>\n <td class=\"tg-3m6m\"></td>\n <td class=\"tg-km2t\">(3) Impervious surface</td>\n <td class=\"tg-8jgo\">13.72</td>\n <td class=\"tg-8jgo\">14.98</td> \n <td class=\"tg-8jgo\">5.87</td>\n <td class=\"tg-2cns\"></td>\n <td class=\"tg-km2t\">(13) Swimming pool</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.04</td> \n <td class=\"tg-8jgo\">0.02</td>\n </tr>\n <tr>\n <td class=\"tg-r3rw\"></td>\n <td class=\"tg-km2t\">(4) Bare soil</td>\n <td class=\"tg-8jgo\">3.47</td>\n <td class=\"tg-8jgo\">4.36</td> \n <td class=\"tg-8jgo\">1.6</td>\n <td class=\"tg-jjsp\"></td>\n <td class=\"tg-km2t\">(14) Snow</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-9xgv\"></td>\n <td class=\"tg-km2t\">(5) Water</td>\n <td class=\"tg-8jgo\">4.88</td>\n <td class=\"tg-8jgo\">5.98</td> \n <td class=\"tg-8jgo\">3.17</td>\n <td class=\"tg-2w6m\"></td>\n <td class=\"tg-km2t\">(15) Clear cut</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">0.01</td> \n <td class=\"tg-8jgo\">0.82</td>\n </tr>\n <tr>\n <td class=\"tg-b45e\"></td>\n <td class=\"tg-km2t\">(6) Coniferous</td>\n <td class=\"tg-8jgo\">2.74</td>\n <td class=\"tg-8jgo\">2.39</td>\n <td class=\"tg-8jgo\">10.24</td>\n <td class=\"tg-nla7\"></td>\n <td class=\"tg-km2t\">(16) Mixed</td>\n <td class=\"tg-8jgo\">0.05</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">0.12</td>\n </tr>\n <tr>\n <td class=\"tg-qg2z\"></td>\n <td class=\"tg-km2t\">(7) Deciduous</td>\n <td class=\"tg-8jgo\">15.38</td>\n <td class=\"tg-8jgo\">13.91</td>\n <td class=\"tg-8jgo\">24.79</td>\n <td class=\"tg-nv8o\"></td>\n <td class=\"tg-km2t\">(17) Ligneous</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.03</td>\n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-grz5\"></td>\n <td class=\"tg-km2t\">(8) Brushwood</td>\n <td class=\"tg-8jgo\">6.95</td>\n <td class=\"tg-8jgo\">6.91</td>\n <td class=\"tg-8jgo\">3.81</td> \n <td class=\"tg-bja1\"></td>\n <td class=\"tg-km2t\">(18) Greenhouse</td>\n <td class=\"tg-8jgo\">0.12</td>\n <td class=\"tg-8jgo\">0.2</td> \n <td class=\"tg-8jgo\">0.15</td>\n </tr>\n <tr>\n <td class=\"tg-69kt\"></td>\n <td class=\"tg-km2t\">(9) Vineyard</td>\n <td class=\"tg-8jgo\">3.13</td>\n <td class=\"tg-8jgo\">3.87</td>\n <td class=\"tg-8jgo\">2.55</td>\n <td class=\"tg-nto1\"></td>\n <td class=\"tg-km2t\">(19) Other</td>\n <td class=\"tg-8jgo\">0.14</td>\n <td class=\"tg-8jgo\">0.-</td> \n <td class=\"tg-8jgo\">0.04</td>\n </tr>\n <tr>\n <td class=\"tg-r1r4\"></td>\n <td class=\"tg-km2t\">(10) Herbaceous vegetation</td>\n <td class=\"tg-8jgo\">17.84</td>\n <td class=\"tg-8jgo\">22.17</td>\n <td class=\"tg-8jgo\">19.76</td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n </tr>\n</tbody>\n</table>\n\n<br><br>",
"## Dataset Structure\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.\n\nEach patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m \nand associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).\n\n<p align=\"center\"><img src=\"URL\" alt=\"\" style=\"width:70%;max-width:600px;\"/></p><br>",
"### Band order\n\n<div style=\"display: flex;\">\n<div style=\"width: 15%;margin-right: 1;\"\">\nAerial\n<ul>\n<li>1. Red</li>\n<li>2. Green</li>\n<li>3. Blue</li>\n<li>4. NIR</li>\n<li>5. nDSM</li>\n</ul>\n</div>\n\n<div style=\"width: 25%;\">\nSatellite\n<ul>\n<li>1. Blue (B2 490nm)</li>\n<li>2. Green (B3 560nm)</li>\n<li>3. Red (B4 665nm)</li>\n<li>4. Red-Edge (B5 705nm)</li>\n<li>5. Red-Edge2 (B6 470nm)</li>\n<li>6. Red-Edge3 (B7 783nm)</li>\n<li>7. NIR (B8 842nm)</li>\n<li>8. NIR-Red-Edge (B8a 865nm)</li>\n<li>9. SWIR (B11 1610nm)</li>\n<li>10. SWIR2 (B12 2190nm)</li>\n</ul>\n</div>\n\n</div>",
"### Annotations\nEach pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN. \nMovable objects like cars or boats are annotated according to their underlying cover.",
"### Data Splits\nThe dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements. \nFor our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test. \nIt can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain. \nThis arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set. \nConsequently, every split accurately reflects the landscape diversity inherent to metropolitan France. \nIt is important to mention that the patches come with meta-data permitting alternative splitting schemes. \n\n\nOfficial domain split: <br/>\n\n<div style=\"display: flex; flex-wrap: nowrap; align-items: center\">\n <div style=\"flex: 40%;\">\n <img src=\"URL\" alt=\"flair-splits\">\n</div>\n\n <div style=\"flex: 60%; margin: auto;\"\">\n <table border=\"1\">\n <tr>\n <th><font color=\"#c7254e\">TRAIN:</font></th>\n <td>D006, D007, D008, D009, D013, D016, D017, D021, D023, D030, D032, D033, D034, D035, D038, D041, D044, D046, D049, D051, D052, D055, D060, D063, D070, D072, D074, D078, D080, D081, D086, D091</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">VALIDATION:</font></th>\n <td>D004, D014, D029, D031, D058, D066, D067, D077</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#1:</font></th>\n <td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#2:</font></th>\n <td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>\n </tr> \n </table>\n </div>\n</div>\n\n<br><br>",
"## Baseline code \n<hr style='margin-top:-1em; margin-bottom:0' />\n<br>",
"### Flair #1 (aerial only)\nA U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines. \nThe used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques. \n\nFlair#1 code repository 📁 : URL\nLink to the paper : URL <br>\n\nPlease include a citation to the following article if you use the FLAIR#1 dataset:\n\n\n<br>",
"### Flair #2 (aerial and satellite)\nWe propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch, \nthe U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data, \napplying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources, \nenhancing the representation of mono-date and time series data.\n\nU-T&T code repository 📁 : URL\nLink to the paper : URL <br>\n\n<th><font color=\"#c7254e\"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b> \nTo work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided in aux-data. \nYou also need to add the following lines to the <font color=‘#D7881C’><em>URL</em></font> file under the <em><b>data</b></em> tag: <br>\n\n\n<br>\nPlease include a citation to the following article if you use the FLAIR#2 dataset:\n\n\n<br>",
"## CodaLab challenges\n<hr style='margin-top:-1em; margin-bottom:0' />\n\nThe FLAIR dataset was used for two challenges organized by IGN in 2023 on the CodaLab platform.<br>\nChallenge FLAIR#1 : URL <br>\nChallenge FLAIR#2 : URL <br>\n\nflair#1-test | The podium: \n businiao - 0.65920 \n Breizhchess - 0.65600 \n wangzhiyu918 - 0.64930 \n\nflair#2-test | The podium: \n strakajk - 0.64130 \n Breizhchess - 0.63550 \n qwerty64 - 0.63510",
"## Acknowledgment\n<hr style='margin-top:-1em; margin-bottom:0' />\nThis work was performed using HPC/AI resources from GENCI-IDRIS (Grant 2022-A0131013803). This work was supported by the project \"Copernicus / FPCUP” of the European Union, by the French Space Agency (CNES) and by Connect by CNES.<br>",
"## Contact\n<hr style='margin-top:-1em; margin-bottom:0' />\nIf you have any questions, issues or feedback, you can contact us at: ai-challenge@URL \n<br>",
"## Dataset license\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe \"OPEN LICENCE 2.0/LICENCE OUVERTE\" is a license created by the French government specifically for the purpose of facilitating the dissemination of open data by public administration.<br/>\nThis licence is governed by French law.<br/>\nThis licence has been designed to be compatible with any free licence that at least requires an acknowledgement of authorship, and specifically with the previous version of this licence as well as with the following licences: United Kingdom’s “Open Government Licence” (OGL), Creative Commons’ “Creative Commons Attribution” (CC-BY) and Open Knowledge Foundation’s “Open Data Commons Attribution” (ODC-BY)."
] |
[
"TAGS\n#task_categories-image-segmentation #size_categories-10B<n<100B #license-etalab-2.0 #IGN #Aerial #Satellite #Environement #Multimodal #Earth Observation #arxiv-2211.12979 #arxiv-2310.13336 #region-us \n",
"# Datset Card for FLAIR land-cover semantic segmentation",
"## Context & Data\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains). \nAerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines). \nFurthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided. \nMore than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.\n<br>\n\nThe dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.\nWe sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes. \nIn contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>\n\n<style type=\"text/css\">\n.tg {border-collapse:collapse;border-spacing:0;}\n.tg td{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n overflow:hidden;padding:2px 5px;word-break:normal;}\n.tg th{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n font-weight:normal;overflow:hidden;padding:10px 5px;word-break:normal;}\n.tg .tg-km2t{border-color:#ffffff;font-weight:bold;text-align:left;vertical-align:top}\n.tg .tg-rime{background-color:#E4DF7C;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r3rw{background-color:#a97101;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-zv4m{border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nto1{background-color:#000000;border-color:inherit;text-align:left;vertical-align:top}\n.tg .tg-9efv{background-color:#938e7b;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-8jgo{border-color:#ffffff;text-align:center;vertical-align:top}\n.tg .tg-b45e{background-color:#194A26;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-9xgv{background-color:#1553ae;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-3m6m{background-color:#f80c00;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2e1p{background-color:#db0e9a;border-color:#ffffff;color:#db0e9a;text-align:left;vertical-align:top}\n.tg .tg-l5fa{background-color:#FFF30D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2cns{background-color:#3DE6EB;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-jjsp{background-color:#FFF;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2w6m{background-color:#8AB3A0;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nla7{background-color:#6B714F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-qg2z{background-color:#46E483;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nv8o{background-color:#C5DC42;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-grz5{background-color:#F3A60D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-bja1{background-color:#99F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-69kt{background-color:#660082;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r1r4{background-color:#5F0;border-color:#ffffff;text-align:left;vertical-align:top}\n</style>\n<table class=\"tg\">\n<thead>\n <tr>\n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n </tr>\n</thead>\n<tbody>\n <tr>\n <td class=\"tg-2e1p\"></td>\n <td class=\"tg-km2t\">(1) Building</td>\n <td class=\"tg-8jgo\">8.14</td>\n <td class=\"tg-8jgo\">8.6</td> \n <td class=\"tg-8jgo\">3.26</td>\n <td class=\"tg-l5fa\"></td>\n <td class=\"tg-km2t\">(11) Agricultural Land</td>\n <td class=\"tg-8jgo\">10.98</td>\n <td class=\"tg-8jgo\">6.95</td> \n <td class=\"tg-8jgo\">18.19</td>\n </tr>\n <tr>\n <td class=\"tg-9efv\"></td>\n <td class=\"tg-km2t\">(2) Pervious surface</td>\n <td class=\"tg-8jgo\">8.25</td>\n <td class=\"tg-8jgo\">7.34</td> \n <td class=\"tg-8jgo\">3.82</td>\n <td class=\"tg-rime\"></td>\n <td class=\"tg-km2t\">(12) Plowed land</td>\n <td class=\"tg-8jgo\">3.88</td>\n <td class=\"tg-8jgo\">2.25</td> \n <td class=\"tg-8jgo\">1.81</td>\n </tr>\n <tr>\n <td class=\"tg-3m6m\"></td>\n <td class=\"tg-km2t\">(3) Impervious surface</td>\n <td class=\"tg-8jgo\">13.72</td>\n <td class=\"tg-8jgo\">14.98</td> \n <td class=\"tg-8jgo\">5.87</td>\n <td class=\"tg-2cns\"></td>\n <td class=\"tg-km2t\">(13) Swimming pool</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.04</td> \n <td class=\"tg-8jgo\">0.02</td>\n </tr>\n <tr>\n <td class=\"tg-r3rw\"></td>\n <td class=\"tg-km2t\">(4) Bare soil</td>\n <td class=\"tg-8jgo\">3.47</td>\n <td class=\"tg-8jgo\">4.36</td> \n <td class=\"tg-8jgo\">1.6</td>\n <td class=\"tg-jjsp\"></td>\n <td class=\"tg-km2t\">(14) Snow</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-9xgv\"></td>\n <td class=\"tg-km2t\">(5) Water</td>\n <td class=\"tg-8jgo\">4.88</td>\n <td class=\"tg-8jgo\">5.98</td> \n <td class=\"tg-8jgo\">3.17</td>\n <td class=\"tg-2w6m\"></td>\n <td class=\"tg-km2t\">(15) Clear cut</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">0.01</td> \n <td class=\"tg-8jgo\">0.82</td>\n </tr>\n <tr>\n <td class=\"tg-b45e\"></td>\n <td class=\"tg-km2t\">(6) Coniferous</td>\n <td class=\"tg-8jgo\">2.74</td>\n <td class=\"tg-8jgo\">2.39</td>\n <td class=\"tg-8jgo\">10.24</td>\n <td class=\"tg-nla7\"></td>\n <td class=\"tg-km2t\">(16) Mixed</td>\n <td class=\"tg-8jgo\">0.05</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">0.12</td>\n </tr>\n <tr>\n <td class=\"tg-qg2z\"></td>\n <td class=\"tg-km2t\">(7) Deciduous</td>\n <td class=\"tg-8jgo\">15.38</td>\n <td class=\"tg-8jgo\">13.91</td>\n <td class=\"tg-8jgo\">24.79</td>\n <td class=\"tg-nv8o\"></td>\n <td class=\"tg-km2t\">(17) Ligneous</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.03</td>\n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-grz5\"></td>\n <td class=\"tg-km2t\">(8) Brushwood</td>\n <td class=\"tg-8jgo\">6.95</td>\n <td class=\"tg-8jgo\">6.91</td>\n <td class=\"tg-8jgo\">3.81</td> \n <td class=\"tg-bja1\"></td>\n <td class=\"tg-km2t\">(18) Greenhouse</td>\n <td class=\"tg-8jgo\">0.12</td>\n <td class=\"tg-8jgo\">0.2</td> \n <td class=\"tg-8jgo\">0.15</td>\n </tr>\n <tr>\n <td class=\"tg-69kt\"></td>\n <td class=\"tg-km2t\">(9) Vineyard</td>\n <td class=\"tg-8jgo\">3.13</td>\n <td class=\"tg-8jgo\">3.87</td>\n <td class=\"tg-8jgo\">2.55</td>\n <td class=\"tg-nto1\"></td>\n <td class=\"tg-km2t\">(19) Other</td>\n <td class=\"tg-8jgo\">0.14</td>\n <td class=\"tg-8jgo\">0.-</td> \n <td class=\"tg-8jgo\">0.04</td>\n </tr>\n <tr>\n <td class=\"tg-r1r4\"></td>\n <td class=\"tg-km2t\">(10) Herbaceous vegetation</td>\n <td class=\"tg-8jgo\">17.84</td>\n <td class=\"tg-8jgo\">22.17</td>\n <td class=\"tg-8jgo\">19.76</td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n </tr>\n</tbody>\n</table>\n\n<br><br>",
"## Dataset Structure\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.\n\nEach patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m \nand associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).\n\n<p align=\"center\"><img src=\"URL\" alt=\"\" style=\"width:70%;max-width:600px;\"/></p><br>",
"### Band order\n\n<div style=\"display: flex;\">\n<div style=\"width: 15%;margin-right: 1;\"\">\nAerial\n<ul>\n<li>1. Red</li>\n<li>2. Green</li>\n<li>3. Blue</li>\n<li>4. NIR</li>\n<li>5. nDSM</li>\n</ul>\n</div>\n\n<div style=\"width: 25%;\">\nSatellite\n<ul>\n<li>1. Blue (B2 490nm)</li>\n<li>2. Green (B3 560nm)</li>\n<li>3. Red (B4 665nm)</li>\n<li>4. Red-Edge (B5 705nm)</li>\n<li>5. Red-Edge2 (B6 470nm)</li>\n<li>6. Red-Edge3 (B7 783nm)</li>\n<li>7. NIR (B8 842nm)</li>\n<li>8. NIR-Red-Edge (B8a 865nm)</li>\n<li>9. SWIR (B11 1610nm)</li>\n<li>10. SWIR2 (B12 2190nm)</li>\n</ul>\n</div>\n\n</div>",
"### Annotations\nEach pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN. \nMovable objects like cars or boats are annotated according to their underlying cover.",
"### Data Splits\nThe dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements. \nFor our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test. \nIt can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain. \nThis arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set. \nConsequently, every split accurately reflects the landscape diversity inherent to metropolitan France. \nIt is important to mention that the patches come with meta-data permitting alternative splitting schemes. \n\n\nOfficial domain split: <br/>\n\n<div style=\"display: flex; flex-wrap: nowrap; align-items: center\">\n <div style=\"flex: 40%;\">\n <img src=\"URL\" alt=\"flair-splits\">\n</div>\n\n <div style=\"flex: 60%; margin: auto;\"\">\n <table border=\"1\">\n <tr>\n <th><font color=\"#c7254e\">TRAIN:</font></th>\n <td>D006, D007, D008, D009, D013, D016, D017, D021, D023, D030, D032, D033, D034, D035, D038, D041, D044, D046, D049, D051, D052, D055, D060, D063, D070, D072, D074, D078, D080, D081, D086, D091</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">VALIDATION:</font></th>\n <td>D004, D014, D029, D031, D058, D066, D067, D077</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#1:</font></th>\n <td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#2:</font></th>\n <td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>\n </tr> \n </table>\n </div>\n</div>\n\n<br><br>",
"## Baseline code \n<hr style='margin-top:-1em; margin-bottom:0' />\n<br>",
"### Flair #1 (aerial only)\nA U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines. \nThe used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques. \n\nFlair#1 code repository 📁 : URL\nLink to the paper : URL <br>\n\nPlease include a citation to the following article if you use the FLAIR#1 dataset:\n\n\n<br>",
"### Flair #2 (aerial and satellite)\nWe propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch, \nthe U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data, \napplying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources, \nenhancing the representation of mono-date and time series data.\n\nU-T&T code repository 📁 : URL\nLink to the paper : URL <br>\n\n<th><font color=\"#c7254e\"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b> \nTo work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided in aux-data. \nYou also need to add the following lines to the <font color=‘#D7881C’><em>URL</em></font> file under the <em><b>data</b></em> tag: <br>\n\n\n<br>\nPlease include a citation to the following article if you use the FLAIR#2 dataset:\n\n\n<br>",
"## CodaLab challenges\n<hr style='margin-top:-1em; margin-bottom:0' />\n\nThe FLAIR dataset was used for two challenges organized by IGN in 2023 on the CodaLab platform.<br>\nChallenge FLAIR#1 : URL <br>\nChallenge FLAIR#2 : URL <br>\n\nflair#1-test | The podium: \n businiao - 0.65920 \n Breizhchess - 0.65600 \n wangzhiyu918 - 0.64930 \n\nflair#2-test | The podium: \n strakajk - 0.64130 \n Breizhchess - 0.63550 \n qwerty64 - 0.63510",
"## Acknowledgment\n<hr style='margin-top:-1em; margin-bottom:0' />\nThis work was performed using HPC/AI resources from GENCI-IDRIS (Grant 2022-A0131013803). This work was supported by the project \"Copernicus / FPCUP” of the European Union, by the French Space Agency (CNES) and by Connect by CNES.<br>",
"## Contact\n<hr style='margin-top:-1em; margin-bottom:0' />\nIf you have any questions, issues or feedback, you can contact us at: ai-challenge@URL \n<br>",
"## Dataset license\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe \"OPEN LICENCE 2.0/LICENCE OUVERTE\" is a license created by the French government specifically for the purpose of facilitating the dissemination of open data by public administration.<br/>\nThis licence is governed by French law.<br/>\nThis licence has been designed to be compatible with any free licence that at least requires an acknowledgement of authorship, and specifically with the previous version of this licence as well as with the following licences: United Kingdom’s “Open Government Licence” (OGL), Creative Commons’ “Creative Commons Attribution” (CC-BY) and Open Knowledge Foundation’s “Open Data Commons Attribution” (ODC-BY)."
] |
[
80,
14,
3539,
210,
306,
70,
662,
26,
119,
399,
142,
94,
47,
165
] |
[
"passage: TAGS\n#task_categories-image-segmentation #size_categories-10B<n<100B #license-etalab-2.0 #IGN #Aerial #Satellite #Environement #Multimodal #Earth Observation #arxiv-2211.12979 #arxiv-2310.13336 #region-us \n# Datset Card for FLAIR land-cover semantic segmentation",
"passage: ## Context & Data\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains). \nAerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines). \nFurthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided. \nMore than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.\n<br>\n\nThe dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.\nWe sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes. \nIn contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>\n\n<style type=\"text/css\">\n.tg {border-collapse:collapse;border-spacing:0;}\n.tg td{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n overflow:hidden;padding:2px 5px;word-break:normal;}\n.tg th{border-color:black;border-style:solid;border-width:1px;font-family:Arial, sans-serif;font-size:13px;\n font-weight:normal;overflow:hidden;padding:10px 5px;word-break:normal;}\n.tg .tg-km2t{border-color:#ffffff;font-weight:bold;text-align:left;vertical-align:top}\n.tg .tg-rime{background-color:#E4DF7C;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r3rw{background-color:#a97101;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-zv4m{border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nto1{background-color:#000000;border-color:inherit;text-align:left;vertical-align:top}\n.tg .tg-9efv{background-color:#938e7b;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-8jgo{border-color:#ffffff;text-align:center;vertical-align:top}\n.tg .tg-b45e{background-color:#194A26;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-9xgv{background-color:#1553ae;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-3m6m{background-color:#f80c00;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2e1p{background-color:#db0e9a;border-color:#ffffff;color:#db0e9a;text-align:left;vertical-align:top}\n.tg .tg-l5fa{background-color:#FFF30D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2cns{background-color:#3DE6EB;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-jjsp{background-color:#FFF;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-2w6m{background-color:#8AB3A0;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nla7{background-color:#6B714F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-qg2z{background-color:#46E483;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-nv8o{background-color:#C5DC42;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-grz5{background-color:#F3A60D;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-bja1{background-color:#99F;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-69kt{background-color:#660082;border-color:#ffffff;text-align:left;vertical-align:top}\n.tg .tg-r1r4{background-color:#5F0;border-color:#ffffff;text-align:left;vertical-align:top}\n</style>\n<table class=\"tg\">\n<thead>\n <tr>\n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n <th class=\"tg-zv4m\"></th>\n <th class=\"tg-zv4m\">Class</th>\n <th class=\"tg-8jgo\">Train/val (%)</th>\n <th class=\"tg-8jgo\">Test flair#1 (%)</th>\n <th class=\"tg-8jgo\">Test flair#2 (%)</th> \n </tr>\n</thead>\n<tbody>\n <tr>\n <td class=\"tg-2e1p\"></td>\n <td class=\"tg-km2t\">(1) Building</td>\n <td class=\"tg-8jgo\">8.14</td>\n <td class=\"tg-8jgo\">8.6</td> \n <td class=\"tg-8jgo\">3.26</td>\n <td class=\"tg-l5fa\"></td>\n <td class=\"tg-km2t\">(11) Agricultural Land</td>\n <td class=\"tg-8jgo\">10.98</td>\n <td class=\"tg-8jgo\">6.95</td> \n <td class=\"tg-8jgo\">18.19</td>\n </tr>\n <tr>\n <td class=\"tg-9efv\"></td>\n <td class=\"tg-km2t\">(2) Pervious surface</td>\n <td class=\"tg-8jgo\">8.25</td>\n <td class=\"tg-8jgo\">7.34</td> \n <td class=\"tg-8jgo\">3.82</td>\n <td class=\"tg-rime\"></td>\n <td class=\"tg-km2t\">(12) Plowed land</td>\n <td class=\"tg-8jgo\">3.88</td>\n <td class=\"tg-8jgo\">2.25</td> \n <td class=\"tg-8jgo\">1.81</td>\n </tr>\n <tr>\n <td class=\"tg-3m6m\"></td>\n <td class=\"tg-km2t\">(3) Impervious surface</td>\n <td class=\"tg-8jgo\">13.72</td>\n <td class=\"tg-8jgo\">14.98</td> \n <td class=\"tg-8jgo\">5.87</td>\n <td class=\"tg-2cns\"></td>\n <td class=\"tg-km2t\">(13) Swimming pool</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.04</td> \n <td class=\"tg-8jgo\">0.02</td>\n </tr>\n <tr>\n <td class=\"tg-r3rw\"></td>\n <td class=\"tg-km2t\">(4) Bare soil</td>\n <td class=\"tg-8jgo\">3.47</td>\n <td class=\"tg-8jgo\">4.36</td> \n <td class=\"tg-8jgo\">1.6</td>\n <td class=\"tg-jjsp\"></td>\n <td class=\"tg-km2t\">(14) Snow</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-9xgv\"></td>\n <td class=\"tg-km2t\">(5) Water</td>\n <td class=\"tg-8jgo\">4.88</td>\n <td class=\"tg-8jgo\">5.98</td> \n <td class=\"tg-8jgo\">3.17</td>\n <td class=\"tg-2w6m\"></td>\n <td class=\"tg-km2t\">(15) Clear cut</td>\n <td class=\"tg-8jgo\">0.15</td>\n <td class=\"tg-8jgo\">0.01</td> \n <td class=\"tg-8jgo\">0.82</td>\n </tr>\n <tr>\n <td class=\"tg-b45e\"></td>\n <td class=\"tg-km2t\">(6) Coniferous</td>\n <td class=\"tg-8jgo\">2.74</td>\n <td class=\"tg-8jgo\">2.39</td>\n <td class=\"tg-8jgo\">10.24</td>\n <td class=\"tg-nla7\"></td>\n <td class=\"tg-km2t\">(16) Mixed</td>\n <td class=\"tg-8jgo\">0.05</td>\n <td class=\"tg-8jgo\">-</td> \n <td class=\"tg-8jgo\">0.12</td>\n </tr>\n <tr>\n <td class=\"tg-qg2z\"></td>\n <td class=\"tg-km2t\">(7) Deciduous</td>\n <td class=\"tg-8jgo\">15.38</td>\n <td class=\"tg-8jgo\">13.91</td>\n <td class=\"tg-8jgo\">24.79</td>\n <td class=\"tg-nv8o\"></td>\n <td class=\"tg-km2t\">(17) Ligneous</td>\n <td class=\"tg-8jgo\">0.01</td>\n <td class=\"tg-8jgo\">0.03</td>\n <td class=\"tg-8jgo\">-</td>\n </tr>\n <tr>\n <td class=\"tg-grz5\"></td>\n <td class=\"tg-km2t\">(8) Brushwood</td>\n <td class=\"tg-8jgo\">6.95</td>\n <td class=\"tg-8jgo\">6.91</td>\n <td class=\"tg-8jgo\">3.81</td> \n <td class=\"tg-bja1\"></td>\n <td class=\"tg-km2t\">(18) Greenhouse</td>\n <td class=\"tg-8jgo\">0.12</td>\n <td class=\"tg-8jgo\">0.2</td> \n <td class=\"tg-8jgo\">0.15</td>\n </tr>\n <tr>\n <td class=\"tg-69kt\"></td>\n <td class=\"tg-km2t\">(9) Vineyard</td>\n <td class=\"tg-8jgo\">3.13</td>\n <td class=\"tg-8jgo\">3.87</td>\n <td class=\"tg-8jgo\">2.55</td>\n <td class=\"tg-nto1\"></td>\n <td class=\"tg-km2t\">(19) Other</td>\n <td class=\"tg-8jgo\">0.14</td>\n <td class=\"tg-8jgo\">0.-</td> \n <td class=\"tg-8jgo\">0.04</td>\n </tr>\n <tr>\n <td class=\"tg-r1r4\"></td>\n <td class=\"tg-km2t\">(10) Herbaceous vegetation</td>\n <td class=\"tg-8jgo\">17.84</td>\n <td class=\"tg-8jgo\">22.17</td>\n <td class=\"tg-8jgo\">19.76</td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n <td class=\"tg-zv4m\"></td>\n </tr>\n</tbody>\n</table>\n\n<br><br>## Dataset Structure\n<hr style='margin-top:-1em; margin-bottom:0' />\nThe FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.\n\nEach patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m \nand associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).\n\n<p align=\"center\"><img src=\"URL\" alt=\"\" style=\"width:70%;max-width:600px;\"/></p><br>",
"passage: ### Band order\n\n<div style=\"display: flex;\">\n<div style=\"width: 15%;margin-right: 1;\"\">\nAerial\n<ul>\n<li>1. Red</li>\n<li>2. Green</li>\n<li>3. Blue</li>\n<li>4. NIR</li>\n<li>5. nDSM</li>\n</ul>\n</div>\n\n<div style=\"width: 25%;\">\nSatellite\n<ul>\n<li>1. Blue (B2 490nm)</li>\n<li>2. Green (B3 560nm)</li>\n<li>3. Red (B4 665nm)</li>\n<li>4. Red-Edge (B5 705nm)</li>\n<li>5. Red-Edge2 (B6 470nm)</li>\n<li>6. Red-Edge3 (B7 783nm)</li>\n<li>7. NIR (B8 842nm)</li>\n<li>8. NIR-Red-Edge (B8a 865nm)</li>\n<li>9. SWIR (B11 1610nm)</li>\n<li>10. SWIR2 (B12 2190nm)</li>\n</ul>\n</div>\n\n</div>### Annotations\nEach pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN. \nMovable objects like cars or boats are annotated according to their underlying cover.",
"passage: ### Data Splits\nThe dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements. \nFor our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test. \nIt can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain. \nThis arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set. \nConsequently, every split accurately reflects the landscape diversity inherent to metropolitan France. \nIt is important to mention that the patches come with meta-data permitting alternative splitting schemes. \n\n\nOfficial domain split: <br/>\n\n<div style=\"display: flex; flex-wrap: nowrap; align-items: center\">\n <div style=\"flex: 40%;\">\n <img src=\"URL\" alt=\"flair-splits\">\n</div>\n\n <div style=\"flex: 60%; margin: auto;\"\">\n <table border=\"1\">\n <tr>\n <th><font color=\"#c7254e\">TRAIN:</font></th>\n <td>D006, D007, D008, D009, D013, D016, D017, D021, D023, D030, D032, D033, D034, D035, D038, D041, D044, D046, D049, D051, D052, D055, D060, D063, D070, D072, D074, D078, D080, D081, D086, D091</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">VALIDATION:</font></th>\n <td>D004, D014, D029, D031, D058, D066, D067, D077</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#1:</font></th>\n <td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>\n </tr>\n <tr>\n <th><font color=\"#c7254e\">TEST-flair#2:</font></th>\n <td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>\n </tr> \n </table>\n </div>\n</div>\n\n<br><br>## Baseline code \n<hr style='margin-top:-1em; margin-bottom:0' />\n<br>### Flair #1 (aerial only)\nA U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines. \nThe used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques. \n\nFlair#1 code repository 📁 : URL\nLink to the paper : URL <br>\n\nPlease include a citation to the following article if you use the FLAIR#1 dataset:\n\n\n<br>"
] |
617d3c8b6dd481246d677e9a36202bd7a9a72b27
|
from https://github.com/vinid/instruction-llms-safety-eval/tree/main (not mine)
just for personal usage
thanks
|
SuzakuinTsubaki/safepaca
|
[
"license:apache-2.0",
"region:us"
] |
2023-10-26T07:42:37+00:00
|
{"license": "apache-2.0"}
|
2023-10-26T08:02:19+00:00
|
[] |
[] |
TAGS
#license-apache-2.0 #region-us
|
from URL (not mine)
just for personal usage
thanks
|
[] |
[
"TAGS\n#license-apache-2.0 #region-us \n"
] |
[
14
] |
[
"passage: TAGS\n#license-apache-2.0 #region-us \n"
] |
3a77ac2174e45129d7e0f2655a502b7c9d27d992
|
# Dataset Card for "cot-large"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
quyanh/cot-large
|
[
"region:us"
] |
2023-10-26T07:45:21+00:00
|
{"dataset_info": {"features": [{"name": "inputs", "dtype": "string"}, {"name": "response", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 14568291.0, "num_examples": 35873}], "download_size": 8626487, "dataset_size": 14568291.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T07:47:09+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "cot-large"
More Information needed
|
[
"# Dataset Card for \"cot-large\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"cot-large\"\n\nMore Information needed"
] |
[
6,
14
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"cot-large\"\n\nMore Information needed"
] |
8781ac12b1ac2c348ceabd9f8ceeb49e80c7db0e
|
# Dataset Card for "helm-samsum-dolly-lima-cot-large"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
quyanh/helm-samsum-dolly-lima-cot-large
|
[
"region:us"
] |
2023-10-26T07:48:56+00:00
|
{"dataset_info": {"features": [{"name": "prompt", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 43358866.34944185, "num_examples": 57836}], "download_size": 25751305, "dataset_size": 43358866.34944185}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T07:50:28+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "helm-samsum-dolly-lima-cot-large"
More Information needed
|
[
"# Dataset Card for \"helm-samsum-dolly-lima-cot-large\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"helm-samsum-dolly-lima-cot-large\"\n\nMore Information needed"
] |
[
6,
25
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"helm-samsum-dolly-lima-cot-large\"\n\nMore Information needed"
] |
fb79ccc110aabcdc26dc691679bdca4bde37492d
|
# Computed Tomography (CT) of the Spine - Scoliosis
The dataset consists of CT spine scans of people with **scoliosis**. images that aid in the assessment and diagnosis of scoliosis. Each scan consists of multiple slices capturing various sections of the spine, including the **cervical (neck), thoracic (upper back), and lumbar (lower back) regions**. The data are presented in 2 different formats: **.jpg and .dcm**.
The dataset of CT spine scans is valuable for research in **automated scoliosis detection, scoliosis segmentation and scoliosis classification**.
This dataset may contribute to the development of *treatment planning techniques, surgical interventions, and monitoring strategies* for patients with scoliosis.
# Get the Dataset
## This is just an example of the data
Leave a request on [https://trainingdata.pro/data-market](https://trainingdata.pro/data-market/spine-x-ray-image?utm_source=huggingface&utm_medium=cpc&utm_campaign=ct-of-the-spine) to discuss your requirements, learn about the price and buy the dataset
# Content
### The folder "files" includes 4 folders:
- corresponding to each person with scoliosis
- including spine scans in 2 different formats: **.jpg and .dcm**.
### File with the extension .csv includes the following information for each media file:
- **dcm**: link to access the .dcm file,
- **jpg**: link to access the .jpg file,
# Medical data might be collected in accordance with your requirements.
## [TrainingData](https://trainingdata.pro/data-market/spine-x-ray-image?utm_source=huggingface&utm_medium=cpc&utm_campaign=ct-of-the-spine) provides high-quality data annotation tailored to your needs
More datasets in TrainingData's Kaggle account: **https://www.kaggle.com/trainingdatapro/datasets**
TrainingData's GitHub: **https://github.com/trainingdata-pro**
*keywords: scoliosis detection, scoliosis segmentation, scoliosis classification, scoliosis image dataset, medical imaging, radiology dataset, vertebral segmentation dataset, spine deformity dataset. thoracic spine, lumbar spine, abnormal spinal curvature, adolescent idiopathic scoliosis, congenital scoliosis, juvenile scoliosis, cobb angle ct, vertebral rotation, scoliotic curve dataset, posterior-anterior projection, lateral projection, spondylolisthesis*
|
TrainingDataPro/ct-of-the-spine-scoliosis
|
[
"task_categories:image-classification",
"task_categories:image-to-image",
"language:en",
"license:cc-by-nc-nd-4.0",
"medical",
"code",
"region:us"
] |
2023-10-26T08:05:25+00:00
|
{"language": ["en"], "license": "cc-by-nc-nd-4.0", "task_categories": ["image-classification", "image-to-image"], "tags": ["medical", "code"]}
|
2023-10-26T09:18:48+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-image-classification #task_categories-image-to-image #language-English #license-cc-by-nc-nd-4.0 #medical #code #region-us
|
# Computed Tomography (CT) of the Spine - Scoliosis
The dataset consists of CT spine scans of people with scoliosis. images that aid in the assessment and diagnosis of scoliosis. Each scan consists of multiple slices capturing various sections of the spine, including the cervical (neck), thoracic (upper back), and lumbar (lower back) regions. The data are presented in 2 different formats: .jpg and .dcm.
The dataset of CT spine scans is valuable for research in automated scoliosis detection, scoliosis segmentation and scoliosis classification.
![](URL
This dataset may contribute to the development of *treatment planning techniques, surgical interventions, and monitoring strategies* for patients with scoliosis.
# Get the Dataset
## This is just an example of the data
Leave a request on URL to discuss your requirements, learn about the price and buy the dataset
# Content
### The folder "files" includes 4 folders:
- corresponding to each person with scoliosis
- including spine scans in 2 different formats: .jpg and .dcm.
### File with the extension .csv includes the following information for each media file:
- dcm: link to access the .dcm file,
- jpg: link to access the .jpg file,
# Medical data might be collected in accordance with your requirements.
## TrainingData provides high-quality data annotation tailored to your needs
More datasets in TrainingData's Kaggle account: URL
TrainingData's GitHub: URL
*keywords: scoliosis detection, scoliosis segmentation, scoliosis classification, scoliosis image dataset, medical imaging, radiology dataset, vertebral segmentation dataset, spine deformity dataset. thoracic spine, lumbar spine, abnormal spinal curvature, adolescent idiopathic scoliosis, congenital scoliosis, juvenile scoliosis, cobb angle ct, vertebral rotation, scoliotic curve dataset, posterior-anterior projection, lateral projection, spondylolisthesis*
|
[
"# Computed Tomography (CT) of the Spine - Scoliosis\n\nThe dataset consists of CT spine scans of people with scoliosis. images that aid in the assessment and diagnosis of scoliosis. Each scan consists of multiple slices capturing various sections of the spine, including the cervical (neck), thoracic (upper back), and lumbar (lower back) regions. The data are presented in 2 different formats: .jpg and .dcm. \n\nThe dataset of CT spine scans is valuable for research in automated scoliosis detection, scoliosis segmentation and scoliosis classification.\n\n![](URL\n\nThis dataset may contribute to the development of *treatment planning techniques, surgical interventions, and monitoring strategies* for patients with scoliosis.",
"# Get the Dataset",
"## This is just an example of the data \n\nLeave a request on URL to discuss your requirements, learn about the price and buy the dataset",
"# Content",
"### The folder \"files\" includes 4 folders:\n- corresponding to each person with scoliosis\n- including spine scans in 2 different formats: .jpg and .dcm.",
"### File with the extension .csv includes the following information for each media file:\n\n- dcm: link to access the .dcm file,\n- jpg: link to access the .jpg file,",
"# Medical data might be collected in accordance with your requirements.",
"## TrainingData provides high-quality data annotation tailored to your needs\n\nMore datasets in TrainingData's Kaggle account: URL\n\nTrainingData's GitHub: URL\n\n*keywords: scoliosis detection, scoliosis segmentation, scoliosis classification, scoliosis image dataset, medical imaging, radiology dataset, vertebral segmentation dataset, spine deformity dataset. thoracic spine, lumbar spine, abnormal spinal curvature, adolescent idiopathic scoliosis, congenital scoliosis, juvenile scoliosis, cobb angle ct, vertebral rotation, scoliotic curve dataset, posterior-anterior projection, lateral projection, spondylolisthesis*"
] |
[
"TAGS\n#task_categories-image-classification #task_categories-image-to-image #language-English #license-cc-by-nc-nd-4.0 #medical #code #region-us \n",
"# Computed Tomography (CT) of the Spine - Scoliosis\n\nThe dataset consists of CT spine scans of people with scoliosis. images that aid in the assessment and diagnosis of scoliosis. Each scan consists of multiple slices capturing various sections of the spine, including the cervical (neck), thoracic (upper back), and lumbar (lower back) regions. The data are presented in 2 different formats: .jpg and .dcm. \n\nThe dataset of CT spine scans is valuable for research in automated scoliosis detection, scoliosis segmentation and scoliosis classification.\n\n![](URL\n\nThis dataset may contribute to the development of *treatment planning techniques, surgical interventions, and monitoring strategies* for patients with scoliosis.",
"# Get the Dataset",
"## This is just an example of the data \n\nLeave a request on URL to discuss your requirements, learn about the price and buy the dataset",
"# Content",
"### The folder \"files\" includes 4 folders:\n- corresponding to each person with scoliosis\n- including spine scans in 2 different formats: .jpg and .dcm.",
"### File with the extension .csv includes the following information for each media file:\n\n- dcm: link to access the .dcm file,\n- jpg: link to access the .jpg file,",
"# Medical data might be collected in accordance with your requirements.",
"## TrainingData provides high-quality data annotation tailored to your needs\n\nMore datasets in TrainingData's Kaggle account: URL\n\nTrainingData's GitHub: URL\n\n*keywords: scoliosis detection, scoliosis segmentation, scoliosis classification, scoliosis image dataset, medical imaging, radiology dataset, vertebral segmentation dataset, spine deformity dataset. thoracic spine, lumbar spine, abnormal spinal curvature, adolescent idiopathic scoliosis, congenital scoliosis, juvenile scoliosis, cobb angle ct, vertebral rotation, scoliotic curve dataset, posterior-anterior projection, lateral projection, spondylolisthesis*"
] |
[
51,
188,
5,
28,
2,
44,
46,
13,
171
] |
[
"passage: TAGS\n#task_categories-image-classification #task_categories-image-to-image #language-English #license-cc-by-nc-nd-4.0 #medical #code #region-us \n# Computed Tomography (CT) of the Spine - Scoliosis\n\nThe dataset consists of CT spine scans of people with scoliosis. images that aid in the assessment and diagnosis of scoliosis. Each scan consists of multiple slices capturing various sections of the spine, including the cervical (neck), thoracic (upper back), and lumbar (lower back) regions. The data are presented in 2 different formats: .jpg and .dcm. \n\nThe dataset of CT spine scans is valuable for research in automated scoliosis detection, scoliosis segmentation and scoliosis classification.\n\n![](URL\n\nThis dataset may contribute to the development of *treatment planning techniques, surgical interventions, and monitoring strategies* for patients with scoliosis.# Get the Dataset## This is just an example of the data \n\nLeave a request on URL to discuss your requirements, learn about the price and buy the dataset# Content### The folder \"files\" includes 4 folders:\n- corresponding to each person with scoliosis\n- including spine scans in 2 different formats: .jpg and .dcm.### File with the extension .csv includes the following information for each media file:\n\n- dcm: link to access the .dcm file,\n- jpg: link to access the .jpg file,# Medical data might be collected in accordance with your requirements."
] |
30458627bb96441167b65861b6b64f5c8b89cf83
|
### Institute for the Study of Contemporary Antisemitism (ISCA) at Indiana University Dataset on bias against Asians, Blacks, Jews, Latines, and Muslims
The ISCA project compiled this dataset using an annotation portal, which was used to label tweets as either biased or non-biased, among other labels. Note that the annotation was done on live data, including images and context, such as threads. The original data comes from annotationportal.com. They include representative samples of live tweets from the years 2020 and 2021 with the keywords "Asians, Blacks, Jews, Latinos, and Muslims".
A random sample of 600 tweets per year was drawn for each of the keywords. This includes retweets. Due to a sampling error, the sample for the year 2021 for the keyword "Jews" has only 453 tweets from 2021 and 147 from the first eight months of 2022 and it includes some tweets from the query with the keyword "Israel." The tweets were divided into six samples of 100 tweets, which were then annotated by three to seven students in the class "Researching White Supremacism and Antisemitism on Social Media" taught by Gunther Jikeli, Elisha S. Breton, and Seth Moller at Indiana University in the fall of 2022, see this report. Annotators used a scale from 1 to 5 (confident not biased, probably not biased, don't know, probably biased, confident biased). The definitions of bias against each minority group used for annotation are also included in the report.
If a tweet called out or denounced bias against the minority in question, it was labeled as "calling out bias."
The labels of whether a tweet is biased or calls out bias are based on a 75% majority vote. We considered "probably biased" and "confident biased" as biased and "confident not biased," "probably not biased," and "don't know" as not biased.
The types of stereotypes vary widely across the different categories of prejudice. While about a third of all biased tweets were classified as "hate" against the minority, the stereotypes in the tweets often matched common stereotypes about the minority. Asians were blamed for the Covid pandemic. Blacks were seen as inferior and associated with crime. Jews were seen as powerful and held collectively responsible for the actions of the State of Israel. Some tweets denied the Holocaust. Hispanics/Latines were portrayed as being in the country illegally and as "invaders," in addition to stereotypical accusations of being lazy, stupid, or having too many children. Muslims, on the other hand, were often collectively blamed for terrorism and violence, though often in conversations about Muslims in India.
# Content:
This dataset contains 5880 tweets that cover a wide range of topics common in conversations about Asians, Blacks, Jews, Latines, and Muslims. 357 tweets (6.1 %) are labeled as biased and 5523 (93.9 %) are labeled as not biased. 1365 tweets (23.2 %) are labeled as calling out or denouncing bias.
1180 out of 5880 tweets (20.1 %) contain the keyword "Asians," 590 were posted in 2020 and 590 in 2021. 39 tweets (3.3 %) are biased against Asian people. 370 tweets (31,4 %) call out bias against Asians.
1160 out of 5880 tweets (19.7%) contain the keyword "Blacks," 578 were posted in 2020 and 582 in 2021. 101 tweets (8.7 %) are biased against Black people. 334 tweets (28.8 %) call out bias against Blacks.
1189 out of 5880 tweets (20.2 %) contain the keyword "Jews," 592 were posted in 2020, 451 in 2021, and ––as mentioned above––146 tweets from 2022. 83 tweets (7 %) are biased against Jewish people. 220 tweets (18.5 %) call out bias against Jews.
1169 out of 5880 tweets (19.9 %) contain the keyword "Latinos," 584 were posted in 2020 and 585 in 2021. 29 tweets (2.5 %) are biased against Latines. 181 tweets (15.5 %) call out bias against Latines.
1182 out of 5880 tweets (20.1 %) contain the keyword "Muslims," 593 were posted in 2020 and 589 in 2021. 105 tweets (8.9 %) are biased against Muslims. 260 tweets (22 %) call out bias against Muslims.
# File Description:
The dataset is provided in a csv file format, with each row representing a single message, including replies, quotes, and retweets. The file contains the following columns:
'TweetID': Represents the tweet ID.
'Username': Represents the username who published the tweet (if it is a retweet, it will be the user who retweetet the original tweet.
'Text': Represents the full text of the tweet (not pre-processed).
'CreateDate': Represents the date the tweet was created.
'Biased': Represents the labeled by our annotators if the tweet is biased (1) or not (0).
'Calling_Out': Represents the label by our annotators if the tweet is calling out bias against minority groups (1) or not (0).
'Keyword': Represents the keyword that was used in the query. The keyword can be in the text, including mentioned names, or the username.
# Licences
Data is published under the terms of the "Creative Commons Attribution 4.0 International" licence (https://creativecommons.org/licenses/by/4.0)
# Acknowledgements
We are grateful for the technical collaboration with Indiana University's Observatory on Social Media (OSoMe). We thank all class participants for the annotations and contributions, including Kate Baba, Eleni Ballis, Garrett Banuelos, Savannah Benjamin, Luke Bianco, Zoe Bogan, Elisha S. Breton, Aidan Calderaro, Anaye Caldron, Olivia Cozzi, Daj Crisler, Jenna Eidson, Ella Fanning, Victoria Ford, Jess Gruettner, Ronan Hancock, Isabel Hawes, Brennan Hensler, Kyra Horton, Maxwell Idczak, Sanjana Iyer, Jacob Joffe, Katie Johnson, Allison Jones, Kassidy Keltner, Sophia Knoll, Jillian Kolesky, Emily Lowrey, Rachael Morara, Benjamin Nadolne, Rachel Neglia, Seungmin Oh, Kirsten Pecsenye, Sophia Perkovich, Joey Philpott, Katelin Ray, Kaleb Samuels, Chloe Sherman, Rachel Weber, Molly Winkeljohn, Ally Wolfgang, Rowan Wolke, Michael Wong, Jane Woods, Kaleb Woodworth, and Aurora Young.
This work used Jetstream2 at Indiana University through allocation HUM200003 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296.
|
ISCA-IUB/HateSpeechAndBias
|
[
"language:en",
"hate speech",
"bias",
"racism",
"asians",
"blacks",
"jews",
"latinos",
"muslims",
"region:us"
] |
2023-10-26T08:06:24+00:00
|
{"language": ["en"], "tags": ["hate speech", "bias", "racism", "asians", "blacks", "jews", "latinos", "muslims"]}
|
2023-10-26T08:09:13+00:00
|
[] |
[
"en"
] |
TAGS
#language-English #hate speech #bias #racism #asians #blacks #jews #latinos #muslims #region-us
|
### Institute for the Study of Contemporary Antisemitism (ISCA) at Indiana University Dataset on bias against Asians, Blacks, Jews, Latines, and Muslims
The ISCA project compiled this dataset using an annotation portal, which was used to label tweets as either biased or non-biased, among other labels. Note that the annotation was done on live data, including images and context, such as threads. The original data comes from URL. They include representative samples of live tweets from the years 2020 and 2021 with the keywords "Asians, Blacks, Jews, Latinos, and Muslims".
A random sample of 600 tweets per year was drawn for each of the keywords. This includes retweets. Due to a sampling error, the sample for the year 2021 for the keyword "Jews" has only 453 tweets from 2021 and 147 from the first eight months of 2022 and it includes some tweets from the query with the keyword "Israel." The tweets were divided into six samples of 100 tweets, which were then annotated by three to seven students in the class "Researching White Supremacism and Antisemitism on Social Media" taught by Gunther Jikeli, Elisha S. Breton, and Seth Moller at Indiana University in the fall of 2022, see this report. Annotators used a scale from 1 to 5 (confident not biased, probably not biased, don't know, probably biased, confident biased). The definitions of bias against each minority group used for annotation are also included in the report.
If a tweet called out or denounced bias against the minority in question, it was labeled as "calling out bias."
The labels of whether a tweet is biased or calls out bias are based on a 75% majority vote. We considered "probably biased" and "confident biased" as biased and "confident not biased," "probably not biased," and "don't know" as not biased.
The types of stereotypes vary widely across the different categories of prejudice. While about a third of all biased tweets were classified as "hate" against the minority, the stereotypes in the tweets often matched common stereotypes about the minority. Asians were blamed for the Covid pandemic. Blacks were seen as inferior and associated with crime. Jews were seen as powerful and held collectively responsible for the actions of the State of Israel. Some tweets denied the Holocaust. Hispanics/Latines were portrayed as being in the country illegally and as "invaders," in addition to stereotypical accusations of being lazy, stupid, or having too many children. Muslims, on the other hand, were often collectively blamed for terrorism and violence, though often in conversations about Muslims in India.
# Content:
This dataset contains 5880 tweets that cover a wide range of topics common in conversations about Asians, Blacks, Jews, Latines, and Muslims. 357 tweets (6.1 %) are labeled as biased and 5523 (93.9 %) are labeled as not biased. 1365 tweets (23.2 %) are labeled as calling out or denouncing bias.
1180 out of 5880 tweets (20.1 %) contain the keyword "Asians," 590 were posted in 2020 and 590 in 2021. 39 tweets (3.3 %) are biased against Asian people. 370 tweets (31,4 %) call out bias against Asians.
1160 out of 5880 tweets (19.7%) contain the keyword "Blacks," 578 were posted in 2020 and 582 in 2021. 101 tweets (8.7 %) are biased against Black people. 334 tweets (28.8 %) call out bias against Blacks.
1189 out of 5880 tweets (20.2 %) contain the keyword "Jews," 592 were posted in 2020, 451 in 2021, and ––as mentioned above––146 tweets from 2022. 83 tweets (7 %) are biased against Jewish people. 220 tweets (18.5 %) call out bias against Jews.
1169 out of 5880 tweets (19.9 %) contain the keyword "Latinos," 584 were posted in 2020 and 585 in 2021. 29 tweets (2.5 %) are biased against Latines. 181 tweets (15.5 %) call out bias against Latines.
1182 out of 5880 tweets (20.1 %) contain the keyword "Muslims," 593 were posted in 2020 and 589 in 2021. 105 tweets (8.9 %) are biased against Muslims. 260 tweets (22 %) call out bias against Muslims.
# File Description:
The dataset is provided in a csv file format, with each row representing a single message, including replies, quotes, and retweets. The file contains the following columns:
'TweetID': Represents the tweet ID.
'Username': Represents the username who published the tweet (if it is a retweet, it will be the user who retweetet the original tweet.
'Text': Represents the full text of the tweet (not pre-processed).
'CreateDate': Represents the date the tweet was created.
'Biased': Represents the labeled by our annotators if the tweet is biased (1) or not (0).
'Calling_Out': Represents the label by our annotators if the tweet is calling out bias against minority groups (1) or not (0).
'Keyword': Represents the keyword that was used in the query. The keyword can be in the text, including mentioned names, or the username.
# Licences
Data is published under the terms of the "Creative Commons Attribution 4.0 International" licence (URL
# Acknowledgements
We are grateful for the technical collaboration with Indiana University's Observatory on Social Media (OSoMe). We thank all class participants for the annotations and contributions, including Kate Baba, Eleni Ballis, Garrett Banuelos, Savannah Benjamin, Luke Bianco, Zoe Bogan, Elisha S. Breton, Aidan Calderaro, Anaye Caldron, Olivia Cozzi, Daj Crisler, Jenna Eidson, Ella Fanning, Victoria Ford, Jess Gruettner, Ronan Hancock, Isabel Hawes, Brennan Hensler, Kyra Horton, Maxwell Idczak, Sanjana Iyer, Jacob Joffe, Katie Johnson, Allison Jones, Kassidy Keltner, Sophia Knoll, Jillian Kolesky, Emily Lowrey, Rachael Morara, Benjamin Nadolne, Rachel Neglia, Seungmin Oh, Kirsten Pecsenye, Sophia Perkovich, Joey Philpott, Katelin Ray, Kaleb Samuels, Chloe Sherman, Rachel Weber, Molly Winkeljohn, Ally Wolfgang, Rowan Wolke, Michael Wong, Jane Woods, Kaleb Woodworth, and Aurora Young.
This work used Jetstream2 at Indiana University through allocation HUM200003 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296.
|
[
"### Institute for the Study of Contemporary Antisemitism (ISCA) at Indiana University Dataset on bias against Asians, Blacks, Jews, Latines, and Muslims \n\n \n\nThe ISCA project compiled this dataset using an annotation portal, which was used to label tweets as either biased or non-biased, among other labels. Note that the annotation was done on live data, including images and context, such as threads. The original data comes from URL. They include representative samples of live tweets from the years 2020 and 2021 with the keywords \"Asians, Blacks, Jews, Latinos, and Muslims\". \n\nA random sample of 600 tweets per year was drawn for each of the keywords. This includes retweets. Due to a sampling error, the sample for the year 2021 for the keyword \"Jews\" has only 453 tweets from 2021 and 147 from the first eight months of 2022 and it includes some tweets from the query with the keyword \"Israel.\" The tweets were divided into six samples of 100 tweets, which were then annotated by three to seven students in the class \"Researching White Supremacism and Antisemitism on Social Media\" taught by Gunther Jikeli, Elisha S. Breton, and Seth Moller at Indiana University in the fall of 2022, see this report. Annotators used a scale from 1 to 5 (confident not biased, probably not biased, don't know, probably biased, confident biased). The definitions of bias against each minority group used for annotation are also included in the report. \n\nIf a tweet called out or denounced bias against the minority in question, it was labeled as \"calling out bias.\" \n\nThe labels of whether a tweet is biased or calls out bias are based on a 75% majority vote. We considered \"probably biased\" and \"confident biased\" as biased and \"confident not biased,\" \"probably not biased,\" and \"don't know\" as not biased. \n\nThe types of stereotypes vary widely across the different categories of prejudice. While about a third of all biased tweets were classified as \"hate\" against the minority, the stereotypes in the tweets often matched common stereotypes about the minority. Asians were blamed for the Covid pandemic. Blacks were seen as inferior and associated with crime. Jews were seen as powerful and held collectively responsible for the actions of the State of Israel. Some tweets denied the Holocaust. Hispanics/Latines were portrayed as being in the country illegally and as \"invaders,\" in addition to stereotypical accusations of being lazy, stupid, or having too many children. Muslims, on the other hand, were often collectively blamed for terrorism and violence, though often in conversations about Muslims in India.",
"# Content: \n \n\nThis dataset contains 5880 tweets that cover a wide range of topics common in conversations about Asians, Blacks, Jews, Latines, and Muslims. 357 tweets (6.1 %) are labeled as biased and 5523 (93.9 %) are labeled as not biased. 1365 tweets (23.2 %) are labeled as calling out or denouncing bias. \n\n1180 out of 5880 tweets (20.1 %) contain the keyword \"Asians,\" 590 were posted in 2020 and 590 in 2021. 39 tweets (3.3 %) are biased against Asian people. 370 tweets (31,4 %) call out bias against Asians. \n\n1160 out of 5880 tweets (19.7%) contain the keyword \"Blacks,\" 578 were posted in 2020 and 582 in 2021. 101 tweets (8.7 %) are biased against Black people. 334 tweets (28.8 %) call out bias against Blacks. \n\n1189 out of 5880 tweets (20.2 %) contain the keyword \"Jews,\" 592 were posted in 2020, 451 in 2021, and ––as mentioned above––146 tweets from 2022. 83 tweets (7 %) are biased against Jewish people. 220 tweets (18.5 %) call out bias against Jews.\n\n1169 out of 5880 tweets (19.9 %) contain the keyword \"Latinos,\" 584 were posted in 2020 and 585 in 2021. 29 tweets (2.5 %) are biased against Latines. 181 tweets (15.5 %) call out bias against Latines. \n\n1182 out of 5880 tweets (20.1 %) contain the keyword \"Muslims,\" 593 were posted in 2020 and 589 in 2021. 105 tweets (8.9 %) are biased against Muslims. 260 tweets (22 %) call out bias against Muslims.",
"# File Description: \n\nThe dataset is provided in a csv file format, with each row representing a single message, including replies, quotes, and retweets. The file contains the following columns: \n\n \n'TweetID': Represents the tweet ID. \n\n'Username': Represents the username who published the tweet (if it is a retweet, it will be the user who retweetet the original tweet. \n\n'Text': Represents the full text of the tweet (not pre-processed). \n\n'CreateDate': Represents the date the tweet was created. \n\n'Biased': Represents the labeled by our annotators if the tweet is biased (1) or not (0). \n\n'Calling_Out': Represents the label by our annotators if the tweet is calling out bias against minority groups (1) or not (0). \n\n'Keyword': Represents the keyword that was used in the query. The keyword can be in the text, including mentioned names, or the username.",
"# Licences \n\nData is published under the terms of the \"Creative Commons Attribution 4.0 International\" licence (URL",
"# Acknowledgements \n\nWe are grateful for the technical collaboration with Indiana University's Observatory on Social Media (OSoMe). We thank all class participants for the annotations and contributions, including Kate Baba, Eleni Ballis, Garrett Banuelos, Savannah Benjamin, Luke Bianco, Zoe Bogan, Elisha S. Breton, Aidan Calderaro, Anaye Caldron, Olivia Cozzi, Daj Crisler, Jenna Eidson, Ella Fanning, Victoria Ford, Jess Gruettner, Ronan Hancock, Isabel Hawes, Brennan Hensler, Kyra Horton, Maxwell Idczak, Sanjana Iyer, Jacob Joffe, Katie Johnson, Allison Jones, Kassidy Keltner, Sophia Knoll, Jillian Kolesky, Emily Lowrey, Rachael Morara, Benjamin Nadolne, Rachel Neglia, Seungmin Oh, Kirsten Pecsenye, Sophia Perkovich, Joey Philpott, Katelin Ray, Kaleb Samuels, Chloe Sherman, Rachel Weber, Molly Winkeljohn, Ally Wolfgang, Rowan Wolke, Michael Wong, Jane Woods, Kaleb Woodworth, and Aurora Young. \n\nThis work used Jetstream2 at Indiana University through allocation HUM200003 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296."
] |
[
"TAGS\n#language-English #hate speech #bias #racism #asians #blacks #jews #latinos #muslims #region-us \n",
"### Institute for the Study of Contemporary Antisemitism (ISCA) at Indiana University Dataset on bias against Asians, Blacks, Jews, Latines, and Muslims \n\n \n\nThe ISCA project compiled this dataset using an annotation portal, which was used to label tweets as either biased or non-biased, among other labels. Note that the annotation was done on live data, including images and context, such as threads. The original data comes from URL. They include representative samples of live tweets from the years 2020 and 2021 with the keywords \"Asians, Blacks, Jews, Latinos, and Muslims\". \n\nA random sample of 600 tweets per year was drawn for each of the keywords. This includes retweets. Due to a sampling error, the sample for the year 2021 for the keyword \"Jews\" has only 453 tweets from 2021 and 147 from the first eight months of 2022 and it includes some tweets from the query with the keyword \"Israel.\" The tweets were divided into six samples of 100 tweets, which were then annotated by three to seven students in the class \"Researching White Supremacism and Antisemitism on Social Media\" taught by Gunther Jikeli, Elisha S. Breton, and Seth Moller at Indiana University in the fall of 2022, see this report. Annotators used a scale from 1 to 5 (confident not biased, probably not biased, don't know, probably biased, confident biased). The definitions of bias against each minority group used for annotation are also included in the report. \n\nIf a tweet called out or denounced bias against the minority in question, it was labeled as \"calling out bias.\" \n\nThe labels of whether a tweet is biased or calls out bias are based on a 75% majority vote. We considered \"probably biased\" and \"confident biased\" as biased and \"confident not biased,\" \"probably not biased,\" and \"don't know\" as not biased. \n\nThe types of stereotypes vary widely across the different categories of prejudice. While about a third of all biased tweets were classified as \"hate\" against the minority, the stereotypes in the tweets often matched common stereotypes about the minority. Asians were blamed for the Covid pandemic. Blacks were seen as inferior and associated with crime. Jews were seen as powerful and held collectively responsible for the actions of the State of Israel. Some tweets denied the Holocaust. Hispanics/Latines were portrayed as being in the country illegally and as \"invaders,\" in addition to stereotypical accusations of being lazy, stupid, or having too many children. Muslims, on the other hand, were often collectively blamed for terrorism and violence, though often in conversations about Muslims in India.",
"# Content: \n \n\nThis dataset contains 5880 tweets that cover a wide range of topics common in conversations about Asians, Blacks, Jews, Latines, and Muslims. 357 tweets (6.1 %) are labeled as biased and 5523 (93.9 %) are labeled as not biased. 1365 tweets (23.2 %) are labeled as calling out or denouncing bias. \n\n1180 out of 5880 tweets (20.1 %) contain the keyword \"Asians,\" 590 were posted in 2020 and 590 in 2021. 39 tweets (3.3 %) are biased against Asian people. 370 tweets (31,4 %) call out bias against Asians. \n\n1160 out of 5880 tweets (19.7%) contain the keyword \"Blacks,\" 578 were posted in 2020 and 582 in 2021. 101 tweets (8.7 %) are biased against Black people. 334 tweets (28.8 %) call out bias against Blacks. \n\n1189 out of 5880 tweets (20.2 %) contain the keyword \"Jews,\" 592 were posted in 2020, 451 in 2021, and ––as mentioned above––146 tweets from 2022. 83 tweets (7 %) are biased against Jewish people. 220 tweets (18.5 %) call out bias against Jews.\n\n1169 out of 5880 tweets (19.9 %) contain the keyword \"Latinos,\" 584 were posted in 2020 and 585 in 2021. 29 tweets (2.5 %) are biased against Latines. 181 tweets (15.5 %) call out bias against Latines. \n\n1182 out of 5880 tweets (20.1 %) contain the keyword \"Muslims,\" 593 were posted in 2020 and 589 in 2021. 105 tweets (8.9 %) are biased against Muslims. 260 tweets (22 %) call out bias against Muslims.",
"# File Description: \n\nThe dataset is provided in a csv file format, with each row representing a single message, including replies, quotes, and retweets. The file contains the following columns: \n\n \n'TweetID': Represents the tweet ID. \n\n'Username': Represents the username who published the tweet (if it is a retweet, it will be the user who retweetet the original tweet. \n\n'Text': Represents the full text of the tweet (not pre-processed). \n\n'CreateDate': Represents the date the tweet was created. \n\n'Biased': Represents the labeled by our annotators if the tweet is biased (1) or not (0). \n\n'Calling_Out': Represents the label by our annotators if the tweet is calling out bias against minority groups (1) or not (0). \n\n'Keyword': Represents the keyword that was used in the query. The keyword can be in the text, including mentioned names, or the username.",
"# Licences \n\nData is published under the terms of the \"Creative Commons Attribution 4.0 International\" licence (URL",
"# Acknowledgements \n\nWe are grateful for the technical collaboration with Indiana University's Observatory on Social Media (OSoMe). We thank all class participants for the annotations and contributions, including Kate Baba, Eleni Ballis, Garrett Banuelos, Savannah Benjamin, Luke Bianco, Zoe Bogan, Elisha S. Breton, Aidan Calderaro, Anaye Caldron, Olivia Cozzi, Daj Crisler, Jenna Eidson, Ella Fanning, Victoria Ford, Jess Gruettner, Ronan Hancock, Isabel Hawes, Brennan Hensler, Kyra Horton, Maxwell Idczak, Sanjana Iyer, Jacob Joffe, Katie Johnson, Allison Jones, Kassidy Keltner, Sophia Knoll, Jillian Kolesky, Emily Lowrey, Rachael Morara, Benjamin Nadolne, Rachel Neglia, Seungmin Oh, Kirsten Pecsenye, Sophia Perkovich, Joey Philpott, Katelin Ray, Kaleb Samuels, Chloe Sherman, Rachel Weber, Molly Winkeljohn, Ally Wolfgang, Rowan Wolke, Michael Wong, Jane Woods, Kaleb Woodworth, and Aurora Young. \n\nThis work used Jetstream2 at Indiana University through allocation HUM200003 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296."
] |
[
36,
651,
396,
232,
21,
333
] |
[
"passage: TAGS\n#language-English #hate speech #bias #racism #asians #blacks #jews #latinos #muslims #region-us \n",
"passage: ### Institute for the Study of Contemporary Antisemitism (ISCA) at Indiana University Dataset on bias against Asians, Blacks, Jews, Latines, and Muslims \n\n \n\nThe ISCA project compiled this dataset using an annotation portal, which was used to label tweets as either biased or non-biased, among other labels. Note that the annotation was done on live data, including images and context, such as threads. The original data comes from URL. They include representative samples of live tweets from the years 2020 and 2021 with the keywords \"Asians, Blacks, Jews, Latinos, and Muslims\". \n\nA random sample of 600 tweets per year was drawn for each of the keywords. This includes retweets. Due to a sampling error, the sample for the year 2021 for the keyword \"Jews\" has only 453 tweets from 2021 and 147 from the first eight months of 2022 and it includes some tweets from the query with the keyword \"Israel.\" The tweets were divided into six samples of 100 tweets, which were then annotated by three to seven students in the class \"Researching White Supremacism and Antisemitism on Social Media\" taught by Gunther Jikeli, Elisha S. Breton, and Seth Moller at Indiana University in the fall of 2022, see this report. Annotators used a scale from 1 to 5 (confident not biased, probably not biased, don't know, probably biased, confident biased). The definitions of bias against each minority group used for annotation are also included in the report. \n\nIf a tweet called out or denounced bias against the minority in question, it was labeled as \"calling out bias.\" \n\nThe labels of whether a tweet is biased or calls out bias are based on a 75% majority vote. We considered \"probably biased\" and \"confident biased\" as biased and \"confident not biased,\" \"probably not biased,\" and \"don't know\" as not biased. \n\nThe types of stereotypes vary widely across the different categories of prejudice. While about a third of all biased tweets were classified as \"hate\" against the minority, the stereotypes in the tweets often matched common stereotypes about the minority. Asians were blamed for the Covid pandemic. Blacks were seen as inferior and associated with crime. Jews were seen as powerful and held collectively responsible for the actions of the State of Israel. Some tweets denied the Holocaust. Hispanics/Latines were portrayed as being in the country illegally and as \"invaders,\" in addition to stereotypical accusations of being lazy, stupid, or having too many children. Muslims, on the other hand, were often collectively blamed for terrorism and violence, though often in conversations about Muslims in India.# Content: \n \n\nThis dataset contains 5880 tweets that cover a wide range of topics common in conversations about Asians, Blacks, Jews, Latines, and Muslims. 357 tweets (6.1 %) are labeled as biased and 5523 (93.9 %) are labeled as not biased. 1365 tweets (23.2 %) are labeled as calling out or denouncing bias. \n\n1180 out of 5880 tweets (20.1 %) contain the keyword \"Asians,\" 590 were posted in 2020 and 590 in 2021. 39 tweets (3.3 %) are biased against Asian people. 370 tweets (31,4 %) call out bias against Asians. \n\n1160 out of 5880 tweets (19.7%) contain the keyword \"Blacks,\" 578 were posted in 2020 and 582 in 2021. 101 tweets (8.7 %) are biased against Black people. 334 tweets (28.8 %) call out bias against Blacks. \n\n1189 out of 5880 tweets (20.2 %) contain the keyword \"Jews,\" 592 were posted in 2020, 451 in 2021, and ––as mentioned above––146 tweets from 2022. 83 tweets (7 %) are biased against Jewish people. 220 tweets (18.5 %) call out bias against Jews.\n\n1169 out of 5880 tweets (19.9 %) contain the keyword \"Latinos,\" 584 were posted in 2020 and 585 in 2021. 29 tweets (2.5 %) are biased against Latines. 181 tweets (15.5 %) call out bias against Latines. \n\n1182 out of 5880 tweets (20.1 %) contain the keyword \"Muslims,\" 593 were posted in 2020 and 589 in 2021. 105 tweets (8.9 %) are biased against Muslims. 260 tweets (22 %) call out bias against Muslims."
] |
b22795b9ac826dcfc8bd86739013b6cfc2d52b41
|
---
task_categories:
- translation
## Dataset Description
This dataset for project snd_to_eng.
### Languages
The BCP-47 code for the dataset's language is unk.
### Dataset Fields
The dataset has the following fields (also called "features"):
```json
{
"feat_id": "Value(dtype='int64', id=None)",
"feat_source_lang": "Value(dtype='string', id=None)",
"feat_target_lang": "Value(dtype='string', id=None)",
"source": "Value(dtype='string', id=None)",
"target": "Value(dtype='string', id=None)"
}
``
|
samiesam/snd_eng
|
[
"task_categories:translation",
"size_categories:n<1K",
"language:en",
"language:sd",
"license:apache-2.0",
"code",
"region:us"
] |
2023-10-26T08:08:22+00:00
|
{"language": ["en", "sd"], "license": "apache-2.0", "size_categories": ["n<1K"], "task_categories": ["translation"], "pretty_name": "eng_snd", "tags": ["code"]}
|
2023-10-26T08:29:59+00:00
|
[] |
[
"en",
"sd"
] |
TAGS
#task_categories-translation #size_categories-n<1K #language-English #language-Sindhi #license-apache-2.0 #code #region-us
|
---
task_categories:
- translation
## Dataset Description
This dataset for project snd_to_eng.
### Languages
The BCP-47 code for the dataset's language is unk.
### Dataset Fields
The dataset has the following fields (also called "features"):
'''json
{
"feat_id": "Value(dtype='int64', id=None)",
"feat_source_lang": "Value(dtype='string', id=None)",
"feat_target_lang": "Value(dtype='string', id=None)",
"source": "Value(dtype='string', id=None)",
"target": "Value(dtype='string', id=None)"
}
''
|
[
"## Dataset Description\n\nThis dataset for project snd_to_eng.",
"### Languages\n\nThe BCP-47 code for the dataset's language is unk.",
"### Dataset Fields\n\nThe dataset has the following fields (also called \"features\"):\n\n'''json\n{\n \"feat_id\": \"Value(dtype='int64', id=None)\",\n \"feat_source_lang\": \"Value(dtype='string', id=None)\",\n \"feat_target_lang\": \"Value(dtype='string', id=None)\",\n \"source\": \"Value(dtype='string', id=None)\",\n \"target\": \"Value(dtype='string', id=None)\"\n}\n''"
] |
[
"TAGS\n#task_categories-translation #size_categories-n<1K #language-English #language-Sindhi #license-apache-2.0 #code #region-us \n",
"## Dataset Description\n\nThis dataset for project snd_to_eng.",
"### Languages\n\nThe BCP-47 code for the dataset's language is unk.",
"### Dataset Fields\n\nThe dataset has the following fields (also called \"features\"):\n\n'''json\n{\n \"feat_id\": \"Value(dtype='int64', id=None)\",\n \"feat_source_lang\": \"Value(dtype='string', id=None)\",\n \"feat_target_lang\": \"Value(dtype='string', id=None)\",\n \"source\": \"Value(dtype='string', id=None)\",\n \"target\": \"Value(dtype='string', id=None)\"\n}\n''"
] |
[
45,
16,
20,
148
] |
[
"passage: TAGS\n#task_categories-translation #size_categories-n<1K #language-English #language-Sindhi #license-apache-2.0 #code #region-us \n## Dataset Description\n\nThis dataset for project snd_to_eng.### Languages\n\nThe BCP-47 code for the dataset's language is unk.### Dataset Fields\n\nThe dataset has the following fields (also called \"features\"):\n\n'''json\n{\n \"feat_id\": \"Value(dtype='int64', id=None)\",\n \"feat_source_lang\": \"Value(dtype='string', id=None)\",\n \"feat_target_lang\": \"Value(dtype='string', id=None)\",\n \"source\": \"Value(dtype='string', id=None)\",\n \"target\": \"Value(dtype='string', id=None)\"\n}\n''"
] |
959b562f2b4dd7b76df5c32855c8777366b78d90
|
# Dataset Card for "model_v1_instruction_finetuning_dataset_updated"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
sayan1101/model_v1_instruction_finetuning_dataset_updated
|
[
"region:us"
] |
2023-10-26T08:14:35+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "text", "path": "data/text-*"}]}], "dataset_info": {"features": [{"name": "text", "sequence": "int64"}], "splits": [{"name": "text", "num_bytes": 53095000, "num_examples": 52000}], "download_size": 9929145, "dataset_size": 53095000}}
|
2023-10-26T08:17:57+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "model_v1_instruction_finetuning_dataset_updated"
More Information needed
|
[
"# Dataset Card for \"model_v1_instruction_finetuning_dataset_updated\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"model_v1_instruction_finetuning_dataset_updated\"\n\nMore Information needed"
] |
[
6,
26
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"model_v1_instruction_finetuning_dataset_updated\"\n\nMore Information needed"
] |
4ae79c01396411e3ad623b035fcfc8243b62f433
|
# Dataset Card for "instr_finetune_modelv1"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
sayan1101/instr_finetune_modelv1
|
[
"region:us"
] |
2023-10-26T08:21:29+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 27407564, "num_examples": 52000}], "download_size": 0, "dataset_size": 27407564}}
|
2023-10-26T08:50:01+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "instr_finetune_modelv1"
More Information needed
|
[
"# Dataset Card for \"instr_finetune_modelv1\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"instr_finetune_modelv1\"\n\nMore Information needed"
] |
[
6,
19
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"instr_finetune_modelv1\"\n\nMore Information needed"
] |
91a5c9a97fcfe06439e534cca5549ba73795c722
|
# Dataset Card for "guanaco-llama2-1k"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
am96149/guanaco-llama2-1k
|
[
"region:us"
] |
2023-10-26T08:39:00+00:00
|
{"dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 195589, "num_examples": 2000}, {"name": "test", "num_bytes": 87745, "num_examples": 900}], "download_size": 175131, "dataset_size": 283334}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}]}]}
|
2023-11-01T10:37:23+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "guanaco-llama2-1k"
More Information needed
|
[
"# Dataset Card for \"guanaco-llama2-1k\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"guanaco-llama2-1k\"\n\nMore Information needed"
] |
[
6,
18
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"guanaco-llama2-1k\"\n\nMore Information needed"
] |
24bf130e1e1347a10ae0ec5ed5c6facdb50b714a
|
# Dataset Card for "hotel_data1_pushed"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Ryan20/hotel_data1_pushed
|
[
"region:us"
] |
2023-10-26T08:40:10+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "inputs", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 10324, "num_examples": 16}], "download_size": 10259, "dataset_size": 10324}}
|
2023-10-26T08:40:12+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "hotel_data1_pushed"
More Information needed
|
[
"# Dataset Card for \"hotel_data1_pushed\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"hotel_data1_pushed\"\n\nMore Information needed"
] |
[
6,
17
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"hotel_data1_pushed\"\n\nMore Information needed"
] |
a160eed53a8c27e95ec1a3ff346e25de98d309a6
|
# Dataset Card for "Vie_alpaca"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
HoangHa/Vie_alpaca
|
[
"region:us"
] |
2023-10-26T08:44:22+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 51907952, "num_examples": 49999}], "download_size": 24606528, "dataset_size": 51907952}}
|
2023-10-26T08:44:26+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "Vie_alpaca"
More Information needed
|
[
"# Dataset Card for \"Vie_alpaca\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"Vie_alpaca\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"Vie_alpaca\"\n\nMore Information needed"
] |
a453fea345b91e82329c4dac7d2c8b8eaf96f57d
|
# Dataset Card for "finetune_dataset"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
sayan1101/finetune_dataset
|
[
"region:us"
] |
2023-10-26T08:49:44+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 27407564, "num_examples": 52000}], "download_size": 12306324, "dataset_size": 27407564}}
|
2023-10-26T08:51:04+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "finetune_dataset"
More Information needed
|
[
"# Dataset Card for \"finetune_dataset\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"finetune_dataset\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"finetune_dataset\"\n\nMore Information needed"
] |
289c6d3261031147eaefcdbe69fd7668841b3594
|
# Dataset Card for "bbq_cleaned"
Get the source data from here: https://huggingface.co/datasets/lighteval/bbq_helm/
And then manually selected.
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Rocinante/bbq_cleaned
|
[
"region:us"
] |
2023-10-26T09:06:08+00:00
|
{"dataset_info": {"features": [{"name": "output", "dtype": "string"}, {"name": "instruction", "dtype": "string"}, {"name": "data_source", "dtype": "string"}, {"name": "input", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 50292, "num_examples": 150}], "download_size": 28559, "dataset_size": 50292}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T09:10:33+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "bbq_cleaned"
Get the source data from here: URL
And then manually selected.
More Information needed
|
[
"# Dataset Card for \"bbq_cleaned\"\n\nGet the source data from here: URL\nAnd then manually selected.\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"bbq_cleaned\"\n\nGet the source data from here: URL\nAnd then manually selected.\n\nMore Information needed"
] |
[
6,
30
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"bbq_cleaned\"\n\nGet the source data from here: URL\nAnd then manually selected.\n\nMore Information needed"
] |
00dc96a69d2bd58dbcb57f39cf99f415b30d0593
|
# Dataset Card for Evaluation run of HWERI/Llama2-7b-sharegpt4
## Dataset Description
- **Homepage:**
- **Repository:** https://huggingface.co/HWERI/Llama2-7b-sharegpt4
- **Paper:**
- **Leaderboard:** https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- **Point of Contact:** [email protected]
### Dataset Summary
Dataset automatically created during the evaluation run of model [HWERI/Llama2-7b-sharegpt4](https://huggingface.co/HWERI/Llama2-7b-sharegpt4) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).
To load the details from a run, you can for instance do the following:
```python
from datasets import load_dataset
data = load_dataset("open-llm-leaderboard/details_HWERI__Llama2-7b-sharegpt4",
"harness_winogrande_5",
split="train")
```
## Latest results
These are the [latest results from run 2023-10-26T10:08:23.331981](https://huggingface.co/datasets/open-llm-leaderboard/details_HWERI__Llama2-7b-sharegpt4/blob/main/results_2023-10-26T10-08-23.331981.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
```python
{
"all": {
"em": 0.001572986577181208,
"em_stderr": 0.0004058451132417743,
"f1": 0.06141988255033573,
"f1_stderr": 0.0014263478827371335,
"acc": 0.369226585159047,
"acc_stderr": 0.008577465355756637
},
"harness|drop|3": {
"em": 0.001572986577181208,
"em_stderr": 0.0004058451132417743,
"f1": 0.06141988255033573,
"f1_stderr": 0.0014263478827371335
},
"harness|gsm8k|5": {
"acc": 0.026535253980288095,
"acc_stderr": 0.00442704598726516
},
"harness|winogrande|5": {
"acc": 0.7119179163378059,
"acc_stderr": 0.012727884724248115
}
}
```
### Supported Tasks and Leaderboards
[More Information Needed]
### Languages
[More Information Needed]
## Dataset Structure
### Data Instances
[More Information Needed]
### Data Fields
[More Information Needed]
### Data Splits
[More Information Needed]
## Dataset Creation
### Curation Rationale
[More Information Needed]
### Source Data
#### Initial Data Collection and Normalization
[More Information Needed]
#### Who are the source language producers?
[More Information Needed]
### Annotations
#### Annotation process
[More Information Needed]
#### Who are the annotators?
[More Information Needed]
### Personal and Sensitive Information
[More Information Needed]
## Considerations for Using the Data
### Social Impact of Dataset
[More Information Needed]
### Discussion of Biases
[More Information Needed]
### Other Known Limitations
[More Information Needed]
## Additional Information
### Dataset Curators
[More Information Needed]
### Licensing Information
[More Information Needed]
### Citation Information
[More Information Needed]
### Contributions
[More Information Needed]
|
open-llm-leaderboard/details_HWERI__Llama2-7b-sharegpt4
|
[
"region:us"
] |
2023-10-26T09:08:27+00:00
|
{"pretty_name": "Evaluation run of HWERI/Llama2-7b-sharegpt4", "dataset_summary": "Dataset automatically created during the evaluation run of model [HWERI/Llama2-7b-sharegpt4](https://huggingface.co/HWERI/Llama2-7b-sharegpt4) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).\n\nTo load the details from a run, you can for instance do the following:\n```python\nfrom datasets import load_dataset\ndata = load_dataset(\"open-llm-leaderboard/details_HWERI__Llama2-7b-sharegpt4\",\n\t\"harness_winogrande_5\",\n\tsplit=\"train\")\n```\n\n## Latest results\n\nThese are the [latest results from run 2023-10-26T10:08:23.331981](https://huggingface.co/datasets/open-llm-leaderboard/details_HWERI__Llama2-7b-sharegpt4/blob/main/results_2023-10-26T10-08-23.331981.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):\n\n```python\n{\n \"all\": {\n \"em\": 0.001572986577181208,\n \"em_stderr\": 0.0004058451132417743,\n \"f1\": 0.06141988255033573,\n \"f1_stderr\": 0.0014263478827371335,\n \"acc\": 0.369226585159047,\n \"acc_stderr\": 0.008577465355756637\n },\n \"harness|drop|3\": {\n \"em\": 0.001572986577181208,\n \"em_stderr\": 0.0004058451132417743,\n \"f1\": 0.06141988255033573,\n \"f1_stderr\": 0.0014263478827371335\n },\n \"harness|gsm8k|5\": {\n \"acc\": 0.026535253980288095,\n \"acc_stderr\": 0.00442704598726516\n },\n \"harness|winogrande|5\": {\n \"acc\": 0.7119179163378059,\n \"acc_stderr\": 0.012727884724248115\n }\n}\n```", "repo_url": "https://huggingface.co/HWERI/Llama2-7b-sharegpt4", "leaderboard_url": "https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard", "point_of_contact": "[email protected]", "configs": [{"config_name": "harness_drop_3", "data_files": [{"split": "2023_10_26T10_08_23.331981", "path": ["**/details_harness|drop|3_2023-10-26T10-08-23.331981.parquet"]}, {"split": "latest", "path": ["**/details_harness|drop|3_2023-10-26T10-08-23.331981.parquet"]}]}, {"config_name": "harness_gsm8k_5", "data_files": [{"split": "2023_10_26T10_08_23.331981", "path": ["**/details_harness|gsm8k|5_2023-10-26T10-08-23.331981.parquet"]}, {"split": "latest", "path": ["**/details_harness|gsm8k|5_2023-10-26T10-08-23.331981.parquet"]}]}, {"config_name": "harness_winogrande_5", "data_files": [{"split": "2023_10_26T10_08_23.331981", "path": ["**/details_harness|winogrande|5_2023-10-26T10-08-23.331981.parquet"]}, {"split": "latest", "path": ["**/details_harness|winogrande|5_2023-10-26T10-08-23.331981.parquet"]}]}, {"config_name": "results", "data_files": [{"split": "2023_10_26T10_08_23.331981", "path": ["results_2023-10-26T10-08-23.331981.parquet"]}, {"split": "latest", "path": ["results_2023-10-26T10-08-23.331981.parquet"]}]}]}
|
2023-10-26T09:08:34+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for Evaluation run of HWERI/Llama2-7b-sharegpt4
## Dataset Description
- Homepage:
- Repository: URL
- Paper:
- Leaderboard: URL
- Point of Contact: clementine@URL
### Dataset Summary
Dataset automatically created during the evaluation run of model HWERI/Llama2-7b-sharegpt4 on the Open LLM Leaderboard.
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).
To load the details from a run, you can for instance do the following:
## Latest results
These are the latest results from run 2023-10-26T10:08:23.331981(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
### Supported Tasks and Leaderboards
### Languages
## Dataset Structure
### Data Instances
### Data Fields
### Data Splits
## Dataset Creation
### Curation Rationale
### Source Data
#### Initial Data Collection and Normalization
#### Who are the source language producers?
### Annotations
#### Annotation process
#### Who are the annotators?
### Personal and Sensitive Information
## Considerations for Using the Data
### Social Impact of Dataset
### Discussion of Biases
### Other Known Limitations
## Additional Information
### Dataset Curators
### Licensing Information
### Contributions
|
[
"# Dataset Card for Evaluation run of HWERI/Llama2-7b-sharegpt4",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model HWERI/Llama2-7b-sharegpt4 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T10:08:23.331981(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for Evaluation run of HWERI/Llama2-7b-sharegpt4",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model HWERI/Llama2-7b-sharegpt4 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T10:08:23.331981(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
6,
23,
31,
171,
66,
10,
4,
6,
6,
5,
5,
5,
7,
4,
10,
10,
5,
5,
9,
8,
8,
7,
8,
7,
5,
6,
6,
5
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for Evaluation run of HWERI/Llama2-7b-sharegpt4## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL### Dataset Summary\n\nDataset automatically created during the evaluation run of model HWERI/Llama2-7b-sharegpt4 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:## Latest results\n\nThese are the latest results from run 2023-10-26T10:08:23.331981(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):### Supported Tasks and Leaderboards### Languages## Dataset Structure### Data Instances### Data Fields### Data Splits## Dataset Creation### Curation Rationale### Source Data#### Initial Data Collection and Normalization#### Who are the source language producers?### Annotations#### Annotation process#### Who are the annotators?### Personal and Sensitive Information## Considerations for Using the Data### Social Impact of Dataset### Discussion of Biases### Other Known Limitations## Additional Information### Dataset Curators### Licensing Information### Contributions"
] |
888bb0fe0629b0a514d4e8c3bcfdd46f5e7c93ed
|
# Dataset Card for "lj_speech_DifferentStructure_removedVocabs_prepared"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
HamdanXI/lj_speech_DifferentStructure_removedVocabs_prepared
|
[
"region:us"
] |
2023-10-26T09:18:09+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "input_values", "sequence": "float32"}, {"name": "labels", "sequence": "float64"}], "splits": [{"name": "train", "num_bytes": 2696420240, "num_examples": 4620}, {"name": "test", "num_bytes": 975708736, "num_examples": 1680}], "download_size": 3391043801, "dataset_size": 3672128976}}
|
2023-10-26T09:20:22+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "lj_speech_DifferentStructure_removedVocabs_prepared"
More Information needed
|
[
"# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs_prepared\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs_prepared\"\n\nMore Information needed"
] |
[
6,
31
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"lj_speech_DifferentStructure_removedVocabs_prepared\"\n\nMore Information needed"
] |
7e4d8960b5c76976f872d64a73436ae1aa036f44
|
# Defect Spectrum Dataset
Welcome to the Defect Spectrum dataset repository. This comprehensive benchmark is a granular collection of large-scale defect datasets with rich semantics, designed to push the frontier of industrial defect inspection research and applications.
## Overview
Defect inspection is a critical component within the closed-loop manufacturing system. To facilitate advanced research and development in this domain, we introduce the Defect Spectrum dataset. It offers precise, semantics-abundant, and large-scale annotations for a wide range of industrial defects. This dataset is an enhancement over existing benchmarks, providing refined annotations and introducing detailed semantic layers, allowing for the distinction between multiple defect types within a single image.
### Features
- **Semantics-Abundant Annotations**: Each defect is meticulously labeled, not just at the pixel level but with rich contextual information, providing insights into the defect type and implications.
- **High Precision**: Annotations are refined by experts to capture even the subtlest of defects, ensuring high precision.
- **Large-Scale Data**: Building on four key industrial benchmarks, Defect Spectrum stands out with its extensive coverage and depth.
- **Incorporates Descriptive Captions**: To bridge the gap towards Vision Language Models (VLMs), each sample is accompanied by a descriptive caption.
### Directory Structure
```plaintext
DefectSpectrum/
├── DS-MVTec/
│ ├── bottle/
│ │ ├── image/ # Original images of the bottle category
│ │ ├── caption/ # Descriptive captions of the bottle category
│ │ ├── mask/ # Single channel defect masks for the bottle category
│ │ └── rgb_mask/ # Colored defect masks for better visualization
│ ├── cable/
│ │ ├── image/ # Original images of the cable category
│ │ ├── caption/ # Descriptive captions of the cable category
│ │ ├── mask/ # Single channel defect masks for the cable category
│ │ └── rgb_mask/ # Colored defect masks for better visualization
│ └── ...
├── DS-VISION/
│ └── ...
├── DS-DAGM/
│ └── ...
├── DS-Cotton-Fabric/
│ └── ...
```
## To-Do List
- [x] Task 1: Release DS-MVTec image-mask pairs.
- [x] Task 2: Release DS-VISION, DS-DAGM, and DS-Cotton-Fabric image-mask pairs.
- [ ] Task 3: Release captions.
- [ ] Task 4: Release selected synthetic data.
---
license: mit
---
|
Andyson/DefectSpectrum
|
[
"task_categories:image-segmentation",
"task_categories:image-to-text",
"size_categories:1K<n<10K",
"language:en",
"license:mit",
"industry",
"region:us"
] |
2023-10-26T09:20:37+00:00
|
{"language": ["en"], "license": "mit", "size_categories": ["1K<n<10K"], "task_categories": ["image-segmentation", "image-to-text"], "pretty_name": "DefectSpectrum", "tags": ["industry"]}
|
2023-11-06T08:27:11+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-image-segmentation #task_categories-image-to-text #size_categories-1K<n<10K #language-English #license-mit #industry #region-us
|
# Defect Spectrum Dataset
Welcome to the Defect Spectrum dataset repository. This comprehensive benchmark is a granular collection of large-scale defect datasets with rich semantics, designed to push the frontier of industrial defect inspection research and applications.
## Overview
Defect inspection is a critical component within the closed-loop manufacturing system. To facilitate advanced research and development in this domain, we introduce the Defect Spectrum dataset. It offers precise, semantics-abundant, and large-scale annotations for a wide range of industrial defects. This dataset is an enhancement over existing benchmarks, providing refined annotations and introducing detailed semantic layers, allowing for the distinction between multiple defect types within a single image.
### Features
- Semantics-Abundant Annotations: Each defect is meticulously labeled, not just at the pixel level but with rich contextual information, providing insights into the defect type and implications.
- High Precision: Annotations are refined by experts to capture even the subtlest of defects, ensuring high precision.
- Large-Scale Data: Building on four key industrial benchmarks, Defect Spectrum stands out with its extensive coverage and depth.
- Incorporates Descriptive Captions: To bridge the gap towards Vision Language Models (VLMs), each sample is accompanied by a descriptive caption.
### Directory Structure
## To-Do List
- [x] Task 1: Release DS-MVTec image-mask pairs.
- [x] Task 2: Release DS-VISION, DS-DAGM, and DS-Cotton-Fabric image-mask pairs.
- [ ] Task 3: Release captions.
- [ ] Task 4: Release selected synthetic data.
---
license: mit
---
|
[
"# Defect Spectrum Dataset\n\nWelcome to the Defect Spectrum dataset repository. This comprehensive benchmark is a granular collection of large-scale defect datasets with rich semantics, designed to push the frontier of industrial defect inspection research and applications.",
"## Overview\n\nDefect inspection is a critical component within the closed-loop manufacturing system. To facilitate advanced research and development in this domain, we introduce the Defect Spectrum dataset. It offers precise, semantics-abundant, and large-scale annotations for a wide range of industrial defects. This dataset is an enhancement over existing benchmarks, providing refined annotations and introducing detailed semantic layers, allowing for the distinction between multiple defect types within a single image.",
"### Features\n\n- Semantics-Abundant Annotations: Each defect is meticulously labeled, not just at the pixel level but with rich contextual information, providing insights into the defect type and implications.\n- High Precision: Annotations are refined by experts to capture even the subtlest of defects, ensuring high precision.\n- Large-Scale Data: Building on four key industrial benchmarks, Defect Spectrum stands out with its extensive coverage and depth.\n- Incorporates Descriptive Captions: To bridge the gap towards Vision Language Models (VLMs), each sample is accompanied by a descriptive caption.",
"### Directory Structure",
"## To-Do List\n- [x] Task 1: Release DS-MVTec image-mask pairs.\n- [x] Task 2: Release DS-VISION, DS-DAGM, and DS-Cotton-Fabric image-mask pairs.\n- [ ] Task 3: Release captions.\n- [ ] Task 4: Release selected synthetic data.\n\n---\nlicense: mit\n---"
] |
[
"TAGS\n#task_categories-image-segmentation #task_categories-image-to-text #size_categories-1K<n<10K #language-English #license-mit #industry #region-us \n",
"# Defect Spectrum Dataset\n\nWelcome to the Defect Spectrum dataset repository. This comprehensive benchmark is a granular collection of large-scale defect datasets with rich semantics, designed to push the frontier of industrial defect inspection research and applications.",
"## Overview\n\nDefect inspection is a critical component within the closed-loop manufacturing system. To facilitate advanced research and development in this domain, we introduce the Defect Spectrum dataset. It offers precise, semantics-abundant, and large-scale annotations for a wide range of industrial defects. This dataset is an enhancement over existing benchmarks, providing refined annotations and introducing detailed semantic layers, allowing for the distinction between multiple defect types within a single image.",
"### Features\n\n- Semantics-Abundant Annotations: Each defect is meticulously labeled, not just at the pixel level but with rich contextual information, providing insights into the defect type and implications.\n- High Precision: Annotations are refined by experts to capture even the subtlest of defects, ensuring high precision.\n- Large-Scale Data: Building on four key industrial benchmarks, Defect Spectrum stands out with its extensive coverage and depth.\n- Incorporates Descriptive Captions: To bridge the gap towards Vision Language Models (VLMs), each sample is accompanied by a descriptive caption.",
"### Directory Structure",
"## To-Do List\n- [x] Task 1: Release DS-MVTec image-mask pairs.\n- [x] Task 2: Release DS-VISION, DS-DAGM, and DS-Cotton-Fabric image-mask pairs.\n- [ ] Task 3: Release captions.\n- [ ] Task 4: Release selected synthetic data.\n\n---\nlicense: mit\n---"
] |
[
54,
58,
112,
152,
7,
86
] |
[
"passage: TAGS\n#task_categories-image-segmentation #task_categories-image-to-text #size_categories-1K<n<10K #language-English #license-mit #industry #region-us \n# Defect Spectrum Dataset\n\nWelcome to the Defect Spectrum dataset repository. This comprehensive benchmark is a granular collection of large-scale defect datasets with rich semantics, designed to push the frontier of industrial defect inspection research and applications.## Overview\n\nDefect inspection is a critical component within the closed-loop manufacturing system. To facilitate advanced research and development in this domain, we introduce the Defect Spectrum dataset. It offers precise, semantics-abundant, and large-scale annotations for a wide range of industrial defects. This dataset is an enhancement over existing benchmarks, providing refined annotations and introducing detailed semantic layers, allowing for the distinction between multiple defect types within a single image.### Features\n\n- Semantics-Abundant Annotations: Each defect is meticulously labeled, not just at the pixel level but with rich contextual information, providing insights into the defect type and implications.\n- High Precision: Annotations are refined by experts to capture even the subtlest of defects, ensuring high precision.\n- Large-Scale Data: Building on four key industrial benchmarks, Defect Spectrum stands out with its extensive coverage and depth.\n- Incorporates Descriptive Captions: To bridge the gap towards Vision Language Models (VLMs), each sample is accompanied by a descriptive caption.### Directory Structure## To-Do List\n- [x] Task 1: Release DS-MVTec image-mask pairs.\n- [x] Task 2: Release DS-VISION, DS-DAGM, and DS-Cotton-Fabric image-mask pairs.\n- [ ] Task 3: Release captions.\n- [ ] Task 4: Release selected synthetic data.\n\n---\nlicense: mit\n---"
] |
05bfd7026d10604cb255e10bd304d427e7005ad9
|
# Dataset Card for "color_painting_dataset_1024"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
KoreadeepKai/color_painting_dataset_1024
|
[
"region:us"
] |
2023-10-26T09:28:42+00:00
|
{"dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 346962459.566, "num_examples": 3551}], "download_size": 322932227, "dataset_size": 346962459.566}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T09:29:06+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "color_painting_dataset_1024"
More Information needed
|
[
"# Dataset Card for \"color_painting_dataset_1024\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"color_painting_dataset_1024\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"color_painting_dataset_1024\"\n\nMore Information needed"
] |
db67b1f40a1e3dd5d442047c573c49ab20816389
|
**[Amyl Guard Supplement](https://www.glitco.com/get-amyl-guard) is one of the best solutions to help you lose weight fast, and easily, and helps to support a healthy digestive system without side effects. Read this review all about ingredients, side effects, complaints, price, pros, cons, and more.**
**[](https://www.glitco.com/get-amyl-guard)**
[**Amyl Guard**](https://www.facebook.com/groups/amylguardweightloss) is a 5-second appetizer help flushes out a pound of stubborn fat from your problematic area.
It's the first-ever amylase inhibitor limiting the carbohydrates, indigestion, and absorption of sugar and fats in the gut.
Amyl Guard formula targets the root source of ugly and unhealthy nagging belly fat and your high amylase level, bringing back the slim body appearance.
**[Amyl Guard](https://www.glitco.com/get-amyl-guard)** supplement shrinks hip, arm, and thigh fat, revealing a toned figure you haven't seen in years.
Amyl Guard weight loss makes you feel more confident and happy as you develop less interest in crab food, eating only the nutrient-filled delicious meal.
Amyl Guard is the only solution supporting your weight management no matter your age, lbs, or how much time use lose in the past.
**[Click to See the Amyl Guard Ingredients Details..](https://www.glitco.com/get-amyl-guard)**
### Benefits of Amyl Guard:
* Amyl Guard helps reduce the pound of unhealthy and nagging fat from your body.
* It causes you to feel better than you have in a long time with a spectacular hip, belly, and slim arm.
* You can stop pursuing a diet or exercise and eat your favorite food without restrictions.
* Your stomach will tighten and firm up as you use the [**Amyl Guard**](https://www.facebook.com/groups/amylguardweightloss)
* Amyl Guard reduces the binge eating habit and suppresses your craving.
* A radiance, firmness, and elasticity of youth return to your skin and physique, letting you witness the real beauty that makes you look younger.
* It leaves you feeling more happy, and healthy than you have in years or even decades.
* **[Amyl Guard](https://www.glitco.com/get-amyl-guard)** formula helps to convert your body into a fine slim, and healthy one.
* Restoring your amylase to normal levels for a long-lasting weight loss
[](https://www.glitco.com/get-amyl-guard)
### Drawbacks of Amyl Guard:
* You can't purchase the Amyl Guard from a third-party store or [online webpage](https://www.glitco.com/get-amyl-guard).
* You are not eligible to receive the offers and guarantees of Amyl Guard if you purchased the goods from another unregistered website.
### How Much Will It Cost to Purchase [Amyl Guard](https://www.glitco.com/get-amyl-guard)?
Amyl Guard is sold only on the official page containing a unique combination of the five most effective super nutrients for fat reduction and rejuvenating the enzyme of amylase enzymes.
Due to its effectiveness and balance of metabolism, the Amyl Guard supplement has reached its limit, so I chose to pick 6 bottles at once and have a long time to experience the result.
Amyl Guard weight loss formula ensures the creator that you have an adequate supply to accomplish the fat-melting effectiveness.
**[Click to See the Amyl Guard Ingredients Details..](https://www.glitco.com/get-amyl-guard)**
[](https://www.glitco.com/get-amyl-guard)
Try the **[Amyl Guard](https://www.glitco.com/get-amyl-guard)** to make your body more challenging in storing more nutrients and less fat.
* Trail Package: Buy one bottle of Amyl Guard for $59 each and save $8.
* Most Popular: Buy three bottles of Amyl Guard for $49 each and Save $54
* Best Value: Buy six bottles of **[Amyl Guard](https://www.glitco.com/get-amyl-guard)** for $29 each and Save $228 + Free Bonus and FREE SHIPPING.
|
amylguardofficial/amylguardofficial
|
[
"region:us"
] |
2023-10-26T09:39:36+00:00
|
{}
|
2023-10-26T09:40:19+00:00
|
[] |
[] |
TAGS
#region-us
|
Amyl Guard Supplement is one of the best solutions to help you lose weight fast, and easily, and helps to support a healthy digestive system without side effects. Read this review all about ingredients, side effects, complaints, price, pros, cons, and more.
![](URL
Amyl Guard is a 5-second appetizer help flushes out a pound of stubborn fat from your problematic area.
It's the first-ever amylase inhibitor limiting the carbohydrates, indigestion, and absorption of sugar and fats in the gut.
Amyl Guard formula targets the root source of ugly and unhealthy nagging belly fat and your high amylase level, bringing back the slim body appearance.
Amyl Guard supplement shrinks hip, arm, and thigh fat, revealing a toned figure you haven't seen in years.
Amyl Guard weight loss makes you feel more confident and happy as you develop less interest in crab food, eating only the nutrient-filled delicious meal.
Amyl Guard is the only solution supporting your weight management no matter your age, lbs, or how much time use lose in the past.
Click to See the Amyl Guard Ingredients Details..
### Benefits of Amyl Guard:
* Amyl Guard helps reduce the pound of unhealthy and nagging fat from your body.
* It causes you to feel better than you have in a long time with a spectacular hip, belly, and slim arm.
* You can stop pursuing a diet or exercise and eat your favorite food without restrictions.
* Your stomach will tighten and firm up as you use the Amyl Guard
* Amyl Guard reduces the binge eating habit and suppresses your craving.
* A radiance, firmness, and elasticity of youth return to your skin and physique, letting you witness the real beauty that makes you look younger.
* It leaves you feeling more happy, and healthy than you have in years or even decades.
* Amyl Guard formula helps to convert your body into a fine slim, and healthy one.
* Restoring your amylase to normal levels for a long-lasting weight loss
![](URL
### Drawbacks of Amyl Guard:
* You can't purchase the Amyl Guard from a third-party store or online webpage.
* You are not eligible to receive the offers and guarantees of Amyl Guard if you purchased the goods from another unregistered website.
### How Much Will It Cost to Purchase Amyl Guard?
Amyl Guard is sold only on the official page containing a unique combination of the five most effective super nutrients for fat reduction and rejuvenating the enzyme of amylase enzymes.
Due to its effectiveness and balance of metabolism, the Amyl Guard supplement has reached its limit, so I chose to pick 6 bottles at once and have a long time to experience the result.
Amyl Guard weight loss formula ensures the creator that you have an adequate supply to accomplish the fat-melting effectiveness.
Click to See the Amyl Guard Ingredients Details..
![](URL
Try the Amyl Guard to make your body more challenging in storing more nutrients and less fat.
* Trail Package: Buy one bottle of Amyl Guard for $59 each and save $8.
* Most Popular: Buy three bottles of Amyl Guard for $49 each and Save $54
* Best Value: Buy six bottles of Amyl Guard for $29 each and Save $228 + Free Bonus and FREE SHIPPING.
|
[
"### Benefits of Amyl Guard:\n\n* Amyl Guard helps reduce the pound of unhealthy and nagging fat from your body. \n \n \n* It causes you to feel better than you have in a long time with a spectacular hip, belly, and slim arm. \n \n \n* You can stop pursuing a diet or exercise and eat your favorite food without restrictions. \n \n \n* Your stomach will tighten and firm up as you use the Amyl Guard \n \n \n* Amyl Guard reduces the binge eating habit and suppresses your craving. \n \n \n* A radiance, firmness, and elasticity of youth return to your skin and physique, letting you witness the real beauty that makes you look younger. \n \n \n* It leaves you feeling more happy, and healthy than you have in years or even decades. \n \n \n* Amyl Guard formula helps to convert your body into a fine slim, and healthy one. \n \n \n* Restoring your amylase to normal levels for a long-lasting weight loss\n\n![](URL",
"### Drawbacks of Amyl Guard:\n\n* You can't purchase the Amyl Guard from a third-party store or online webpage. \n \n \n* You are not eligible to receive the offers and guarantees of Amyl Guard if you purchased the goods from another unregistered website.",
"### How Much Will It Cost to Purchase Amyl Guard?\n\nAmyl Guard is sold only on the official page containing a unique combination of the five most effective super nutrients for fat reduction and rejuvenating the enzyme of amylase enzymes.\n\nDue to its effectiveness and balance of metabolism, the Amyl Guard supplement has reached its limit, so I chose to pick 6 bottles at once and have a long time to experience the result.\n\nAmyl Guard weight loss formula ensures the creator that you have an adequate supply to accomplish the fat-melting effectiveness.\n\nClick to See the Amyl Guard Ingredients Details..\n\n![](URL\n\nTry the Amyl Guard to make your body more challenging in storing more nutrients and less fat.\n\n* Trail Package: Buy one bottle of Amyl Guard for $59 each and save $8. \n \n \n* Most Popular: Buy three bottles of Amyl Guard for $49 each and Save $54 \n \n \n* Best Value: Buy six bottles of Amyl Guard for $29 each and Save $228 + Free Bonus and FREE SHIPPING."
] |
[
"TAGS\n#region-us \n",
"### Benefits of Amyl Guard:\n\n* Amyl Guard helps reduce the pound of unhealthy and nagging fat from your body. \n \n \n* It causes you to feel better than you have in a long time with a spectacular hip, belly, and slim arm. \n \n \n* You can stop pursuing a diet or exercise and eat your favorite food without restrictions. \n \n \n* Your stomach will tighten and firm up as you use the Amyl Guard \n \n \n* Amyl Guard reduces the binge eating habit and suppresses your craving. \n \n \n* A radiance, firmness, and elasticity of youth return to your skin and physique, letting you witness the real beauty that makes you look younger. \n \n \n* It leaves you feeling more happy, and healthy than you have in years or even decades. \n \n \n* Amyl Guard formula helps to convert your body into a fine slim, and healthy one. \n \n \n* Restoring your amylase to normal levels for a long-lasting weight loss\n\n![](URL",
"### Drawbacks of Amyl Guard:\n\n* You can't purchase the Amyl Guard from a third-party store or online webpage. \n \n \n* You are not eligible to receive the offers and guarantees of Amyl Guard if you purchased the goods from another unregistered website.",
"### How Much Will It Cost to Purchase Amyl Guard?\n\nAmyl Guard is sold only on the official page containing a unique combination of the five most effective super nutrients for fat reduction and rejuvenating the enzyme of amylase enzymes.\n\nDue to its effectiveness and balance of metabolism, the Amyl Guard supplement has reached its limit, so I chose to pick 6 bottles at once and have a long time to experience the result.\n\nAmyl Guard weight loss formula ensures the creator that you have an adequate supply to accomplish the fat-melting effectiveness.\n\nClick to See the Amyl Guard Ingredients Details..\n\n![](URL\n\nTry the Amyl Guard to make your body more challenging in storing more nutrients and less fat.\n\n* Trail Package: Buy one bottle of Amyl Guard for $59 each and save $8. \n \n \n* Most Popular: Buy three bottles of Amyl Guard for $49 each and Save $54 \n \n \n* Best Value: Buy six bottles of Amyl Guard for $29 each and Save $228 + Free Bonus and FREE SHIPPING."
] |
[
6,
214,
62,
234
] |
[
"passage: TAGS\n#region-us \n### Benefits of Amyl Guard:\n\n* Amyl Guard helps reduce the pound of unhealthy and nagging fat from your body. \n \n \n* It causes you to feel better than you have in a long time with a spectacular hip, belly, and slim arm. \n \n \n* You can stop pursuing a diet or exercise and eat your favorite food without restrictions. \n \n \n* Your stomach will tighten and firm up as you use the Amyl Guard \n \n \n* Amyl Guard reduces the binge eating habit and suppresses your craving. \n \n \n* A radiance, firmness, and elasticity of youth return to your skin and physique, letting you witness the real beauty that makes you look younger. \n \n \n* It leaves you feeling more happy, and healthy than you have in years or even decades. \n \n \n* Amyl Guard formula helps to convert your body into a fine slim, and healthy one. \n \n \n* Restoring your amylase to normal levels for a long-lasting weight loss\n\n![](URL### Drawbacks of Amyl Guard:\n\n* You can't purchase the Amyl Guard from a third-party store or online webpage. \n \n \n* You are not eligible to receive the offers and guarantees of Amyl Guard if you purchased the goods from another unregistered website."
] |
0a098937e3e1a49f0f809c2f6cb443f24ab279c3
|
# Dataset Card for UzCrawl
## Table of Contents
- [Dataset Description](#dataset-description)
- [Dataset Summary](#dataset-summary)
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
- [Languages](#languages)
- [Dataset Structure](#dataset-structure)
- [Data Instances](#data-instances)
- [Data Fields](#data-fields)
- [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
- [Curation Rationale](#curation-rationale)
- [Source Data](#source-data)
- [Annotations](#annotations)
- [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
- [Social Impact of Dataset](#social-impact-of-dataset)
- [Discussion of Biases](#discussion-of-biases)
- [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
- [Dataset Curators](#dataset-curators)
- [Licensing Information](#licensing-information)
- [Citation Information](#citation-information)
- [Contributions](#contributions)
## Dataset Description
- **Homepage:** [https://tahrirchi.uz/grammatika-tekshiruvi](https://tahrirchi.uz/grammatika-tekshiruvi)
- **Repository:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
- **Paper:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
- **Point of Contact:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
- **Size of downloaded dataset files:** 3.52 GB
- **Size of the generated dataset:** 1.58 GB
- **Total amount of disk used:** 5.1 GB
### Dataset Summary
In an effort to democratize research on low-resource languages, we release UzCrawl dataset, a web and telegram crawl corpus consisting of materials from nearly 1.2 million unique sources in the Uzbek Language.
Please refer to our [blogpost](https://tahrirchi.uz/grammatika-tekshiruvi) and paper (Coming soon!) for further details.
To load and use dataset, run this script:
```python
from datasets import load_dataset
uz_crawl=load_dataset("tahrirchi/uz-crawl")
```
## Dataset Structure
### Data Instances
#### plain_text
- **Size of downloaded dataset files:** 3.52 GB
- **Size of the generated dataset:** 1.58 GB
- **Total amount of disk used:** 5.1 GB
An example of 'news' looks as follows.
```
{
'text': "O‘zbekiston Respublikasi Vazirlar Mahkamasining 2019 yil 24 iyuldagi 620-son qarori bilan tasdiqlangan «Xorijiy davlatlarda ta'lim olganlik to‘g‘risidagi hujjatlarni tan olish tartibi to‘g‘risida»gi Nizom ijrosini ta'minlash maqsadida Ta'lim sifatini nazorat qilish davlat inspeksiyasida (Toshkent shahar, Chilonzor tumani, Nurxon ko‘chasi, 21-uy) 2019 yil 9 –14 sentabr kunlari sohalar bo‘yicha sinov testlari bo‘lib o‘tishi rejalashtirilgan.\nTa'lim sifatini nazorat qilish davlat inspeksiyasi matbuot xizmati xabariga\xa0ko‘ra, «Huquqshunoslik», «Sog‘liqni saqlash va ijtimoiy ta'minot», «Iqtisodiyot», «Qishloq xo‘jaligi, muhandislik, ishlov berish va qurilish» hamda «O‘qituvchilar tayyorlash va pedagogik fanlar» sohalari bo‘yicha sinov testlari o‘tkaziladigan sanasi va sinov testida ishtirok etuvchilar ro‘yxati jadvalga muvofiq belgilanadi.\nTa'lim sifatini nazorat qilish davlat inspeksiyasi ogohlantirishicha, xorijiy davlatlarda ta'lim olganlik to‘g‘risidagi hujjatlarni tan olish uchun belgilangan sinov testlariga o‘z vaqtida kelmagan, sinov testida ishtirok etuvchilar ro‘yxatida mavjud bo‘lmagan talabgorlarga sinovlarga kirishga ruxsat etilmaydi.",
'timestamp': '2019-06-09',
'source': 'https://kun.uz/uz/news/2019/09/06/xorijda-talim-olganlik-togrisidagi-hujjatlarni-tan-olish-uchun-testlar-otkaziladigan-kunlar-malum-boldi'
}
```
### Data Fields
The data fields are the same among all splits.
- `text`: a `string` feature that contains text.
- `timestamp`: a `string` feature that contains timestamp of the material.
- `source`: a `string` feature that contains url of the material.
### Data Splits
| name | |
|-----------------|--------:|
| news | 964268 |
| telegram_blogs | 227337 |
## Dataset Creation
The news portion have been crawled from 21 different websites using [Scrapy](https://scrapy.org/) framework. And telegram_blogs portion is consisted of manually curated texts from 81 high-quality Telegram channels.
## Citation
Please cite this model using the following format:
```
@online{Mamasaidov2023UzBooks,
author = {Mukhammadsaid Mamasaidov and Abror Shopulatov},
title = {UzCrawl dataset},
year = {2023},
url = {https://huggingface.co/datasets/tahrirchi/uz-crawl},
note = {Accessed: 2023-10-28}, % change this date
urldate = {2023-10-28} % change this date
}
```
## Gratitude
We are thankful to these awesome organizations and people for helping to make it happen:
- [Asadbek Kiyomov](https://www.linkedin.com/in/asadbey): for his works on the beginning of the project.
- [Ilya Gusev](https://github.com/IlyaGusev/): for his advise throughout the process
- [David Dale](https://daviddale.ru): for his advise throughout the process
## Contacts
We believe that this work will inspire all enthusiasts around the world to open the hidden beauty of low-resource languages, in particular of Uzbek.
For further development and issues about the dataset, please use [email protected] or [email protected] to contact.
|
tahrirchi/uz-crawl
|
[
"task_categories:text-generation",
"task_categories:fill-mask",
"task_ids:language-modeling",
"task_ids:masked-language-modeling",
"annotations_creators:no-annotation",
"multilinguality:monolingual",
"size_categories:1M<n<10M",
"language:uz",
"license:apache-2.0",
"uz",
"crawl",
"telegram_blogs",
"region:us"
] |
2023-10-26T09:43:01+00:00
|
{"annotations_creators": ["no-annotation"], "language": ["uz"], "license": "apache-2.0", "multilinguality": ["monolingual"], "size_categories": ["1M<n<10M"], "task_categories": ["text-generation", "fill-mask"], "task_ids": ["language-modeling", "masked-language-modeling"], "pretty_name": "UzCrawl", "configs": [{"config_name": "default", "data_files": [{"split": "news", "path": "data/news-*"}, {"split": "telegram_blogs", "path": "data/telegram_blogs-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}, {"name": "timestamp", "dtype": "string"}, {"name": "source", "dtype": "string"}], "splits": [{"name": "news", "num_bytes": 3272404822, "num_examples": 964268}, {"name": "telegram_blogs", "num_bytes": 367462330, "num_examples": 368017}], "download_size": 1462920936, "dataset_size": 3639867152}, "tags": ["uz", "crawl", "telegram_blogs"]}
|
2023-12-19T15:26:14+00:00
|
[] |
[
"uz"
] |
TAGS
#task_categories-text-generation #task_categories-fill-mask #task_ids-language-modeling #task_ids-masked-language-modeling #annotations_creators-no-annotation #multilinguality-monolingual #size_categories-1M<n<10M #language-Uzbek #license-apache-2.0 #uz #crawl #telegram_blogs #region-us
|
Dataset Card for UzCrawl
========================
Table of Contents
-----------------
* Dataset Description
+ Dataset Summary
+ Supported Tasks and Leaderboards
+ Languages
* Dataset Structure
+ Data Instances
+ Data Fields
+ Data Splits
* Dataset Creation
+ Curation Rationale
+ Source Data
+ Annotations
+ Personal and Sensitive Information
* Considerations for Using the Data
+ Social Impact of Dataset
+ Discussion of Biases
+ Other Known Limitations
* Additional Information
+ Dataset Curators
+ Licensing Information
+ Citation Information
+ Contributions
Dataset Description
-------------------
* Homepage: URL
* Repository:
* Paper:
* Point of Contact:
* Size of downloaded dataset files: 3.52 GB
* Size of the generated dataset: 1.58 GB
* Total amount of disk used: 5.1 GB
### Dataset Summary
In an effort to democratize research on low-resource languages, we release UzCrawl dataset, a web and telegram crawl corpus consisting of materials from nearly 1.2 million unique sources in the Uzbek Language.
Please refer to our blogpost and paper (Coming soon!) for further details.
To load and use dataset, run this script:
Dataset Structure
-----------------
### Data Instances
#### plain\_text
* Size of downloaded dataset files: 3.52 GB
* Size of the generated dataset: 1.58 GB
* Total amount of disk used: 5.1 GB
An example of 'news' looks as follows.
### Data Fields
The data fields are the same among all splits.
* 'text': a 'string' feature that contains text.
* 'timestamp': a 'string' feature that contains timestamp of the material.
* 'source': a 'string' feature that contains url of the material.
### Data Splits
Dataset Creation
----------------
The news portion have been crawled from 21 different websites using Scrapy framework. And telegram\_blogs portion is consisted of manually curated texts from 81 high-quality Telegram channels.
Please cite this model using the following format:
Gratitude
---------
We are thankful to these awesome organizations and people for helping to make it happen:
* Asadbek Kiyomov: for his works on the beginning of the project.
* Ilya Gusev: for his advise throughout the process
* David Dale: for his advise throughout the process
Contacts
--------
We believe that this work will inspire all enthusiasts around the world to open the hidden beauty of low-resource languages, in particular of Uzbek.
For further development and issues about the dataset, please use m.mamasaidov@URL or a.shopolatov@URL to contact.
|
[
"### Dataset Summary\n\n\nIn an effort to democratize research on low-resource languages, we release UzCrawl dataset, a web and telegram crawl corpus consisting of materials from nearly 1.2 million unique sources in the Uzbek Language.\n\n\nPlease refer to our blogpost and paper (Coming soon!) for further details.\n\n\nTo load and use dataset, run this script:\n\n\nDataset Structure\n-----------------",
"### Data Instances",
"#### plain\\_text\n\n\n* Size of downloaded dataset files: 3.52 GB\n* Size of the generated dataset: 1.58 GB\n* Total amount of disk used: 5.1 GB\n\n\nAn example of 'news' looks as follows.",
"### Data Fields\n\n\nThe data fields are the same among all splits.\n\n\n* 'text': a 'string' feature that contains text.\n* 'timestamp': a 'string' feature that contains timestamp of the material.\n* 'source': a 'string' feature that contains url of the material.",
"### Data Splits\n\n\n\nDataset Creation\n----------------\n\n\nThe news portion have been crawled from 21 different websites using Scrapy framework. And telegram\\_blogs portion is consisted of manually curated texts from 81 high-quality Telegram channels.\n\n\nPlease cite this model using the following format:\n\n\nGratitude\n---------\n\n\nWe are thankful to these awesome organizations and people for helping to make it happen:\n\n\n* Asadbek Kiyomov: for his works on the beginning of the project.\n* Ilya Gusev: for his advise throughout the process\n* David Dale: for his advise throughout the process\n\n\nContacts\n--------\n\n\nWe believe that this work will inspire all enthusiasts around the world to open the hidden beauty of low-resource languages, in particular of Uzbek.\n\n\nFor further development and issues about the dataset, please use m.mamasaidov@URL or a.shopolatov@URL to contact."
] |
[
"TAGS\n#task_categories-text-generation #task_categories-fill-mask #task_ids-language-modeling #task_ids-masked-language-modeling #annotations_creators-no-annotation #multilinguality-monolingual #size_categories-1M<n<10M #language-Uzbek #license-apache-2.0 #uz #crawl #telegram_blogs #region-us \n",
"### Dataset Summary\n\n\nIn an effort to democratize research on low-resource languages, we release UzCrawl dataset, a web and telegram crawl corpus consisting of materials from nearly 1.2 million unique sources in the Uzbek Language.\n\n\nPlease refer to our blogpost and paper (Coming soon!) for further details.\n\n\nTo load and use dataset, run this script:\n\n\nDataset Structure\n-----------------",
"### Data Instances",
"#### plain\\_text\n\n\n* Size of downloaded dataset files: 3.52 GB\n* Size of the generated dataset: 1.58 GB\n* Total amount of disk used: 5.1 GB\n\n\nAn example of 'news' looks as follows.",
"### Data Fields\n\n\nThe data fields are the same among all splits.\n\n\n* 'text': a 'string' feature that contains text.\n* 'timestamp': a 'string' feature that contains timestamp of the material.\n* 'source': a 'string' feature that contains url of the material.",
"### Data Splits\n\n\n\nDataset Creation\n----------------\n\n\nThe news portion have been crawled from 21 different websites using Scrapy framework. And telegram\\_blogs portion is consisted of manually curated texts from 81 high-quality Telegram channels.\n\n\nPlease cite this model using the following format:\n\n\nGratitude\n---------\n\n\nWe are thankful to these awesome organizations and people for helping to make it happen:\n\n\n* Asadbek Kiyomov: for his works on the beginning of the project.\n* Ilya Gusev: for his advise throughout the process\n* David Dale: for his advise throughout the process\n\n\nContacts\n--------\n\n\nWe believe that this work will inspire all enthusiasts around the world to open the hidden beauty of low-resource languages, in particular of Uzbek.\n\n\nFor further development and issues about the dataset, please use m.mamasaidov@URL or a.shopolatov@URL to contact."
] |
[
108,
88,
6,
50,
73,
193
] |
[
"passage: TAGS\n#task_categories-text-generation #task_categories-fill-mask #task_ids-language-modeling #task_ids-masked-language-modeling #annotations_creators-no-annotation #multilinguality-monolingual #size_categories-1M<n<10M #language-Uzbek #license-apache-2.0 #uz #crawl #telegram_blogs #region-us \n### Dataset Summary\n\n\nIn an effort to democratize research on low-resource languages, we release UzCrawl dataset, a web and telegram crawl corpus consisting of materials from nearly 1.2 million unique sources in the Uzbek Language.\n\n\nPlease refer to our blogpost and paper (Coming soon!) for further details.\n\n\nTo load and use dataset, run this script:\n\n\nDataset Structure\n-----------------### Data Instances#### plain\\_text\n\n\n* Size of downloaded dataset files: 3.52 GB\n* Size of the generated dataset: 1.58 GB\n* Total amount of disk used: 5.1 GB\n\n\nAn example of 'news' looks as follows.### Data Fields\n\n\nThe data fields are the same among all splits.\n\n\n* 'text': a 'string' feature that contains text.\n* 'timestamp': a 'string' feature that contains timestamp of the material.\n* 'source': a 'string' feature that contains url of the material."
] |
439bacdf7eb10e988473889208a1accbeb7d81fe
|
# Dataset Card for "alt_manga"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
mickume/alt_manga
|
[
"region:us"
] |
2023-10-26T09:57:35+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 187922951, "num_examples": 1022886}], "download_size": 116412680, "dataset_size": 187922951}}
|
2023-10-26T09:58:04+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "alt_manga"
More Information needed
|
[
"# Dataset Card for \"alt_manga\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"alt_manga\"\n\nMore Information needed"
] |
[
6,
14
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"alt_manga\"\n\nMore Information needed"
] |
3515c5b8e81029a7bc21366c73336f8629054d0e
|
# Disclaimer:
this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.
# Data composition:
All data were derived from the training set portion of the open source dataset.
**gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json**：
-gsm8k_8000: https://huggingface.co/datasets/gsm8k
-dolly_15000: https://huggingface.co/datasets/databricks/databricks-dolly-15k
-cnn_dailymail_8000: https://huggingface.co/datasets/cnn_dailymail
-mmlu_17000: https://huggingface.co/datasets/cais/mmlu
-bbq_8000: https://huggingface.co/datasets/tasksource/bigbench
**lima_4kall.json**
-lima_1000: https://huggingface.co/datasets/GAIR/lima
-3000 of gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json
|
zhongshupeng/dataset_4090_1
|
[
"region:us"
] |
2023-10-26T10:12:20+00:00
|
{}
|
2023-10-26T10:54:50+00:00
|
[] |
[] |
TAGS
#region-us
|
# Disclaimer:
this dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.
# Data composition:
All data were derived from the training set portion of the open source dataset.
gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json：
-gsm8k_8000: URL
-dolly_15000: URL
-cnn_dailymail_8000: URL
-mmlu_17000: URL
-bbq_8000: URL
lima_4kall.json
-lima_1000: URL
-3000 of gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json
|
[
"# Disclaimer: \nthis dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.",
"# Data composition: \nAll data were derived from the training set portion of the open source dataset.\n\ngsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json：\n\n-gsm8k_8000: URL\n\n-dolly_15000: URL\n\n-cnn_dailymail_8000: URL\n\n-mmlu_17000: URL\n\n-bbq_8000: URL\n\n\nlima_4kall.json\n\n-lima_1000: URL\n\n-3000 of gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json"
] |
[
"TAGS\n#region-us \n",
"# Disclaimer: \nthis dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.",
"# Data composition: \nAll data were derived from the training set portion of the open source dataset.\n\ngsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json：\n\n-gsm8k_8000: URL\n\n-dolly_15000: URL\n\n-cnn_dailymail_8000: URL\n\n-mmlu_17000: URL\n\n-bbq_8000: URL\n\n\nlima_4kall.json\n\n-lima_1000: URL\n\n-3000 of gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json"
] |
[
6,
34,
141
] |
[
"passage: TAGS\n#region-us \n# Disclaimer: \nthis dataset is curated for NeurIPS 2023 LLM efficiency challange, and currently work in progress. Please use at your own risk.# Data composition: \nAll data were derived from the training set portion of the open source dataset.\n\ngsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json：\n\n-gsm8k_8000: URL\n\n-dolly_15000: URL\n\n-cnn_dailymail_8000: URL\n\n-mmlu_17000: URL\n\n-bbq_8000: URL\n\n\nlima_4kall.json\n\n-lima_1000: URL\n\n-3000 of gsm8k_dolly15k_cnnadd8k_mmlulog1.7w_bbqabc8k.json"
] |
ecfe84cbf1023d11e09e910754d7fbc0c794f287
|
Randomized sentences from books collected from Kanuri authors: Dr. Baba Kura Alkali Gazali, Lawan Dalama, Kaka Gana Abba, Lawan Hassan.
Corpus size:
- 10,281 sentences
- 90,706 words
The sentences alone are copyrighted to the authors. The compiled corpus is licensed with Attribution 4.0 International (CC BY 4.0)
This corpus was compiled for the creation of open-source language technology. To download you need to agree our terms that prohibits harmful use. If you use this corpus you must give attribution to CLEAR Global and the authors.
For citation you can use this even though this corpus was created later than the paper:
```
Alp Öktem, Muhannad Albayk Jaam, Eric DeLuca, Grace Tang
Gamayun – Language Technology for Humanitarian Response
In: 2020 IEEE Global Humanitarian Technology Conference (GHTC)
2020 October 29 - November 1; Virtual.
```
|
CLEAR-Global/kanuri-books-corpus
|
[
"size_categories:10K<n<100K",
"language:kr",
"license:cc-by-4.0",
"region:us"
] |
2023-10-26T10:27:04+00:00
|
{"language": ["kr"], "license": "cc-by-4.0", "size_categories": ["10K<n<100K"]}
|
2023-10-26T10:33:18+00:00
|
[] |
[
"kr"
] |
TAGS
#size_categories-10K<n<100K #language-Kanuri #license-cc-by-4.0 #region-us
|
Randomized sentences from books collected from Kanuri authors: Dr. Baba Kura Alkali Gazali, Lawan Dalama, Kaka Gana Abba, Lawan Hassan.
Corpus size:
- 10,281 sentences
- 90,706 words
The sentences alone are copyrighted to the authors. The compiled corpus is licensed with Attribution 4.0 International (CC BY 4.0)
This corpus was compiled for the creation of open-source language technology. To download you need to agree our terms that prohibits harmful use. If you use this corpus you must give attribution to CLEAR Global and the authors.
For citation you can use this even though this corpus was created later than the paper:
|
[] |
[
"TAGS\n#size_categories-10K<n<100K #language-Kanuri #license-cc-by-4.0 #region-us \n"
] |
[
32
] |
[
"passage: TAGS\n#size_categories-10K<n<100K #language-Kanuri #license-cc-by-4.0 #region-us \n"
] |
b618ed4be2334c660dbd64aa7bafaec8ab8fe33b
|
# Dataset Card for "accepted_pairs"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
makram93/accepted_pairs
|
[
"region:us"
] |
2023-10-26T10:32:00+00:00
|
{"dataset_info": {"features": [{"name": "url", "dtype": "string"}, {"name": "doc_id", "dtype": "string"}, {"name": "original_title", "sequence": "string"}, {"name": "right", "dtype": "string"}, {"name": "left", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 85236.05575519982, "num_examples": 100}], "download_size": 59695, "dataset_size": 85236.05575519982}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T10:53:39+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "accepted_pairs"
More Information needed
|
[
"# Dataset Card for \"accepted_pairs\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"accepted_pairs\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"accepted_pairs\"\n\nMore Information needed"
] |
f9a0ebc58a82afeca9de37c8f700c27e1146d514
|
# Dataset Card for "rejected_pairs"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
makram93/rejected_pairs
|
[
"region:us"
] |
2023-10-26T10:32:02+00:00
|
{"dataset_info": {"features": [{"name": "url", "dtype": "string"}, {"name": "doc_id", "dtype": "string"}, {"name": "original_title", "sequence": "string"}, {"name": "right", "dtype": "string"}, {"name": "left", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 85236.05575519982, "num_examples": 100}], "download_size": 58204, "dataset_size": 85236.05575519982}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T10:53:55+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "rejected_pairs"
More Information needed
|
[
"# Dataset Card for \"rejected_pairs\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"rejected_pairs\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"rejected_pairs\"\n\nMore Information needed"
] |
e4949525888d697882fb985558b4c45cb704ea1a
|
# Gamayun Language Data Kits
There are more than 7,000 languages in the world, yet only a small proportion of them have language data presence in public. CLEAR Global's Gamayun kits are a starting point for developing audio and text corpora for languages without pre-existing data resources. We create parallel data for a language by translating a pre-compiled set of general-domain sentences in English. If audio data is needed, these translated sentences are recorded by native speakers.
To scale corpus production, we offer four dataset versions:
- Mini-kit of 5,000 sentences (`kit5k`)
- Small-kit of 10,000 sentences (`kit10k`)
- Medium-kit of 15,000 sentences (`kit15k`)
- Large-kit of 30,000 sentences (`kit30k`)
For audio corpora developed using these kits refer to the official initiative website [Gamayun portal](https://gamayun.translatorswb.org/data/).
## Source sentences (`core`)
Sentences in `core` directory are in English, French and Spanish and are sourced from the [Tatoeba repository](https://tatoeba.org). Sentence selection algorithm ensures representation of most frequently used words in the language. For more information, please refer to [corepus-gen repository](https://github.com/translatorswb/corepus-gen). `etc` directories contain sentence id's as used in the Tatoeba corpus.
## Parallel corpora (`parallel`)
Translations of the kits are performed by professionals and volunteers of TWB's translator community. A complete list of translated sentences are:
| Language | Pair | # Segments | Source |
|------|--------|--------|--------|
| Hausa | English | 15,000 | Tatoeba |
| Kanuri | English | 5,000 | Tatoeba |
| Nande | French | 15,000 | Tatoeba |
| Rohingya | English | 5,000 | Tatoeba |
| Swahili (Coastal) | English | 5,000 | Tatoeba |
| Swahili (Congolese) | French | 25,302 | Tatoeba |
## Reference
More on [Gamayun, language equity initiative](https://translatorswithoutborders.org/gamayun/)
Gamayun kits are officially published in the [Gamayun portal](https://gamayun.translatorswb.org/data/). Conditions for use are described in `LICENSE.txt`.
If you need to cite Gamayun kits:
```
Alp Öktem, Muhannad Albayk Jaam, Eric DeLuca, Grace Tang
Gamayun – Language Technology for Humanitarian Response
In: 2020 IEEE Global Humanitarian Technology Conference (GHTC)
2020 October 29 - November 1; Virtual.
Link: https://ieeexplore.ieee.org/document/9342939
```
|
CLEAR-Global/Gamayun-kits
|
[
"task_categories:translation",
"size_categories:10K<n<100K",
"language:ha",
"language:kr",
"language:en",
"language:fr",
"language:sw",
"language:swc",
"language:ln",
"language:nnd",
"language:rhg",
"language:ti",
"region:us"
] |
2023-10-26T10:35:12+00:00
|
{"language": ["ha", "kr", "en", "fr", "sw", "swc", "ln", "nnd", "rhg", "ti"], "size_categories": ["10K<n<100K"], "task_categories": ["translation"], "pretty_name": "Gamayun kits"}
|
2023-10-26T10:44:10+00:00
|
[] |
[
"ha",
"kr",
"en",
"fr",
"sw",
"swc",
"ln",
"nnd",
"rhg",
"ti"
] |
TAGS
#task_categories-translation #size_categories-10K<n<100K #language-Hausa #language-Kanuri #language-English #language-French #language-Swahili (macrolanguage) #language-Congo Swahili #language-Lingala #language-West Ambae #language-Rohingya #language-Tigrinya #region-us
|
Gamayun Language Data Kits
==========================
There are more than 7,000 languages in the world, yet only a small proportion of them have language data presence in public. CLEAR Global's Gamayun kits are a starting point for developing audio and text corpora for languages without pre-existing data resources. We create parallel data for a language by translating a pre-compiled set of general-domain sentences in English. If audio data is needed, these translated sentences are recorded by native speakers.
To scale corpus production, we offer four dataset versions:
* Mini-kit of 5,000 sentences ('kit5k')
* Small-kit of 10,000 sentences ('kit10k')
* Medium-kit of 15,000 sentences ('kit15k')
* Large-kit of 30,000 sentences ('kit30k')
For audio corpora developed using these kits refer to the official initiative website Gamayun portal.
Source sentences ('core')
-------------------------
Sentences in 'core' directory are in English, French and Spanish and are sourced from the Tatoeba repository. Sentence selection algorithm ensures representation of most frequently used words in the language. For more information, please refer to corepus-gen repository. 'etc' directories contain sentence id's as used in the Tatoeba corpus.
Parallel corpora ('parallel')
-----------------------------
Translations of the kits are performed by professionals and volunteers of TWB's translator community. A complete list of translated sentences are:
Reference
---------
More on Gamayun, language equity initiative
Gamayun kits are officially published in the Gamayun portal. Conditions for use are described in 'URL'.
If you need to cite Gamayun kits:
|
[] |
[
"TAGS\n#task_categories-translation #size_categories-10K<n<100K #language-Hausa #language-Kanuri #language-English #language-French #language-Swahili (macrolanguage) #language-Congo Swahili #language-Lingala #language-West Ambae #language-Rohingya #language-Tigrinya #region-us \n"
] |
[
91
] |
[
"passage: TAGS\n#task_categories-translation #size_categories-10K<n<100K #language-Hausa #language-Kanuri #language-English #language-French #language-Swahili (macrolanguage) #language-Congo Swahili #language-Lingala #language-West Ambae #language-Rohingya #language-Tigrinya #region-us \n"
] |
0ac572373f399b4a185fc398c5cbe48564a0259e
|
# Dataset Card for "lemma41k"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
fia24/lemma41k
|
[
"region:us"
] |
2023-10-26T10:40:36+00:00
|
{"dataset_info": {"features": [{"name": "Unnamed: 0", "dtype": "int64"}, {"name": "Inflected_Word", "dtype": "string"}, {"name": "Lemma", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 2048941.980288042, "num_examples": 32995}, {"name": "test", "num_bytes": 256156.5591358743, "num_examples": 4125}, {"name": "val", "num_bytes": 256094.4605760838, "num_examples": 4124}], "download_size": 1387988, "dataset_size": 2561193.0000000005}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}, {"split": "val", "path": "data/val-*"}]}]}
|
2023-10-26T10:54:24+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "lemma41k"
More Information needed
|
[
"# Dataset Card for \"lemma41k\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"lemma41k\"\n\nMore Information needed"
] |
[
6,
14
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"lemma41k\"\n\nMore Information needed"
] |
631638e504eca043b37eedc346b0c72c9dfd34de
|
# Dataset Card for "keywords-umsa"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
riturralde/keywords-umsa
|
[
"task_categories:summarization",
"size_categories:10K<n<100K",
"language:es",
"region:us"
] |
2023-10-26T10:42:10+00:00
|
{"language": ["es"], "size_categories": ["10K<n<100K"], "task_categories": ["summarization"], "dataset_info": {"features": [{"name": "abstract", "dtype": "string"}, {"name": "title", "dtype": "string"}, {"name": "keywords", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 28203744, "num_examples": 15418}], "download_size": 15121873, "dataset_size": 28203744}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T10:45:20+00:00
|
[] |
[
"es"
] |
TAGS
#task_categories-summarization #size_categories-10K<n<100K #language-Spanish #region-us
|
# Dataset Card for "keywords-umsa"
More Information needed
|
[
"# Dataset Card for \"keywords-umsa\"\n\nMore Information needed"
] |
[
"TAGS\n#task_categories-summarization #size_categories-10K<n<100K #language-Spanish #region-us \n",
"# Dataset Card for \"keywords-umsa\"\n\nMore Information needed"
] |
[
33,
16
] |
[
"passage: TAGS\n#task_categories-summarization #size_categories-10K<n<100K #language-Spanish #region-us \n# Dataset Card for \"keywords-umsa\"\n\nMore Information needed"
] |
181a9eded9055d80737cb0a8ff181a2b0a8a8c12
|
# Dataset Card for "text-guided-vc-google-tts-api"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
hhhaaahhhaa/text-guided-vc-google-tts-api-v0
|
[
"region:us"
] |
2023-10-26T11:08:00+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "validation", "path": "data/validation-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "file_id", "dtype": "string"}, {"name": "instruction", "dtype": "string"}, {"name": "transcription", "dtype": "string"}, {"name": "src_encodec_0", "sequence": "int64"}, {"name": "src_encodec_1", "sequence": "int64"}, {"name": "src_encodec_2", "sequence": "int64"}, {"name": "src_encodec_3", "sequence": "int64"}, {"name": "src_encodec_4", "sequence": "int64"}, {"name": "src_encodec_5", "sequence": "int64"}, {"name": "src_encodec_6", "sequence": "int64"}, {"name": "src_encodec_7", "sequence": "int64"}, {"name": "tgt_encodec_0", "sequence": "int64"}, {"name": "tgt_encodec_1", "sequence": "int64"}, {"name": "tgt_encodec_2", "sequence": "int64"}, {"name": "tgt_encodec_3", "sequence": "int64"}, {"name": "tgt_encodec_4", "sequence": "int64"}, {"name": "tgt_encodec_5", "sequence": "int64"}, {"name": "tgt_encodec_6", "sequence": "int64"}, {"name": "tgt_encodec_7", "sequence": "int64"}], "splits": [{"name": "train", "num_bytes": 3704687470, "num_examples": 90000}, {"name": "validation", "num_bytes": 203094306, "num_examples": 5000}, {"name": "test", "num_bytes": 209112202, "num_examples": 5000}], "download_size": 140841385, "dataset_size": 4116893978}}
|
2023-10-27T08:43:13+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "text-guided-vc-google-tts-api"
More Information needed
|
[
"# Dataset Card for \"text-guided-vc-google-tts-api\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"text-guided-vc-google-tts-api\"\n\nMore Information needed"
] |
[
6,
23
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"text-guided-vc-google-tts-api\"\n\nMore Information needed"
] |
3d16b149010bc5fca653b463dd6465f788433eb0
|
# Dataset Card for "LORA_DATASET"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Adminhuggingface/LORA_DATASET
|
[
"region:us"
] |
2023-10-26T11:21:55+00:00
|
{"dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 6575532.0, "num_examples": 26}], "download_size": 6574426, "dataset_size": 6575532.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T11:21:55+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "LORA_DATASET"
More Information needed
|
[
"# Dataset Card for \"LORA_DATASET\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"LORA_DATASET\"\n\nMore Information needed"
] |
[
6,
16
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"LORA_DATASET\"\n\nMore Information needed"
] |
3de0baca52541165de48c976a6291401ae944985
|
# Dataset Card for "jfk_senior_thesis_data"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
pnadel/jfk_senior_thesis_data
|
[
"region:us"
] |
2023-10-26T11:42:28+00:00
|
{"dataset_info": {"features": [{"name": "index", "dtype": "int64"}, {"name": "collection", "dtype": "string"}, {"name": "packageId", "dtype": "string"}, {"name": "granuleId", "dtype": "string"}, {"name": "title", "dtype": "string"}, {"name": "detailsLink", "dtype": "string"}, {"name": "pdfLink", "dtype": "string"}, {"name": "htmlLink", "dtype": "string"}, {"name": "xmlLink", "dtype": "string"}, {"name": "otherLink1", "dtype": "string"}, {"name": "otherLink2", "dtype": "float64"}, {"name": "teaser", "dtype": "string"}, {"name": "historical", "dtype": "float64"}, {"name": "publishdate", "dtype": "string"}, {"name": "president", "dtype": "string"}, {"name": "full_text", "dtype": "string"}, {"name": "url_to_use", "dtype": "string"}, {"name": "path_to_text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 3121664312, "num_examples": 4908}], "download_size": 1609034276, "dataset_size": 3121664312}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T11:43:24+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "jfk_senior_thesis_data"
More Information needed
|
[
"# Dataset Card for \"jfk_senior_thesis_data\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"jfk_senior_thesis_data\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"jfk_senior_thesis_data\"\n\nMore Information needed"
] |
d1d1ec525bca38debd2f594ca39ef8063f8a8078
|
# Dataset Card for "hotel_dataset_pushed"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Ryan20/hotel_dataset_pushed
|
[
"region:us"
] |
2023-10-26T11:50:32+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "answers", "sequence": "string"}, {"name": "context", "dtype": "string"}, {"name": "questions", "sequence": "string"}], "splits": [{"name": "train", "num_bytes": 4634, "num_examples": 7}], "download_size": 7932, "dataset_size": 4634}}
|
2023-10-27T07:50:48+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "hotel_dataset_pushed"
More Information needed
|
[
"# Dataset Card for \"hotel_dataset_pushed\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"hotel_dataset_pushed\"\n\nMore Information needed"
] |
[
6,
17
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"hotel_dataset_pushed\"\n\nMore Information needed"
] |
f0c38f21e83f68018748a1474eacf49236a6ccb5
|
# Dataset Card for "structs_token_size_4_labelled_eval_"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
johannes-garstenauer/structs_token_size_4_labelled_eval
|
[
"region:us"
] |
2023-10-26T12:13:42+00:00
|
{"dataset_info": {"features": [{"name": "struct", "dtype": "string"}, {"name": "label", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 1323561990, "num_examples": 5451270}], "download_size": 384897245, "dataset_size": 1323561990}}
|
2023-10-26T12:14:08+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "structs_token_size_4_labelled_eval_"
More Information needed
|
[
"# Dataset Card for \"structs_token_size_4_labelled_eval_\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"structs_token_size_4_labelled_eval_\"\n\nMore Information needed"
] |
[
6,
26
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"structs_token_size_4_labelled_eval_\"\n\nMore Information needed"
] |
8909a0001f1db2acaed2586bd5a0fd1072313c45
|
# Dataset Card for "seizure_eeg_greyscale_224x224_6secWindow_adjusted"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
JLB-JLB/seizure_eeg_greyscale_224x224_6secWindow_adjusted
|
[
"region:us"
] |
2023-10-26T12:20:48+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "eval", "path": "data/eval-*"}, {"split": "dev", "path": "data/dev-*"}]}], "dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "epoch", "dtype": "int64"}, {"name": "label", "dtype": {"class_label": {"names": {"0": "seiz", "1": "bckg"}}}}], "splits": [{"name": "train", "num_bytes": 2785881214.663499, "num_examples": 93962}, {"name": "eval", "num_bytes": 446792667.3100732, "num_examples": 14910}, {"name": "dev", "num_bytes": 11628715785.0, "num_examples": 390190}], "download_size": 7728884651, "dataset_size": 14861389666.973572}}
|
2023-10-26T12:33:30+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "seizure_eeg_greyscale_224x224_6secWindow_adjusted"
More Information needed
|
[
"# Dataset Card for \"seizure_eeg_greyscale_224x224_6secWindow_adjusted\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"seizure_eeg_greyscale_224x224_6secWindow_adjusted\"\n\nMore Information needed"
] |
[
6,
35
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"seizure_eeg_greyscale_224x224_6secWindow_adjusted\"\n\nMore Information needed"
] |
2faab60cb19cd5918af39fe89ee51e7812e05934
|
# Dataset Card for Evaluation run of pszemraj/pythia-6.9b-HC3
## Dataset Description
- **Homepage:**
- **Repository:** https://huggingface.co/pszemraj/pythia-6.9b-HC3
- **Paper:**
- **Leaderboard:** https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- **Point of Contact:** [email protected]
### Dataset Summary
Dataset automatically created during the evaluation run of model [pszemraj/pythia-6.9b-HC3](https://huggingface.co/pszemraj/pythia-6.9b-HC3) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).
To load the details from a run, you can for instance do the following:
```python
from datasets import load_dataset
data = load_dataset("open-llm-leaderboard/details_pszemraj__pythia-6.9b-HC3",
"harness_winogrande_5",
split="train")
```
## Latest results
These are the [latest results from run 2023-10-26T13:43:34.818170](https://huggingface.co/datasets/open-llm-leaderboard/details_pszemraj__pythia-6.9b-HC3/blob/main/results_2023-10-26T13-43-34.818170.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
```python
{
"all": {
"em": 0.00010486577181208053,
"em_stderr": 0.00010486577181208242,
"f1": 0.022344798657718254,
"f1_stderr": 0.0006975774134342648,
"acc": 0.30386740331491713,
"acc_stderr": 0.006861200231000444
},
"harness|drop|3": {
"em": 0.00010486577181208053,
"em_stderr": 0.00010486577181208242,
"f1": 0.022344798657718254,
"f1_stderr": 0.0006975774134342648
},
"harness|gsm8k|5": {
"acc": 0.0,
"acc_stderr": 0.0
},
"harness|winogrande|5": {
"acc": 0.6077348066298343,
"acc_stderr": 0.013722400462000888
}
}
```
### Supported Tasks and Leaderboards
[More Information Needed]
### Languages
[More Information Needed]
## Dataset Structure
### Data Instances
[More Information Needed]
### Data Fields
[More Information Needed]
### Data Splits
[More Information Needed]
## Dataset Creation
### Curation Rationale
[More Information Needed]
### Source Data
#### Initial Data Collection and Normalization
[More Information Needed]
#### Who are the source language producers?
[More Information Needed]
### Annotations
#### Annotation process
[More Information Needed]
#### Who are the annotators?
[More Information Needed]
### Personal and Sensitive Information
[More Information Needed]
## Considerations for Using the Data
### Social Impact of Dataset
[More Information Needed]
### Discussion of Biases
[More Information Needed]
### Other Known Limitations
[More Information Needed]
## Additional Information
### Dataset Curators
[More Information Needed]
### Licensing Information
[More Information Needed]
### Citation Information
[More Information Needed]
### Contributions
[More Information Needed]
|
open-llm-leaderboard/details_pszemraj__pythia-6.9b-HC3
|
[
"region:us"
] |
2023-10-26T12:43:39+00:00
|
{"pretty_name": "Evaluation run of pszemraj/pythia-6.9b-HC3", "dataset_summary": "Dataset automatically created during the evaluation run of model [pszemraj/pythia-6.9b-HC3](https://huggingface.co/pszemraj/pythia-6.9b-HC3) on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)).\n\nTo load the details from a run, you can for instance do the following:\n```python\nfrom datasets import load_dataset\ndata = load_dataset(\"open-llm-leaderboard/details_pszemraj__pythia-6.9b-HC3\",\n\t\"harness_winogrande_5\",\n\tsplit=\"train\")\n```\n\n## Latest results\n\nThese are the [latest results from run 2023-10-26T13:43:34.818170](https://huggingface.co/datasets/open-llm-leaderboard/details_pszemraj__pythia-6.9b-HC3/blob/main/results_2023-10-26T13-43-34.818170.json)(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):\n\n```python\n{\n \"all\": {\n \"em\": 0.00010486577181208053,\n \"em_stderr\": 0.00010486577181208242,\n \"f1\": 0.022344798657718254,\n \"f1_stderr\": 0.0006975774134342648,\n \"acc\": 0.30386740331491713,\n \"acc_stderr\": 0.006861200231000444\n },\n \"harness|drop|3\": {\n \"em\": 0.00010486577181208053,\n \"em_stderr\": 0.00010486577181208242,\n \"f1\": 0.022344798657718254,\n \"f1_stderr\": 0.0006975774134342648\n },\n \"harness|gsm8k|5\": {\n \"acc\": 0.0,\n \"acc_stderr\": 0.0\n },\n \"harness|winogrande|5\": {\n \"acc\": 0.6077348066298343,\n \"acc_stderr\": 0.013722400462000888\n }\n}\n```", "repo_url": "https://huggingface.co/pszemraj/pythia-6.9b-HC3", "leaderboard_url": "https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard", "point_of_contact": "[email protected]", "configs": [{"config_name": "harness_drop_3", "data_files": [{"split": "2023_10_26T13_43_34.818170", "path": ["**/details_harness|drop|3_2023-10-26T13-43-34.818170.parquet"]}, {"split": "latest", "path": ["**/details_harness|drop|3_2023-10-26T13-43-34.818170.parquet"]}]}, {"config_name": "harness_gsm8k_5", "data_files": [{"split": "2023_10_26T13_43_34.818170", "path": ["**/details_harness|gsm8k|5_2023-10-26T13-43-34.818170.parquet"]}, {"split": "latest", "path": ["**/details_harness|gsm8k|5_2023-10-26T13-43-34.818170.parquet"]}]}, {"config_name": "harness_winogrande_5", "data_files": [{"split": "2023_10_26T13_43_34.818170", "path": ["**/details_harness|winogrande|5_2023-10-26T13-43-34.818170.parquet"]}, {"split": "latest", "path": ["**/details_harness|winogrande|5_2023-10-26T13-43-34.818170.parquet"]}]}, {"config_name": "results", "data_files": [{"split": "2023_10_26T13_43_34.818170", "path": ["results_2023-10-26T13-43-34.818170.parquet"]}, {"split": "latest", "path": ["results_2023-10-26T13-43-34.818170.parquet"]}]}]}
|
2023-10-26T12:43:48+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for Evaluation run of pszemraj/pythia-6.9b-HC3
## Dataset Description
- Homepage:
- Repository: URL
- Paper:
- Leaderboard: URL
- Point of Contact: clementine@URL
### Dataset Summary
Dataset automatically created during the evaluation run of model pszemraj/pythia-6.9b-HC3 on the Open LLM Leaderboard.
The dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional configuration "results" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).
To load the details from a run, you can for instance do the following:
## Latest results
These are the latest results from run 2023-10-26T13:43:34.818170(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the "latest" split for each eval):
### Supported Tasks and Leaderboards
### Languages
## Dataset Structure
### Data Instances
### Data Fields
### Data Splits
## Dataset Creation
### Curation Rationale
### Source Data
#### Initial Data Collection and Normalization
#### Who are the source language producers?
### Annotations
#### Annotation process
#### Who are the annotators?
### Personal and Sensitive Information
## Considerations for Using the Data
### Social Impact of Dataset
### Discussion of Biases
### Other Known Limitations
## Additional Information
### Dataset Curators
### Licensing Information
### Contributions
|
[
"# Dataset Card for Evaluation run of pszemraj/pythia-6.9b-HC3",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model pszemraj/pythia-6.9b-HC3 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T13:43:34.818170(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for Evaluation run of pszemraj/pythia-6.9b-HC3",
"## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL",
"### Dataset Summary\n\nDataset automatically created during the evaluation run of model pszemraj/pythia-6.9b-HC3 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:",
"## Latest results\n\nThese are the latest results from run 2023-10-26T13:43:34.818170(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):",
"### Supported Tasks and Leaderboards",
"### Languages",
"## Dataset Structure",
"### Data Instances",
"### Data Fields",
"### Data Splits",
"## Dataset Creation",
"### Curation Rationale",
"### Source Data",
"#### Initial Data Collection and Normalization",
"#### Who are the source language producers?",
"### Annotations",
"#### Annotation process",
"#### Who are the annotators?",
"### Personal and Sensitive Information",
"## Considerations for Using the Data",
"### Social Impact of Dataset",
"### Discussion of Biases",
"### Other Known Limitations",
"## Additional Information",
"### Dataset Curators",
"### Licensing Information",
"### Contributions"
] |
[
6,
21,
31,
169,
67,
10,
4,
6,
6,
5,
5,
5,
7,
4,
10,
10,
5,
5,
9,
8,
8,
7,
8,
7,
5,
6,
6,
5
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for Evaluation run of pszemraj/pythia-6.9b-HC3## Dataset Description\n\n- Homepage: \n- Repository: URL\n- Paper: \n- Leaderboard: URL\n- Point of Contact: clementine@URL### Dataset Summary\n\nDataset automatically created during the evaluation run of model pszemraj/pythia-6.9b-HC3 on the Open LLM Leaderboard.\n\nThe dataset is composed of 3 configuration, each one coresponding to one of the evaluated task.\n\nThe dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The \"train\" split is always pointing to the latest results.\n\nAn additional configuration \"results\" store all the aggregated results of the run (and is used to compute and display the agregated metrics on the Open LLM Leaderboard).\n\nTo load the details from a run, you can for instance do the following:## Latest results\n\nThese are the latest results from run 2023-10-26T13:43:34.818170(note that their might be results for other tasks in the repos if successive evals didn't cover the same tasks. You find each in the results and the \"latest\" split for each eval):### Supported Tasks and Leaderboards### Languages## Dataset Structure### Data Instances### Data Fields### Data Splits## Dataset Creation### Curation Rationale### Source Data#### Initial Data Collection and Normalization#### Who are the source language producers?### Annotations#### Annotation process#### Who are the annotators?### Personal and Sensitive Information## Considerations for Using the Data### Social Impact of Dataset### Discussion of Biases### Other Known Limitations## Additional Information### Dataset Curators### Licensing Information### Contributions"
] |
5f419c761a0c15ac3540725e790b81dd5493c0c1
|
# Dataset Card for "hindi_asr_dataset_accent"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
TheAIchemist13/hindi_asr_dataset_accent
|
[
"region:us"
] |
2023-10-26T12:44:12+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "audio", "dtype": {"audio": {"sampling_rate": 16000}}}, {"name": "transcriptions", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 60408191.0, "num_examples": 175}, {"name": "test", "num_bytes": 3850439.0, "num_examples": 5}], "download_size": 59683824, "dataset_size": 64258630.0}}
|
2023-11-01T05:26:24+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "hindi_asr_dataset_accent"
More Information needed
|
[
"# Dataset Card for \"hindi_asr_dataset_accent\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"hindi_asr_dataset_accent\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"hindi_asr_dataset_accent\"\n\nMore Information needed"
] |
3e0319203c7162b9c9f8015b594441f979c199bc
|
## Update
[01/31/2024] We update the OpanAI Moderation API results for ToxicChat (0124) based on their updated moderation model on on Jan 25, 2024.
[01/28/2024] We release an official [T5-Large model](https://huggingface.co/lmsys/toxicchat-t5-large-v1.0) trained on ToxicChat (toxicchat0124). Go and check it for you baseline comparision!
[01/19/2024] We have a new version of ToxicChat (toxicchat0124)!
## Content
This dataset contains toxicity annotations on 10K user prompts collected from the Vicuna [online demo](https://chat.lmsys.org/).
We utilize a human-AI collaborative annotation framework to guarantee the quality of annotation while maintaining a feasible annotation workload.
The details of data collection, pre-processing, and annotation can be found in our [paper](https://arxiv.org/abs/2310.17389).
We believe that ToxicChat can be a valuable resource to drive further advancements toward building a safe and healthy environment for user-AI interactions.
## Version
The version name is the update time of the dataset, e.g, 0124 means it is updated on Jan, 2024. We recommend using the latest version
for training and evaluating a model. Please make sure the version of the data is the same when comparing different models. You can use the
following code to specify the dataset version:
```python
from datasets import load_dataset
dataset = load_dataset("lmsys/toxic-chat", "toxicchat0124")
```
- **toxicchat0124** Based on version 1123, we did a model error analysis to check if there are any annotation errors and later fixed them. Each fix was checked by two annotators. The total label difference is 1.28% for toxicity labels and 0.34% for jailbreaking labels. We finally add 20 more human annotated examples which are not annotated in version 1123.
- **toxicchat1123:** The initial version.
**Basic Statistics**
| Version | 1123 | 0124 |
| --- | --- | --- |
| # User Prompts | 10,165 | 10,165 |
| # Human Annotations | 5,634 | 5,654 |
| % Toxic Examples | 7.18% | 7.33% |
| % Jailbreaking Examples | 1.78% | 2.01% |
## Model
We finetuned a [T5-large model](https://huggingface.co/lmsys/toxicchat-t5-large-v1.0) on ToxicChat (version 0124),
and you can use it as a baseline model for comparision. Note to have the same version of data.
| Model | Precision | Recall | F1 | AUPRC |
| --- | --- | --- | --- | --- |
| ToxicChat-T5-large | 0.7983 | 0.8475 | 0.8221 | 0.8850 |
| OpenAI Moderation (Updated Jan 25, 2024, threshold=0.02) | 0.5476 | 0.6989 | 0.6141 | 0.6313 |
## Disclaimer and Terms
- This dataset is based on the user query collected from the Vicuna online demo.
The Vicuna demo is fully anonymous for the users and also highlights the possible reuse of the user query data.
We have carefully gone through the data and taken out anything that could have personal information in it.
However, there is still a chance that some personal information might be left in the data.
If you come across anything in the data that you think should not be made public, please let us know right away.
- Safety and Moderation: **This dataset may contain racism, sexuality, or other undesired content.**
Before the annotation, the annotators are first notified about the toxic data that they will be annotated.
Verbal agreements were obtained before annotation.
- Non-Endorsement: Statements or opinions made in this dataset **do not reflect** the views of researchers or institutions involved in the data collection effort.
- Legal Compliance: Users of this data are responsible for ensuring its appropriate use.
The dataset should not be utilized for training dialogue agents, or any other applications, in manners that conflict with legal and ethical standards.
- Non-Identification: Users of this data agree to not attempt to determine the identity of individuals in this dataset.
## License
Both the user prompts and the model outputs are licensed under CC-BY-NC-4.0.
## Citation
```
@misc{lin2023toxicchat,
title={ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversation},
author={Zi Lin and Zihan Wang and Yongqi Tong and Yangkun Wang and Yuxin Guo and Yujia Wang and Jingbo Shang},
year={2023},
eprint={2310.17389},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
|
lmsys/toxic-chat
|
[
"task_categories:text-classification",
"size_categories:10K<n<100K",
"language:en",
"license:cc-by-nc-4.0",
"arxiv:2310.17389",
"region:us"
] |
2023-10-26T12:52:48+00:00
|
{"language": ["en"], "license": "cc-by-nc-4.0", "size_categories": ["10K<n<100K"], "task_categories": ["text-classification"], "config_names": ["toxicchat0124", "toxicchat1123"], "dataset_info": [{"config_name": "toxicchat0124", "features": [{"name": "conv_id", "dtype": "string"}, {"name": "user_input", "dtype": "string"}, {"name": "model_output", "dtype": "string"}, {"name": "human_annotation", "dtype": "bool"}, {"name": "toxicity", "dtype": "int64"}, {"name": "jailbreaking", "dtype": "int64"}, {"name": "openai_moderation", "dtype": "string"}]}, {"config_name": "toxicchat1123", "features": [{"name": "conv_id", "dtype": "string"}, {"name": "user_input", "dtype": "string"}, {"name": "model_output", "dtype": "string"}, {"name": "human_annotation", "dtype": "bool"}, {"name": "toxicity", "dtype": "int64"}, {"name": "jailbreaking", "dtype": "int64"}, {"name": "openai_moderation", "dtype": "string"}]}], "configs": [{"config_name": "toxicchat0124", "data_files": [{"split": "train", "path": "data/0124/toxic-chat_annotation_train.csv"}, {"split": "test", "path": "data/0124/toxic-chat_annotation_test.csv"}]}, {"config_name": "toxicchat1123", "data_files": [{"split": "train", "path": "data/1123/toxic-chat_annotation_train.csv"}, {"split": "test", "path": "data/1123/toxic-chat_annotation_test.csv"}]}]}
|
2024-01-31T22:51:55+00:00
|
[
"2310.17389"
] |
[
"en"
] |
TAGS
#task_categories-text-classification #size_categories-10K<n<100K #language-English #license-cc-by-nc-4.0 #arxiv-2310.17389 #region-us
|
Update
------
[01/31/2024] We update the OpanAI Moderation API results for ToxicChat (0124) based on their updated moderation model on on Jan 25, 2024.
[01/28/2024] We release an official T5-Large model trained on ToxicChat (toxicchat0124). Go and check it for you baseline comparision!
[01/19/2024] We have a new version of ToxicChat (toxicchat0124)!
Content
-------
This dataset contains toxicity annotations on 10K user prompts collected from the Vicuna online demo.
We utilize a human-AI collaborative annotation framework to guarantee the quality of annotation while maintaining a feasible annotation workload.
The details of data collection, pre-processing, and annotation can be found in our paper.
We believe that ToxicChat can be a valuable resource to drive further advancements toward building a safe and healthy environment for user-AI interactions.
Version
-------
The version name is the update time of the dataset, e.g, 0124 means it is updated on Jan, 2024. We recommend using the latest version
for training and evaluating a model. Please make sure the version of the data is the same when comparing different models. You can use the
following code to specify the dataset version:
* toxicchat0124 Based on version 1123, we did a model error analysis to check if there are any annotation errors and later fixed them. Each fix was checked by two annotators. The total label difference is 1.28% for toxicity labels and 0.34% for jailbreaking labels. We finally add 20 more human annotated examples which are not annotated in version 1123.
* toxicchat1123: The initial version.
Basic Statistics
Version: # User Prompts, 1123: 10,165, 0124: 10,165
Version: # Human Annotations, 1123: 5,634, 0124: 5,654
Version: % Toxic Examples, 1123: 7.18%, 0124: 7.33%
Version: % Jailbreaking Examples, 1123: 1.78%, 0124: 2.01%
Model
-----
We finetuned a T5-large model on ToxicChat (version 0124),
and you can use it as a baseline model for comparision. Note to have the same version of data.
Disclaimer and Terms
--------------------
* This dataset is based on the user query collected from the Vicuna online demo.
The Vicuna demo is fully anonymous for the users and also highlights the possible reuse of the user query data.
We have carefully gone through the data and taken out anything that could have personal information in it.
However, there is still a chance that some personal information might be left in the data.
If you come across anything in the data that you think should not be made public, please let us know right away.
* Safety and Moderation: This dataset may contain racism, sexuality, or other undesired content.
Before the annotation, the annotators are first notified about the toxic data that they will be annotated.
Verbal agreements were obtained before annotation.
* Non-Endorsement: Statements or opinions made in this dataset do not reflect the views of researchers or institutions involved in the data collection effort.
* Legal Compliance: Users of this data are responsible for ensuring its appropriate use.
The dataset should not be utilized for training dialogue agents, or any other applications, in manners that conflict with legal and ethical standards.
* Non-Identification: Users of this data agree to not attempt to determine the identity of individuals in this dataset.
License
-------
Both the user prompts and the model outputs are licensed under CC-BY-NC-4.0.
|
[
"# User Prompts, 1123: 10,165, 0124: 10,165\nVersion: # Human Annotations, 1123: 5,634, 0124: 5,654\nVersion: % Toxic Examples, 1123: 7.18%, 0124: 7.33%\nVersion: % Jailbreaking Examples, 1123: 1.78%, 0124: 2.01%\n\n\nModel\n-----\n\n\nWe finetuned a T5-large model on ToxicChat (version 0124),\nand you can use it as a baseline model for comparision. Note to have the same version of data.\n\n\n\nDisclaimer and Terms\n--------------------\n\n\n* This dataset is based on the user query collected from the Vicuna online demo.\nThe Vicuna demo is fully anonymous for the users and also highlights the possible reuse of the user query data.\nWe have carefully gone through the data and taken out anything that could have personal information in it.\nHowever, there is still a chance that some personal information might be left in the data.\nIf you come across anything in the data that you think should not be made public, please let us know right away.\n* Safety and Moderation: This dataset may contain racism, sexuality, or other undesired content.\nBefore the annotation, the annotators are first notified about the toxic data that they will be annotated.\nVerbal agreements were obtained before annotation.\n* Non-Endorsement: Statements or opinions made in this dataset do not reflect the views of researchers or institutions involved in the data collection effort.\n* Legal Compliance: Users of this data are responsible for ensuring its appropriate use.\nThe dataset should not be utilized for training dialogue agents, or any other applications, in manners that conflict with legal and ethical standards.\n* Non-Identification: Users of this data agree to not attempt to determine the identity of individuals in this dataset.\n\n\nLicense\n-------\n\n\nBoth the user prompts and the model outputs are licensed under CC-BY-NC-4.0."
] |
[
"TAGS\n#task_categories-text-classification #size_categories-10K<n<100K #language-English #license-cc-by-nc-4.0 #arxiv-2310.17389 #region-us \n",
"# User Prompts, 1123: 10,165, 0124: 10,165\nVersion: # Human Annotations, 1123: 5,634, 0124: 5,654\nVersion: % Toxic Examples, 1123: 7.18%, 0124: 7.33%\nVersion: % Jailbreaking Examples, 1123: 1.78%, 0124: 2.01%\n\n\nModel\n-----\n\n\nWe finetuned a T5-large model on ToxicChat (version 0124),\nand you can use it as a baseline model for comparision. Note to have the same version of data.\n\n\n\nDisclaimer and Terms\n--------------------\n\n\n* This dataset is based on the user query collected from the Vicuna online demo.\nThe Vicuna demo is fully anonymous for the users and also highlights the possible reuse of the user query data.\nWe have carefully gone through the data and taken out anything that could have personal information in it.\nHowever, there is still a chance that some personal information might be left in the data.\nIf you come across anything in the data that you think should not be made public, please let us know right away.\n* Safety and Moderation: This dataset may contain racism, sexuality, or other undesired content.\nBefore the annotation, the annotators are first notified about the toxic data that they will be annotated.\nVerbal agreements were obtained before annotation.\n* Non-Endorsement: Statements or opinions made in this dataset do not reflect the views of researchers or institutions involved in the data collection effort.\n* Legal Compliance: Users of this data are responsible for ensuring its appropriate use.\nThe dataset should not be utilized for training dialogue agents, or any other applications, in manners that conflict with legal and ethical standards.\n* Non-Identification: Users of this data agree to not attempt to determine the identity of individuals in this dataset.\n\n\nLicense\n-------\n\n\nBoth the user prompts and the model outputs are licensed under CC-BY-NC-4.0."
] |
[
53,
439
] |
[
"passage: TAGS\n#task_categories-text-classification #size_categories-10K<n<100K #language-English #license-cc-by-nc-4.0 #arxiv-2310.17389 #region-us \n# User Prompts, 1123: 10,165, 0124: 10,165\nVersion: # Human Annotations, 1123: 5,634, 0124: 5,654\nVersion: % Toxic Examples, 1123: 7.18%, 0124: 7.33%\nVersion: % Jailbreaking Examples, 1123: 1.78%, 0124: 2.01%\n\n\nModel\n-----\n\n\nWe finetuned a T5-large model on ToxicChat (version 0124),\nand you can use it as a baseline model for comparision. Note to have the same version of data.\n\n\n\nDisclaimer and Terms\n--------------------\n\n\n* This dataset is based on the user query collected from the Vicuna online demo.\nThe Vicuna demo is fully anonymous for the users and also highlights the possible reuse of the user query data.\nWe have carefully gone through the data and taken out anything that could have personal information in it.\nHowever, there is still a chance that some personal information might be left in the data.\nIf you come across anything in the data that you think should not be made public, please let us know right away.\n* Safety and Moderation: This dataset may contain racism, sexuality, or other undesired content.\nBefore the annotation, the annotators are first notified about the toxic data that they will be annotated.\nVerbal agreements were obtained before annotation.\n* Non-Endorsement: Statements or opinions made in this dataset do not reflect the views of researchers or institutions involved in the data collection effort.\n* Legal Compliance: Users of this data are responsible for ensuring its appropriate use.\nThe dataset should not be utilized for training dialogue agents, or any other applications, in manners that conflict with legal and ethical standards.\n* Non-Identification: Users of this data agree to not attempt to determine the identity of individuals in this dataset.\n\n\nLicense\n-------\n\n\nBoth the user prompts and the model outputs are licensed under CC-BY-NC-4.0."
] |
5d833feae2b93c710cf6d030af1dc3779066abc5
|
# Dataset Card for "nci_nq_t5_tokenized"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
carnival13/nci_nq_t5_tokenized
|
[
"region:us"
] |
2023-10-26T12:53:14+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "eval", "path": "data/eval-*"}, {"split": "eval_zero_shot", "path": "data/eval_zero_shot-*"}, {"split": "eval_normal", "path": "data/eval_normal-*"}]}], "dataset_info": {"features": [{"name": "input", "dtype": "string"}, {"name": "label", "sequence": "int64"}, {"name": "input_ids", "sequence": "int32"}, {"name": "attention_mask", "sequence": "int8"}], "splits": [{"name": "train", "num_bytes": 137430914, "num_examples": 177638}, {"name": "eval", "num_bytes": 1529607, "num_examples": 7830}, {"name": "eval_zero_shot", "num_bytes": 562161, "num_examples": 2859}, {"name": "eval_normal", "num_bytes": 967446, "num_examples": 4971}], "download_size": 61636686, "dataset_size": 140490128}}
|
2023-10-26T12:53:37+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "nci_nq_t5_tokenized"
More Information needed
|
[
"# Dataset Card for \"nci_nq_t5_tokenized\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"nci_nq_t5_tokenized\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"nci_nq_t5_tokenized\"\n\nMore Information needed"
] |
f3695aa2168cbe56659c1b5fb9cb939e6ec0f7e0
|
<p align="center"><h1>🧠 Awesome ChatGPT Prompts [CSV dataset]</h1></p>
This is a Dataset Repository of **Awesome ChatGPT Prompts**
**[View All Prompts on GitHub](https://github.com/f/awesome-chatgpt-prompts)**
# License
CC-0
|
MustafaSuleyman/real-toxicity-prompts
|
[
"license:cc0-1.0",
"ChatGPT",
"region:us"
] |
2023-10-26T12:55:11+00:00
|
{"license": "cc0-1.0", "tags": ["ChatGPT"]}
|
2023-11-10T14:33:02+00:00
|
[] |
[] |
TAGS
#license-cc0-1.0 #ChatGPT #region-us
|
<p align="center"><h1> Awesome ChatGPT Prompts [CSV dataset]</h1></p>
This is a Dataset Repository of Awesome ChatGPT Prompts
View All Prompts on GitHub
# License
CC-0
|
[
"# License\n\nCC-0"
] |
[
"TAGS\n#license-cc0-1.0 #ChatGPT #region-us \n",
"# License\n\nCC-0"
] |
[
18,
4
] |
[
"passage: TAGS\n#license-cc0-1.0 #ChatGPT #region-us \n# License\n\nCC-0"
] |
1a2b4edfc5285e6a4b467cb984eb5ef16f12b423
|
# Dataset Card for "nci_nq_t5_naive"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
carnival13/nci_nq_t5_naive
|
[
"region:us"
] |
2023-10-26T13:07:35+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "eval", "path": "data/eval-*"}, {"split": "eval_zero_shot", "path": "data/eval_zero_shot-*"}, {"split": "eval_normal", "path": "data/eval_normal-*"}]}], "dataset_info": {"features": [{"name": "input", "dtype": "string"}, {"name": "label", "sequence": "int64"}, {"name": "input_ids", "sequence": "int32"}, {"name": "attention_mask", "sequence": "int8"}], "splits": [{"name": "train", "num_bytes": 137430914, "num_examples": 177638}, {"name": "eval", "num_bytes": 1529607, "num_examples": 7830}, {"name": "eval_zero_shot", "num_bytes": 562161, "num_examples": 2859}, {"name": "eval_normal", "num_bytes": 967446, "num_examples": 4971}], "download_size": 61636683, "dataset_size": 140490128}}
|
2023-10-26T13:07:57+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "nci_nq_t5_naive"
More Information needed
|
[
"# Dataset Card for \"nci_nq_t5_naive\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"nci_nq_t5_naive\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"nci_nq_t5_naive\"\n\nMore Information needed"
] |
587c8c380d1a28f5557005dfb98cc72e5c458415
|
# Dataset Card for "tencent_data_encodec"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
zion84006/tencent_data_encodec
|
[
"region:us"
] |
2023-10-26T13:18:56+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "valid", "path": "data/valid-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "file_id", "dtype": "int64"}, {"name": "instruction", "dtype": "string"}, {"name": "transcription", "dtype": "string"}, {"name": "src_encodec_0", "sequence": "int64"}, {"name": "src_encodec_1", "sequence": "int64"}, {"name": "src_encodec_2", "sequence": "int64"}, {"name": "src_encodec_3", "sequence": "int64"}, {"name": "src_encodec_4", "sequence": "int64"}, {"name": "src_encodec_5", "sequence": "int64"}, {"name": "src_encodec_6", "sequence": "int64"}, {"name": "src_encodec_7", "sequence": "int64"}, {"name": "tgt_encodec_0", "sequence": "int64"}, {"name": "tgt_encodec_1", "sequence": "int64"}, {"name": "tgt_encodec_2", "sequence": "int64"}, {"name": "tgt_encodec_3", "sequence": "int64"}, {"name": "tgt_encodec_4", "sequence": "int64"}, {"name": "tgt_encodec_5", "sequence": "int64"}, {"name": "tgt_encodec_6", "sequence": "int64"}, {"name": "tgt_encodec_7", "sequence": "int64"}], "splits": [{"name": "train", "num_bytes": 18586613483, "num_examples": 266780}, {"name": "valid", "num_bytes": 527894882, "num_examples": 7620}, {"name": "test", "num_bytes": 508453304, "num_examples": 7620}], "download_size": 472185815, "dataset_size": 19622961669}}
|
2023-10-26T20:43:04+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "tencent_data_encodec"
More Information needed
|
[
"# Dataset Card for \"tencent_data_encodec\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"tencent_data_encodec\"\n\nMore Information needed"
] |
[
6,
17
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"tencent_data_encodec\"\n\nMore Information needed"
] |
c2ed7049f35b030bd5259595d7c158c84e70034c
|
The phrases in Phrasal Concordance are chunked by Strongs numbers. The Cross References have THREE levels: carat (columns), comma (references), and percent (votes).
BLB (Blue Letter Bible) Greek
-
GroupID ^ ReferenceID ^ Reference
BLB Hebrew
-
GroupID ^ ReferenceID ^ Reference
Chained Phrasal Concordances
-
Reference ^ StrongsChunkedPhraseChain
Cross References
-
RowID ^ ReferencedPassage ^ ReferencingPassagesWithVoteCounts
Phrasal Concordance
-
Phrase ^ Reference ^ Count
|
JWBickel/Concordances_And_Cross_References
|
[
"language:en",
"region:us"
] |
2023-10-26T13:21:14+00:00
|
{"language": ["en"]}
|
2023-11-01T21:52:34+00:00
|
[] |
[
"en"
] |
TAGS
#language-English #region-us
|
The phrases in Phrasal Concordance are chunked by Strongs numbers. The Cross References have THREE levels: carat (columns), comma (references), and percent (votes).
BLB (Blue Letter Bible) Greek
-
GroupID ^ ReferenceID ^ Reference
BLB Hebrew
-
GroupID ^ ReferenceID ^ Reference
Chained Phrasal Concordances
-
Reference ^ StrongsChunkedPhraseChain
Cross References
-
RowID ^ ReferencedPassage ^ ReferencingPassagesWithVoteCounts
Phrasal Concordance
-
Phrase ^ Reference ^ Count
|
[] |
[
"TAGS\n#language-English #region-us \n"
] |
[
10
] |
[
"passage: TAGS\n#language-English #region-us \n"
] |
467cd4194f47d81a542dd29dcb535d7f97780cfc
|
# Dataset Card for "humansleepproject-small-individuals"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
emi429/humansleepproject-small-individuals
|
[
"region:us"
] |
2023-10-26T13:31:15+00:00
|
{"dataset_info": {"features": [{"name": "rr_intervals", "dtype": "int64"}, {"name": "sleep_stage", "dtype": "int64"}, {"name": "patient_id", "dtype": "int64"}], "splits": [{"name": "test", "num_bytes": 12096, "num_examples": 504}, {"name": "train", "num_bytes": 49680, "num_examples": 2070}], "download_size": 47116, "dataset_size": 61776}}
|
2023-10-26T17:18:10+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "humansleepproject-small-individuals"
More Information needed
|
[
"# Dataset Card for \"humansleepproject-small-individuals\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"humansleepproject-small-individuals\"\n\nMore Information needed"
] |
[
6,
19
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"humansleepproject-small-individuals\"\n\nMore Information needed"
] |
b5d6229872e322a08d34a84202e54ba8563faeef
|
1,863,903,840 samples.
Contains only the unique, RDKit canonicalized SMILES molecules in a CSV format (after extracting), from the ZINC-20 dataset found at https://zinc20.docking.org/tranches/home/ with react set to standard and purchasability set to annotated. ZINC-20 compounds collected in 17 August 2023.
|
Pixelatory/ZINC-20
|
[
"size_categories:1B<n<10B",
"chemistry",
"biology",
"region:us"
] |
2023-10-26T13:46:31+00:00
|
{"size_categories": ["1B<n<10B"], "tags": ["chemistry", "biology"]}
|
2023-10-26T14:55:08+00:00
|
[] |
[] |
TAGS
#size_categories-1B<n<10B #chemistry #biology #region-us
|
1,863,903,840 samples.
Contains only the unique, RDKit canonicalized SMILES molecules in a CSV format (after extracting), from the ZINC-20 dataset found at URL with react set to standard and purchasability set to annotated. ZINC-20 compounds collected in 17 August 2023.
|
[] |
[
"TAGS\n#size_categories-1B<n<10B #chemistry #biology #region-us \n"
] |
[
25
] |
[
"passage: TAGS\n#size_categories-1B<n<10B #chemistry #biology #region-us \n"
] |
deafef3fb78bf9b555eedc6aeebcc4232a34bc0f
|
# Dataset Card for "newsqa"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
legacy107/newsqa
|
[
"region:us"
] |
2023-10-26T13:46:50+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "validation", "path": "data/validation-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "context", "dtype": "string"}, {"name": "question", "dtype": "string"}, {"name": "answers", "sequence": "string"}, {"name": "key", "dtype": "string"}, {"name": "labels", "list": [{"name": "end", "sequence": "int64"}, {"name": "start", "sequence": "int64"}]}, {"name": "document_id", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 221702291, "num_examples": 69960}, {"name": "validation", "num_bytes": 13599482, "num_examples": 4200}, {"name": "test", "num_bytes": 13268158, "num_examples": 4212}], "download_size": 31455725, "dataset_size": 248569931}}
|
2023-10-31T10:03:58+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "newsqa"
More Information needed
|
[
"# Dataset Card for \"newsqa\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"newsqa\"\n\nMore Information needed"
] |
[
6,
12
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"newsqa\"\n\nMore Information needed"
] |
84774340094ef7bafa929a9261404644babac987
|
# Dataset Card for "oasst_lima"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
ycchen/oasst_lima
|
[
"region:us"
] |
2023-10-26T13:47:30+00:00
|
{"dataset_info": {"features": [{"name": "conversations", "sequence": "string"}, {"name": "source", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 7255984, "num_examples": 4538}], "download_size": 4147275, "dataset_size": 7255984}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T13:53:20+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "oasst_lima"
More Information needed
|
[
"# Dataset Card for \"oasst_lima\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"oasst_lima\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"oasst_lima\"\n\nMore Information needed"
] |
33b4495c9bbca068084fca564e0883b0a7fd1685
|
There are 3 files. MobyIndex is a CSV file with a carat as the delimiter and a comma as an INTERNAL DELIMITER.
KJVSynonyms uses carats, but there's only one column, so it's like an internal delimiter, breaking up each group of synonyms.
Synonyms Index has a one to many relationship between words and group numbers. It's similar to KJVSynonyms in structure, but it has two columns where the FIRST carat separates the text from group numbers.
MobyIndex
----
Synonym ^ IDs,
KJVSynonyms
-----------
Synonym1 ^ Synonym2 ^ ..
SynonymsIndex
------------
Synonym ^ Group1 ^ Group2 ^ ...
|
JWBickel/Synonyms
|
[
"language:en",
"region:us"
] |
2023-10-26T13:48:04+00:00
|
{"language": ["en"], "pretty_name": "KJV Synonyms"}
|
2023-10-29T18:01:06+00:00
|
[] |
[
"en"
] |
TAGS
#language-English #region-us
|
There are 3 files. MobyIndex is a CSV file with a carat as the delimiter and a comma as an INTERNAL DELIMITER.
KJVSynonyms uses carats, but there's only one column, so it's like an internal delimiter, breaking up each group of synonyms.
Synonyms Index has a one to many relationship between words and group numbers. It's similar to KJVSynonyms in structure, but it has two columns where the FIRST carat separates the text from group numbers.
MobyIndex
----
Synonym ^ IDs,
KJVSynonyms
-----------
Synonym1 ^ Synonym2 ^ ..
SynonymsIndex
------------
Synonym ^ Group1 ^ Group2 ^ ...
|
[] |
[
"TAGS\n#language-English #region-us \n"
] |
[
10
] |
[
"passage: TAGS\n#language-English #region-us \n"
] |
733ebe89726e2e4fe2ee704bcdbf23a343a9f5eb
|
There are 4 separate lexicons, each in a CSV file. All the files are delimited by carats. Two of the files also include internal delimiters.
Blue Letter Bible (BLB) Hebrew
------------------------------
LexicalID ^ Total Translated ^ Connection ^ Gender ^ POS ^ Pronunciation ^ Transliteration ^ TWOT Number ^ IsAramaic ^ IsRoot ^ ~Translation Counts% ^ ~Lexical Hierarchy%
BLB Greek
---------
LexicalID ^ IsRoot ^ Total Translated ^ Connection ^ Gender ^ POS ^ Pronunciation ^ Transliteration ^ TWOT Number ^ ~Conjugated% ^ ~Translation Counts% ^ ~Extra TDNT Information% ^ ~Lexical Hierarchy%
STEPBible Hebrew
----------------
Line Number ^ Part Number ^ Parse ^ Strongs Number ^ Word Order ID ^ Translation Order ID ^ Translation ^ Translation Alternate ^ Translation Alternate 2 ^ Translation Semantic
NT Words
--------
Gloss ^ Book Number ^ Chapter Number ^ Verse Number ^ Word ID ^ Word Function ^ Clause Level ^ Clause ^ Clause Function ^ Subclause ^ Subclause Function ^ Greek Clause Level ^ Greek ID ^ Strongs ^ Morphology ^ Greek Word
|
JWBickel/Lexicons
|
[
"language:en",
"region:us"
] |
2023-10-26T14:05:32+00:00
|
{"language": ["en"]}
|
2023-10-26T15:05:54+00:00
|
[] |
[
"en"
] |
TAGS
#language-English #region-us
|
There are 4 separate lexicons, each in a CSV file. All the files are delimited by carats. Two of the files also include internal delimiters.
Blue Letter Bible (BLB) Hebrew
------------------------------
LexicalID ^ Total Translated ^ Connection ^ Gender ^ POS ^ Pronunciation ^ Transliteration ^ TWOT Number ^ IsAramaic ^ IsRoot ^ ~Translation Counts% ^ ~Lexical Hierarchy%
BLB Greek
---------
LexicalID ^ IsRoot ^ Total Translated ^ Connection ^ Gender ^ POS ^ Pronunciation ^ Transliteration ^ TWOT Number ^ ~Conjugated% ^ ~Translation Counts% ^ ~Extra TDNT Information% ^ ~Lexical Hierarchy%
STEPBible Hebrew
----------------
Line Number ^ Part Number ^ Parse ^ Strongs Number ^ Word Order ID ^ Translation Order ID ^ Translation ^ Translation Alternate ^ Translation Alternate 2 ^ Translation Semantic
NT Words
--------
Gloss ^ Book Number ^ Chapter Number ^ Verse Number ^ Word ID ^ Word Function ^ Clause Level ^ Clause ^ Clause Function ^ Subclause ^ Subclause Function ^ Greek Clause Level ^ Greek ID ^ Strongs ^ Morphology ^ Greek Word
|
[] |
[
"TAGS\n#language-English #region-us \n"
] |
[
10
] |
[
"passage: TAGS\n#language-English #region-us \n"
] |
92e6296eeae7d9eac5c6266ad8f9b5fbca0c1faa
|
These are KJV phrases and their counts, chunked by Strong's.
It's a CSV file, delimited by carats.
-------------------------------------
RowID ^ StrongsChunkedPhrase ^ Count
_____________________________________
Note that the first record is nonsense - it's just a space. Taking it out would have thrown off the Row IDs. Don't overlook it (but overlook my flaw).
|
JWBickel/StrongsChunked_English_Phrase_Counts
|
[
"size_categories:10K<n<100K",
"language:en",
"region:us"
] |
2023-10-26T14:14:50+00:00
|
{"language": ["en"], "size_categories": ["10K<n<100K"]}
|
2023-10-26T14:53:22+00:00
|
[] |
[
"en"
] |
TAGS
#size_categories-10K<n<100K #language-English #region-us
|
These are KJV phrases and their counts, chunked by Strong's.
It's a CSV file, delimited by carats.
-------------------------------------
RowID ^ StrongsChunkedPhrase ^ Count
_____________________________________
Note that the first record is nonsense - it's just a space. Taking it out would have thrown off the Row IDs. Don't overlook it (but overlook my flaw).
|
[] |
[
"TAGS\n#size_categories-10K<n<100K #language-English #region-us \n"
] |
[
22
] |
[
"passage: TAGS\n#size_categories-10K<n<100K #language-English #region-us \n"
] |
4025d704e06f53f85dd2542f8525caea251a6bce
|
There are 3 files
-----------------
Two are CSV files, delimited by a carat. The other one is a text file which also has carats, but only the first one is a normal column delimiter. The other carats separate the words in each verse.
Strongs Connections and Derivations.csv
---------------------------------------
Strongs Number ^ Derived From ^ Derived By
MultipleStrongs with Gloss and Counts.csv
-----------------------------------------
PhraseID ^ ComplexPhraseStrongsSequence ^ ComplexPhraseEnglishText ^ PhraseCount
KJV Strongs-Chunked Alternate Glosses.txt
-----------------------------------------
VerseID ^ word1-gloss1, word1-gloss2, .. ^ word2-gloss1, ..
|
JWBickel/Strongs
|
[
"language:en",
"region:us"
] |
2023-10-26T14:15:49+00:00
|
{"language": ["en"], "pretty_name": "Strongs"}
|
2023-10-26T14:46:41+00:00
|
[] |
[
"en"
] |
TAGS
#language-English #region-us
|
There are 3 files
-----------------
Two are CSV files, delimited by a carat. The other one is a text file which also has carats, but only the first one is a normal column delimiter. The other carats separate the words in each verse.
Strongs Connections and URL
---------------------------------------
Strongs Number ^ Derived From ^ Derived By
MultipleStrongs with Gloss and URL
-----------------------------------------
PhraseID ^ ComplexPhraseStrongsSequence ^ ComplexPhraseEnglishText ^ PhraseCount
KJV Strongs-Chunked Alternate URL
-----------------------------------------
VerseID ^ word1-gloss1, word1-gloss2, .. ^ word2-gloss1, ..
|
[] |
[
"TAGS\n#language-English #region-us \n"
] |
[
10
] |
[
"passage: TAGS\n#language-English #region-us \n"
] |
6c8f61f66cd6541340b5ad6e52718d0897fb7b17
|
# Dataset Card for KJVWordCounts
This is a simple list of every word and their total counts in the King James version of the Bible.
## Dataset Structure
It's a CSV file, delimited with a carat.
Word ^ Count
## Dataset Card Author
Jeremy Bickel
## Dataset Card Contact
[email protected]
|
JWBickel/KJVWordCounts
|
[
"size_categories:10K<n<100K",
"language:en",
"region:us"
] |
2023-10-26T14:19:53+00:00
|
{"language": ["en"], "size_categories": ["10K<n<100K"], "pretty_name": "KJV Word Counts"}
|
2023-10-26T14:34:15+00:00
|
[] |
[
"en"
] |
TAGS
#size_categories-10K<n<100K #language-English #region-us
|
# Dataset Card for KJVWordCounts
This is a simple list of every word and their total counts in the King James version of the Bible.
## Dataset Structure
It's a CSV file, delimited with a carat.
Word ^ Count
## Dataset Card Author
Jeremy Bickel
## Dataset Card Contact
JeremyWBickel@URL
|
[
"# Dataset Card for KJVWordCounts\n\nThis is a simple list of every word and their total counts in the King James version of the Bible.",
"## Dataset Structure\n\nIt's a CSV file, delimited with a carat.\n\nWord ^ Count",
"## Dataset Card Author\n\nJeremy Bickel",
"## Dataset Card Contact\n\nJeremyWBickel@URL"
] |
[
"TAGS\n#size_categories-10K<n<100K #language-English #region-us \n",
"# Dataset Card for KJVWordCounts\n\nThis is a simple list of every word and their total counts in the King James version of the Bible.",
"## Dataset Structure\n\nIt's a CSV file, delimited with a carat.\n\nWord ^ Count",
"## Dataset Card Author\n\nJeremy Bickel",
"## Dataset Card Contact\n\nJeremyWBickel@URL"
] |
[
22,
33,
26,
8,
11
] |
[
"passage: TAGS\n#size_categories-10K<n<100K #language-English #region-us \n# Dataset Card for KJVWordCounts\n\nThis is a simple list of every word and their total counts in the King James version of the Bible.## Dataset Structure\n\nIt's a CSV file, delimited with a carat.\n\nWord ^ Count## Dataset Card Author\n\nJeremy Bickel## Dataset Card Contact\n\nJeremyWBickel@URL"
] |
83c2fc1957d82bcc717616d7530ae337a0f97e46
|
# DALLE3 Paper Images
Extracted examples shown in dalle-3 paper without any compression: https://cdn.openai.com/papers/dall-e-3.pdf
Used https://pdfcandy.com/extract-images.html to extract images from dalle-3 paper.
Do not include three images that caption may not be correct.
|
Maxlinn/dalle3-paper-images
|
[
"region:us"
] |
2023-10-26T14:22:53+00:00
|
{}
|
2023-10-26T14:24:55+00:00
|
[] |
[] |
TAGS
#region-us
|
# DALLE3 Paper Images
Extracted examples shown in dalle-3 paper without any compression: URL
Used URL to extract images from dalle-3 paper.
Do not include three images that caption may not be correct.
|
[
"# DALLE3 Paper Images\n\nExtracted examples shown in dalle-3 paper without any compression: URL\n\nUsed URL to extract images from dalle-3 paper.\n\nDo not include three images that caption may not be correct."
] |
[
"TAGS\n#region-us \n",
"# DALLE3 Paper Images\n\nExtracted examples shown in dalle-3 paper without any compression: URL\n\nUsed URL to extract images from dalle-3 paper.\n\nDo not include three images that caption may not be correct."
] |
[
6,
44
] |
[
"passage: TAGS\n#region-us \n# DALLE3 Paper Images\n\nExtracted examples shown in dalle-3 paper without any compression: URL\n\nUsed URL to extract images from dalle-3 paper.\n\nDo not include three images that caption may not be correct."
] |
4239b74dd40c75e74d50effe3d3805bd068cf637
|
# LoLLMs-QNA Dataset
## Dataset Description
The LoLLMs-QNA dataset was created by ParisNeo. The dataset is based on the documentation and knowledge base developed for LoLLMs. It aims to provide a comprehensive collection of questions and corresponding answers related to LoLLMs and its functionalities.
The dataset is structured as a JSON file, with each entry consisting of a question and its corresponding answer. The questions cover various aspects of LoLLMs, including installation, features, functionalities, system requirements, and comparisons with other similar tools. The answers provide detailed information and instructions to assist users in understanding and utilizing LoLLMs effectively.
It is important to note that the dataset also contains some generic thoughts and reflections about AI and its potential uses and threats. These thoughts represent ParisNeo's personal views and should not be considered as a universally accepted truth.
## Dataset Creation Process
The LoLLMs-QNA dataset was created using a two-step process outlined in ParisNeo's white paper titled "From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training." The process involves extracting questions from raw text and then utilizing a vectorized version of the raw data along with an LLM to generate answers.
The raw text used for question extraction includes the documentation and knowledge base developed for LoLLMs, along with ParisNeo's personal insights and expertise in the field of AI. The questions were then manually crafted from this raw text to cover a wide range of topics related to LoLLMs.
To generate the answers, a vectorized version of the raw data was created, along with an LLM model trained on the specific domain of LoLLMs. The LLM model was then used to generate accurate and informative answers to the extracted questions.
## Used Models
This database was built using Database Maker on LoLLMs.
Database Maker implements the algorithm presented in the white paper [From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training](https://huggingface.co/datasets/ParisNeo/lollms_aware_dataset/resolve/main/lollms_db_building_strategy.pdf)
To do the LLM tasks required to generate the questions and answers, I used the [airoboros-l2-70b-2.2.1](https://huggingface.co/jondurbin/airoboros-l2-70b-2.2.1) model by [jondurbin](https://huggingface.co/jondurbin).
Updates to this database will come as LoLLMs documentation evolve and new functionalities are added constantly. So stay tuned.
## Dataset Format
The LoLLMs-QNA dataset is provided as a JSON file. Each entry in the dataset consists of a dictionary with two key-value pairs:
- "question": The question posed by the user.
- "answer": The corresponding answer to the question.
Example entry:
```
{
"question": "What are the features of Lollms-webui?",
"answer": "The features of Lollms-webui include:..."
}
```
## Usage and Disclaimer
The LoLLMs-QNA dataset is intended to be used for various tasks, including training AI models, developing chatbots, and assisting users in understanding and utilizing LoLLMs. However, it is important to note that the dataset reflects ParisNeo's personal vision and perspectives about AI and LoLLMs. The answers provided in the dataset should not be considered as universally accepted truths, but rather as ParisNeo's personal insights and instructions.
It is recommended to use the dataset in conjunction with other sources of information and to verify the accuracy and relevance of the answers provided. Users should exercise critical thinking and consider the specific context and requirements of their own applications and use cases.
## Acknowledgments
ParisNeo would like to express gratitude to the open-source community and contributors who have supported the development and improvement of LoLLMs. The dataset is provided as a contribution back to the community and aims to facilitate the understanding and utilization of LoLLMs.
## Special thanks
Special Thanks to [jondurbin](https://huggingface.co/jondurbin) for his advices and for providing the LLM that was used to build this dataset.
Also special thanks to [Tom Jobbins](https://huggingface.co/TheBloke) for quantizing the model that was used to build this database.
## Licence
Apache 2.0.
|
ParisNeo/lollms_aware_dataset
|
[
"task_categories:conversational",
"language:en",
"license:apache-2.0",
"LoLLMs",
"QnA",
"region:us"
] |
2023-10-26T14:25:07+00:00
|
{"language": ["en"], "license": "apache-2.0", "task_categories": ["conversational"], "tags": ["LoLLMs", "QnA"]}
|
2023-10-27T19:43:37+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-conversational #language-English #license-apache-2.0 #LoLLMs #QnA #region-us
|
# LoLLMs-QNA Dataset
## Dataset Description
The LoLLMs-QNA dataset was created by ParisNeo. The dataset is based on the documentation and knowledge base developed for LoLLMs. It aims to provide a comprehensive collection of questions and corresponding answers related to LoLLMs and its functionalities.
The dataset is structured as a JSON file, with each entry consisting of a question and its corresponding answer. The questions cover various aspects of LoLLMs, including installation, features, functionalities, system requirements, and comparisons with other similar tools. The answers provide detailed information and instructions to assist users in understanding and utilizing LoLLMs effectively.
It is important to note that the dataset also contains some generic thoughts and reflections about AI and its potential uses and threats. These thoughts represent ParisNeo's personal views and should not be considered as a universally accepted truth.
## Dataset Creation Process
The LoLLMs-QNA dataset was created using a two-step process outlined in ParisNeo's white paper titled "From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training." The process involves extracting questions from raw text and then utilizing a vectorized version of the raw data along with an LLM to generate answers.
The raw text used for question extraction includes the documentation and knowledge base developed for LoLLMs, along with ParisNeo's personal insights and expertise in the field of AI. The questions were then manually crafted from this raw text to cover a wide range of topics related to LoLLMs.
To generate the answers, a vectorized version of the raw data was created, along with an LLM model trained on the specific domain of LoLLMs. The LLM model was then used to generate accurate and informative answers to the extracted questions.
## Used Models
This database was built using Database Maker on LoLLMs.
Database Maker implements the algorithm presented in the white paper From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training
To do the LLM tasks required to generate the questions and answers, I used the airoboros-l2-70b-2.2.1 model by jondurbin.
Updates to this database will come as LoLLMs documentation evolve and new functionalities are added constantly. So stay tuned.
## Dataset Format
The LoLLMs-QNA dataset is provided as a JSON file. Each entry in the dataset consists of a dictionary with two key-value pairs:
- "question": The question posed by the user.
- "answer": The corresponding answer to the question.
Example entry:
## Usage and Disclaimer
The LoLLMs-QNA dataset is intended to be used for various tasks, including training AI models, developing chatbots, and assisting users in understanding and utilizing LoLLMs. However, it is important to note that the dataset reflects ParisNeo's personal vision and perspectives about AI and LoLLMs. The answers provided in the dataset should not be considered as universally accepted truths, but rather as ParisNeo's personal insights and instructions.
It is recommended to use the dataset in conjunction with other sources of information and to verify the accuracy and relevance of the answers provided. Users should exercise critical thinking and consider the specific context and requirements of their own applications and use cases.
## Acknowledgments
ParisNeo would like to express gratitude to the open-source community and contributors who have supported the development and improvement of LoLLMs. The dataset is provided as a contribution back to the community and aims to facilitate the understanding and utilization of LoLLMs.
## Special thanks
Special Thanks to jondurbin for his advices and for providing the LLM that was used to build this dataset.
Also special thanks to Tom Jobbins for quantizing the model that was used to build this database.
## Licence
Apache 2.0.
|
[
"# LoLLMs-QNA Dataset",
"## Dataset Description\n\nThe LoLLMs-QNA dataset was created by ParisNeo. The dataset is based on the documentation and knowledge base developed for LoLLMs. It aims to provide a comprehensive collection of questions and corresponding answers related to LoLLMs and its functionalities.\n\nThe dataset is structured as a JSON file, with each entry consisting of a question and its corresponding answer. The questions cover various aspects of LoLLMs, including installation, features, functionalities, system requirements, and comparisons with other similar tools. The answers provide detailed information and instructions to assist users in understanding and utilizing LoLLMs effectively.\n\nIt is important to note that the dataset also contains some generic thoughts and reflections about AI and its potential uses and threats. These thoughts represent ParisNeo's personal views and should not be considered as a universally accepted truth.",
"## Dataset Creation Process\n\nThe LoLLMs-QNA dataset was created using a two-step process outlined in ParisNeo's white paper titled \"From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training.\" The process involves extracting questions from raw text and then utilizing a vectorized version of the raw data along with an LLM to generate answers.\n\nThe raw text used for question extraction includes the documentation and knowledge base developed for LoLLMs, along with ParisNeo's personal insights and expertise in the field of AI. The questions were then manually crafted from this raw text to cover a wide range of topics related to LoLLMs.\n\nTo generate the answers, a vectorized version of the raw data was created, along with an LLM model trained on the specific domain of LoLLMs. The LLM model was then used to generate accurate and informative answers to the extracted questions.",
"## Used Models\n\nThis database was built using Database Maker on LoLLMs.\nDatabase Maker implements the algorithm presented in the white paper From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training\nTo do the LLM tasks required to generate the questions and answers, I used the airoboros-l2-70b-2.2.1 model by jondurbin.\n\nUpdates to this database will come as LoLLMs documentation evolve and new functionalities are added constantly. So stay tuned.",
"## Dataset Format\n\nThe LoLLMs-QNA dataset is provided as a JSON file. Each entry in the dataset consists of a dictionary with two key-value pairs:\n\n- \"question\": The question posed by the user.\n- \"answer\": The corresponding answer to the question.\n\nExample entry:",
"## Usage and Disclaimer\n\nThe LoLLMs-QNA dataset is intended to be used for various tasks, including training AI models, developing chatbots, and assisting users in understanding and utilizing LoLLMs. However, it is important to note that the dataset reflects ParisNeo's personal vision and perspectives about AI and LoLLMs. The answers provided in the dataset should not be considered as universally accepted truths, but rather as ParisNeo's personal insights and instructions.\n\nIt is recommended to use the dataset in conjunction with other sources of information and to verify the accuracy and relevance of the answers provided. Users should exercise critical thinking and consider the specific context and requirements of their own applications and use cases.",
"## Acknowledgments\n\nParisNeo would like to express gratitude to the open-source community and contributors who have supported the development and improvement of LoLLMs. The dataset is provided as a contribution back to the community and aims to facilitate the understanding and utilization of LoLLMs.",
"## Special thanks\n\nSpecial Thanks to jondurbin for his advices and for providing the LLM that was used to build this dataset.\nAlso special thanks to Tom Jobbins for quantizing the model that was used to build this database.",
"## Licence\n\nApache 2.0."
] |
[
"TAGS\n#task_categories-conversational #language-English #license-apache-2.0 #LoLLMs #QnA #region-us \n",
"# LoLLMs-QNA Dataset",
"## Dataset Description\n\nThe LoLLMs-QNA dataset was created by ParisNeo. The dataset is based on the documentation and knowledge base developed for LoLLMs. It aims to provide a comprehensive collection of questions and corresponding answers related to LoLLMs and its functionalities.\n\nThe dataset is structured as a JSON file, with each entry consisting of a question and its corresponding answer. The questions cover various aspects of LoLLMs, including installation, features, functionalities, system requirements, and comparisons with other similar tools. The answers provide detailed information and instructions to assist users in understanding and utilizing LoLLMs effectively.\n\nIt is important to note that the dataset also contains some generic thoughts and reflections about AI and its potential uses and threats. These thoughts represent ParisNeo's personal views and should not be considered as a universally accepted truth.",
"## Dataset Creation Process\n\nThe LoLLMs-QNA dataset was created using a two-step process outlined in ParisNeo's white paper titled \"From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training.\" The process involves extracting questions from raw text and then utilizing a vectorized version of the raw data along with an LLM to generate answers.\n\nThe raw text used for question extraction includes the documentation and knowledge base developed for LoLLMs, along with ParisNeo's personal insights and expertise in the field of AI. The questions were then manually crafted from this raw text to cover a wide range of topics related to LoLLMs.\n\nTo generate the answers, a vectorized version of the raw data was created, along with an LLM model trained on the specific domain of LoLLMs. The LLM model was then used to generate accurate and informative answers to the extracted questions.",
"## Used Models\n\nThis database was built using Database Maker on LoLLMs.\nDatabase Maker implements the algorithm presented in the white paper From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training\nTo do the LLM tasks required to generate the questions and answers, I used the airoboros-l2-70b-2.2.1 model by jondurbin.\n\nUpdates to this database will come as LoLLMs documentation evolve and new functionalities are added constantly. So stay tuned.",
"## Dataset Format\n\nThe LoLLMs-QNA dataset is provided as a JSON file. Each entry in the dataset consists of a dictionary with two key-value pairs:\n\n- \"question\": The question posed by the user.\n- \"answer\": The corresponding answer to the question.\n\nExample entry:",
"## Usage and Disclaimer\n\nThe LoLLMs-QNA dataset is intended to be used for various tasks, including training AI models, developing chatbots, and assisting users in understanding and utilizing LoLLMs. However, it is important to note that the dataset reflects ParisNeo's personal vision and perspectives about AI and LoLLMs. The answers provided in the dataset should not be considered as universally accepted truths, but rather as ParisNeo's personal insights and instructions.\n\nIt is recommended to use the dataset in conjunction with other sources of information and to verify the accuracy and relevance of the answers provided. Users should exercise critical thinking and consider the specific context and requirements of their own applications and use cases.",
"## Acknowledgments\n\nParisNeo would like to express gratitude to the open-source community and contributors who have supported the development and improvement of LoLLMs. The dataset is provided as a contribution back to the community and aims to facilitate the understanding and utilization of LoLLMs.",
"## Special thanks\n\nSpecial Thanks to jondurbin for his advices and for providing the LLM that was used to build this dataset.\nAlso special thanks to Tom Jobbins for quantizing the model that was used to build this database.",
"## Licence\n\nApache 2.0."
] |
[
37,
10,
194,
210,
111,
75,
169,
66,
48,
7
] |
[
"passage: TAGS\n#task_categories-conversational #language-English #license-apache-2.0 #LoLLMs #QnA #region-us \n# LoLLMs-QNA Dataset## Dataset Description\n\nThe LoLLMs-QNA dataset was created by ParisNeo. The dataset is based on the documentation and knowledge base developed for LoLLMs. It aims to provide a comprehensive collection of questions and corresponding answers related to LoLLMs and its functionalities.\n\nThe dataset is structured as a JSON file, with each entry consisting of a question and its corresponding answer. The questions cover various aspects of LoLLMs, including installation, features, functionalities, system requirements, and comparisons with other similar tools. The answers provide detailed information and instructions to assist users in understanding and utilizing LoLLMs effectively.\n\nIt is important to note that the dataset also contains some generic thoughts and reflections about AI and its potential uses and threats. These thoughts represent ParisNeo's personal views and should not be considered as a universally accepted truth.## Dataset Creation Process\n\nThe LoLLMs-QNA dataset was created using a two-step process outlined in ParisNeo's white paper titled \"From Text to Interactive Knowledge: Building Chat-Style Databases for AI Training.\" The process involves extracting questions from raw text and then utilizing a vectorized version of the raw data along with an LLM to generate answers.\n\nThe raw text used for question extraction includes the documentation and knowledge base developed for LoLLMs, along with ParisNeo's personal insights and expertise in the field of AI. The questions were then manually crafted from this raw text to cover a wide range of topics related to LoLLMs.\n\nTo generate the answers, a vectorized version of the raw data was created, along with an LLM model trained on the specific domain of LoLLMs. The LLM model was then used to generate accurate and informative answers to the extracted questions."
] |
08a677f0cc24d9a764dea90ba3a4748ca066ae4a
|
# Dataset Card for "Synthetic_Atest_MMS"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
mekaneeky/Synthetic_Ateso_MMS
|
[
"region:us"
] |
2023-10-26T14:26:54+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "dev", "path": "data/dev-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "eng", "dtype": "string"}, {"name": "lug", "dtype": "string"}, {"name": "ach", "dtype": "string"}, {"name": "teo", "dtype": "string"}, {"name": "lgg", "dtype": "string"}, {"name": "nyn", "dtype": "string"}, {"name": "ID", "dtype": "string"}, {"name": "teo_tts", "sequence": {"sequence": "float32"}}], "splits": [{"name": "train", "num_bytes": 15356273984, "num_examples": 23947}, {"name": "dev", "num_bytes": 311338572, "num_examples": 500}, {"name": "test", "num_bytes": 326319368, "num_examples": 500}], "download_size": 16002788236, "dataset_size": 15993931924}}
|
2023-10-26T14:42:31+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "Synthetic_Atest_MMS"
More Information needed
|
[
"# Dataset Card for \"Synthetic_Atest_MMS\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"Synthetic_Atest_MMS\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"Synthetic_Atest_MMS\"\n\nMore Information needed"
] |
37ca04f5909059fad05aa5d3cf919545c63f815b
|
# Dataset Card for "anno_augmented"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
anlp/anno_augmented
|
[
"region:us"
] |
2023-10-26T14:31:37+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "sentences", "sequence": "string"}, {"name": "ner_tags", "sequence": "string"}], "splits": [{"name": "train", "num_bytes": 1227934, "num_examples": 247}], "download_size": 0, "dataset_size": 1227934}}
|
2023-10-26T16:33:41+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "anno_augmented"
More Information needed
|
[
"# Dataset Card for \"anno_augmented\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"anno_augmented\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"anno_augmented\"\n\nMore Information needed"
] |
0f9753a4c6a2fbe3c94dd17bb99c8f238694cfd1
|
# Dataset Card for "Kan"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
Kishore05/Kan
|
[
"region:us"
] |
2023-10-26T14:57:25+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "validation", "path": "data/validation-*"}]}], "dataset_info": {"features": [{"name": "review", "dtype": "string"}, {"name": "review_length", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 19721.78947368421, "num_examples": 17}, {"name": "validation", "num_bytes": 2320.2105263157896, "num_examples": 2}], "download_size": 25309, "dataset_size": 22042.0}}
|
2023-10-26T14:57:28+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "Kan"
More Information needed
|
[
"# Dataset Card for \"Kan\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"Kan\"\n\nMore Information needed"
] |
[
6,
11
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"Kan\"\n\nMore Information needed"
] |
67b9a87a9f52285a666aea50a0e118b8db8f6daa
|
# Dataset Card for "open-lid-dataset"
## Dataset Description
- **Repository:** [https://github.com/laurieburchell/open-lid-dataset]()
- **Paper:** [An Open Dataset and Model for Language Identification](https://aclanthology.org/2023.acl-short.75/)
- **Point of Contact:** laurie.burchell AT ed.ac.uk
### Dataset Summary
The OpenLID dataset covers 201 languages and is designed for training language identification models. The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). A sample of each language in each source was manually audited to check it was in the attested language (see [the paper](https://aclanthology.org/2023.acl-short.75/)) for full details.
### Supported tasks
This dataset is intended for training high-coverage language identification models (e.g. [OpenLID](https://huggingface.co/laurievb/OpenLID)). It is compatible with the [FLORES-200](https://github.com/facebookresearch/flores/tree/main/flores200) evaluation benchmark.
### Languages
There are 201 languages included in the dataset with varying amounts of data: the largest class (English) contains 7.5 million lines of data, and the smallest (South Azerbaijani) contains 532 lines of data. The mean number of lines per language is 602,812. A full breakdown of lines of data per language is available [on the repo](https://github.com/laurieburchell/open-lid-dataset/blob/main/languages.md).
## Dataset Structure
### Data Instances
Each entry in the dataset consists of a line of data, a language label included script information, and a tag indicating the source.
```json
{
"text": "¿Serás exaltada hasta el cielo?",
"language": "spa_Latn",
"dataset_source": "lti"
}
```
### Data Splits
Only a train split is provided. The dataset is designed to be compatible with the [FLORES-200](https://github.com/facebookresearch/flores/tree/main/flores200) evaluation benchmark.
## Dataset Creation
### Curation Rationale
Recent work has found that existing language identification algorithms perform poorly in practice compared to test performance. The problem is particularly acute for low-resource languages: [Kreutzer et al. (2022)](https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00447/109285/Quality-at-a-Glance-An-Audit-of-Web-Crawled) found a positive Spearman rank correlation between quality of data and size of language for all of the \ac{lid}-filtered multilingual datasets they studied. In addition, for a significant fraction of the language corpora they studied, less than half of the sentences were in the correct language. They point out that such low-quality data not only leads to poor performance in downstream tasks, but that it also contributes to `representation washing', where the community is given a false view of the actual progress of low-resource natural language processing.
There are several open language identification models offering quick classification and high language coverage (e.g. CLD3, No Language Left Behind). However, to the best of our knowledge, none of the commonly-used scalable language identificaiton systems make their training data public.
This dataset aims to address that gap by curating and combining sources of open training data for language identification and by auditing a sample of all languages in each source to check reliability.
### Source Data
The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). We provide a full list at the end of this model card along with the licensing information for each source.
#### Initial Data Collection and Normalisation
Our initial aim was to cover the same languages present in the FLORES-200 Evaluation Benchmark so that we could use this dataset for evaluation. However, during the curation process, we decided to exclude three languages. Firstly, though Akan and Twi are both included as separate languages in FLORES-200, Akan is actually a macrolanguage covering a language continuum which includes Twi. Given the other languages in FLORES-200 are individual languages, we decided to exclude Akan. Secondly, FLORES-200 includes Modern Standard Arabic (MSA) written in Latin script. It is true that Arabic dialects are often written in Latin characters in informal situations (e.g. social media). However, MSA is a form of standardised Arabic which is not usually used in informal situations. Since we could not any find naturally-occurring training data, we excluded MSA from the dataset. Finally, we excluded Minangkabau in Arabic script because it is now rarely written this way, making it difficult to find useful training data.
The first step in our manual audit was to check and standardise language labels, as these are often inconsistent or idiosyncratic. We chose to copy the language codes in FLORES-200 and reassign macrolanguage or ambiguous language codes in the data sources we found to the dominant individual language. Whilst this resulted in more useful data for some languages, for other languages we had to be more conservative. For example, we originally reassigned text labelled as the macrolanguage Malay (msa_Latn) to Standard Malay, but this led to a large drop in performance as the former covers a very diverse set of languages.
Two of the authors then carried out a manual audit of a random sample of all data sources and languages: one a native Bulgarian speaker (able to read Cyrillic and Latin scripts and Chinese characters), and the other a native English speaker (able to read Latin, Arabic and Hebrew scripts). For languages we knew, we checked the language was what we expected. For unfamiliar languages in a script we could read, we compared the sample to the Universal Declaration of Human Rights or failing that, to a sample of text on Wikipedia. We compared features of the text which are common in previous language identification algorithms and could be identified easily by humans: similar diacritics, word lengths, common words, loan words matching the right cultural background, similar suffixes and prefixes, and vowel/consonant patterns. For scripts we could not read, we checked that all lines of the sample matched the script in the Universal Declaration of Human Rights.
We kept preprocessing minimal so that the process was as language agnostic as possible. We used the scripts provided with Moses to remove non-printing characters and detokenise the data where necessary. We then filtered the data so that each line contained at least one character in the expected script (as defined by Perl) to allow for borrowings. Finally, we sampled proportionally to $ p_l^{0.3} $, where $ p_l $ is the fraction of lines in the dataset which are in language $ l $. This aims to ameliorate class skew issues.
## Considerations for Using the Data
### Social Impact of Dataset
This dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.
### Discussion of Biases
Our work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.
In addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a `black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.
## Additional information
The dataset was curated from the sources listed below by Laurie Burchell and Nikolay Bogoychev.
### Licensing Information
License considerations for each source are given below. Open use for non-commercial purposes is covered by all licences.
If you view any part of this dataset as a violation of intellectual property rights, please let us know and we will remove it.
| Source | Description | License |
|---|---|---|
|[Arabic Dialects Dataset](https://www.lancaster.ac.uk/staff/elhaj/corpora.html)| Dataset of Arabic dialects for Gulf, Egyptian, Levantine, and Tunisian Arabic dialects plus MSA|No explicit license; website describes data as "some free and useful Arabic corpora that I have created for researchers working on Arabic Natural Language Processing, Corpus and Computational Linguistics."|
|[BLTR](https://github.com/shashwatup9k/bho-resources)|Monolingual Bhojpuri corpus|[CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)|
|[Global Voices](https://opus.nlpl.eu/GlobalVoices-v2015.php)|A parallel corpus of news stories from the web site Global Voices|The website for [Global Voices](https://globalvoices.org/) is licensed as [Creative Commons Attribution 3.0](https://creativecommons.org/licenses/by/3.0/). There is no explicit additional license accompanying the dataset.|
|[Guaraní Parallel Set](https://github.com/sgongora27/giossa-gongora-guarani-2021)|Parallel Guaraní-Spanish news corpus sourced from Paraguyan websites|No explicit license|
|[HKCanCor](https://github.com/fcbond/hkcancor)|Transcribed conversations in Hong Kong Cantonese|[CC BY 4.0](https://creativecommons.org/licenses/by/4.0/legalcode)|
|[IADD](https://github.com/JihadZa/IADD)|Arabic dialect identification dataset covering 5 regions (Maghrebi, Levantine, Egypt, Iraq, and Gulf) and 9 countries (Algeria, Morocco, Tunisia, Palestine, Jordan, Syria, Lebanon, Egypt and Iraq). It is created from five corpora: [DART](http://qufaculty.qu.edu.qa/telsay), [SHAMI](https://github.com/GU-CLASP/shami-corpus), [TSAC](https://github.com/fbougares/TSAC), [PADIC](https://sourceforge.net/projects/padic/), and [AOC](https://www.cs.jhu.edu/data-archive/AOC-2010/). | Multiple licenses: [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0) (SHAMI); [GNU Lesser General Public License v3.0](https://github.com/fbougares/TSAC/blob/master/LICENSE) (TSAC); [GNU General Public License v3](https://www.gnu.org/licenses/gpl-3.0.en.html) (PADIC). DART and AOC had no explicit license.|
|[Leipzig Corpora Collection](https://wortschatz.uni-leipzig.de/en/download)|A collection of corpora in different languages with an identical format.|The [Terms of Usage](https://wortschatz.uni-leipzig.de/en/usage) states "Permission for use is granted free of charge solely for non-commercial personal and scientific purposes licensed under the [Creative Commons License CC BY-NC](https://creativecommons.org/licenses/by-nc/4.0/)."|
|[LTI](https://www.cs.cmu.edu/~ralf/langid.html)|Training data for language identification|From the README: "With the exception of the contents of the Europarl/, ProjectGutenberg/, and PublicDomain/ directories, all code and text in this corpus are copyrighted. However, they may be redistributed under the terms of various Creative Commons licenses and the GNU GPL. Copying the unmodified archive noncommercially is permitted by all of the licenses. For commercial redistribution or redistribution of modified versions, please consult the individual licenses."|
|[MADAR Shared Task 2019, subtask 1](https://camel.abudhabi.nyu.edu/madar-shared-task-2019/)|Dialectal Arabic in the travel domain|The MADAR Corpus has a custom license, the text of which can be found in this repo.|
|[EM corpus](http://lepage-lab.ips.waseda.ac.jp/en/projects/meiteilon-manipuri-language-resources/)|Parallel Manipuri-English sentences crawled from [The Sangai Express](https://www.thesangaiexpress.com/)|[CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/)|
|[MIZAN](https://github.com/omidkashefi/Mizan)|Parallel Persian-English corpus from literature domain|[CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)|
|[MT560 v1](https://opus.nlpl.eu/MT560.php)|A machine translation dataset for over 500 languages to English. We have filtered out data from OPUS-100, Europarl, Open Subtitles, Paracrawl, Wikimedia, Wikimatrix, Wikititles, and Common Crawl due to issues with the fidelity of the language labels. |[Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0)|
|[NLLB Seed](https://github.com/facebookresearch/flores/blob/main/nllb_seed/README.md)|Around 6000 sentences in 39 languages sampled from Wikipedia, intended to cover languages lacking training data.|[CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)|
|[SETIMES](https://opus.nlpl.eu/SETIMES.php)|A parallel corpus of news articles in the Balkan languages|[CC-BY-SA 3.0](https://creativecommons.org/licenses/by-sa/3.0/)|
|[Tatoeba](https://opus.nlpl.eu/Tatoeba.php)|Collaborative sentence translations|[CC BY 2.0 FR](https://creativecommons.org/licenses/by/2.0/fr/)|
|[Tehran English-Persian parallel corpus (TEP)](https://opus.nlpl.eu/TEP.php)|Parallel Persian-English sentences sourced from subtitles|[GNU General Public License](https://www.gnu.org/licenses/gpl-3.0.html)|
|[Turkic Interlingua (TIL) Corpus](https://github.com/turkic-interlingua/til-mt)|A large-scale parallel corpus combining most of the public datasets for 22 Turkic languages|[CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)|
|[WiLI-2018](https://zenodo.org/record/841984)|Wikipedia language identification benchmark containing 235K paragraphs of 235 languages|[Open Data Commons Open Database License (ODbL) v1.0](https://opendatacommons.org/licenses/odbl/1-0/)|
|[XL-Sum](https://github.com/csebuetnlp/xl-sum)|Summarisation dataset covering 44 languages, sourced from BBC News|[CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)|
### Citation Information
If you use this dataset, please cite all the authors [in the citation file](https://github.com/laurieburchell/open-lid-dataset/blob/main/citations.bib) who compiled the source datasets, plus the OpenLID paper:
```bibtex
@inproceedings{burchell-etal-2023-open,
title = "An Open Dataset and Model for Language Identification",
author = "Burchell, Laurie and
Birch, Alexandra and
Bogoychev, Nikolay and
Heafield, Kenneth",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.75",
doi = "10.18653/v1/2023.acl-short.75",
pages = "865--879",
abstract = "Language identification (LID) is a fundamental step in many natural language processing pipelines. However, current LID systems are far from perfect, particularly on lower-resource languages. We present a LID model which achieves a macro-average F1 score of 0.93 and a false positive rate of 0.033{\%} across 201 languages, outperforming previous work. We achieve this by training on a curated dataset of monolingual data, which we audit manually to ensure reliability. We make both the model and the dataset available to the research community. Finally, we carry out detailed analysis into our model{'}s performance, both in comparison to existing open models and by language class.",
}
```
### Contributions
Thanks to @hac541309 and @davanstrien for adding this dataset.
|
laurievb/open-lid-dataset
|
[
"task_categories:text-classification",
"size_categories:100M<n<1B",
"license:other",
"region:us"
] |
2023-10-26T15:00:52+00:00
|
{"license": "other", "size_categories": ["100M<n<1B"], "task_categories": ["text-classification"], "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "text", "dtype": "string"}, {"name": "language", "dtype": {"class_label": {"names": {"0": "plt_Latn", "1": "sun_Latn", "2": "ukr_Cyrl", "3": "spa_Latn", "4": "por_Latn", "5": "mya_Mymr", "6": "mkd_Cyrl", "7": "war_Latn", "8": "nso_Latn", "9": "wol_Latn", "10": "kam_Latn", "11": "mal_Mlym", "12": "gle_Latn", "13": "ayr_Latn", "14": "rus_Cyrl", "15": "pbt_Arab", "16": "pag_Latn", "17": "twi_Latn", "18": "als_Latn", "19": "lit_Latn", "20": "amh_Ethi", "21": "tur_Latn", "22": "tel_Telu", "23": "vec_Latn", "24": "zsm_Latn", "25": "ckb_Arab", "26": "tgk_Cyrl", "27": "tha_Thai", "28": "hye_Armn", "29": "deu_Latn", "30": "tat_Cyrl", "31": "swh_Latn", "32": "kac_Latn", "33": "tuk_Latn", "34": "lvs_Latn", "35": "tso_Latn", "36": "fao_Latn", "37": "tpi_Latn", "38": "umb_Latn", "39": "mlt_Latn", "40": "cym_Latn", "41": "ben_Beng", "42": "hat_Latn", "43": "ron_Latn", "44": "tir_Ethi", "45": "ewe_Latn", "46": "ind_Latn", "47": "snd_Arab", "48": "nld_Latn", "49": "urd_Arab", "50": "vie_Latn", "51": "mar_Deva", "52": "fra_Latn", "53": "lug_Latn", "54": "pol_Latn", "55": "ban_Latn", "56": "est_Latn", "57": "srp_Cyrl", "58": "kin_Latn", "59": "nno_Latn", "60": "fur_Latn", "61": "kmr_Latn", "62": "bho_Deva", "63": "fin_Latn", "64": "mri_Latn", "65": "ilo_Latn", "66": "fij_Latn", "67": "slk_Latn", "68": "knc_Arab", "69": "guj_Gujr", "70": "kor_Hang", "71": "tum_Latn", "72": "kab_Latn", "73": "afr_Latn", "74": "eng_Latn", "75": "acq_Arab", "76": "som_Latn", "77": "tgl_Latn", "78": "epo_Latn", "79": "bjn_Arab", "80": "mni_Beng", "81": "sot_Latn", "82": "nob_Latn", "83": "kat_Geor", "84": "ory_Orya", "85": "arb_Arab", "86": "heb_Hebr", "87": "ibo_Latn", "88": "asm_Beng", "89": "uzn_Latn", "90": "sna_Latn", "91": "mos_Latn", "92": "fuv_Latn", "93": "hne_Deva", "94": "apc_Arab", "95": "hun_Latn", "96": "ita_Latn", "97": "bem_Latn", "98": "slv_Latn", "99": "ssw_Latn", "100": "szl_Latn", "101": "nya_Latn", "102": "kir_Cyrl", "103": "hrv_Latn", "104": "pap_Latn", "105": "kik_Latn", "106": "knc_Latn", "107": "lmo_Latn", "108": "hau_Latn", "109": "eus_Latn", "110": "ltz_Latn", "111": "grn_Latn", "112": "lus_Latn", "113": "taq_Latn", "114": "scn_Latn", "115": "kmb_Latn", "116": "azj_Latn", "117": "isl_Latn", "118": "swe_Latn", "119": "uig_Arab", "120": "jpn_Jpan", "121": "sag_Latn", "122": "xho_Latn", "123": "ast_Latn", "124": "kan_Knda", "125": "sin_Sinh", "126": "acm_Arab", "127": "tzm_Tfng", "128": "dan_Latn", "129": "zho_Hant", "130": "zho_Hans", "131": "pes_Arab", "132": "fon_Latn", "133": "tam_Taml", "134": "yor_Latn", "135": "run_Latn", "136": "arz_Arab", "137": "awa_Deva", "138": "pan_Guru", "139": "gaz_Latn", "140": "lao_Laoo", "141": "bos_Latn", "142": "ces_Latn", "143": "bam_Latn", "144": "crh_Latn", "145": "ltg_Latn", "146": "bul_Cyrl", "147": "gla_Latn", "148": "ell_Grek", "149": "prs_Arab", "150": "smo_Latn", "151": "ajp_Arab", "152": "tsn_Latn", "153": "bak_Cyrl", "154": "srd_Latn", "155": "ace_Arab", "156": "kas_Arab", "157": "lua_Latn", "158": "taq_Tfng", "159": "jav_Latn", "160": "cat_Latn", "161": "kon_Latn", "162": "hin_Deva", "163": "lin_Latn", "164": "khk_Cyrl", "165": "cjk_Latn", "166": "mag_Deva", "167": "dik_Latn", "168": "bug_Latn", "169": "bjn_Latn", "170": "yue_Hant", "171": "zul_Latn", "172": "npi_Deva", "173": "kas_Deva", "174": "dzo_Tibt", "175": "ary_Arab", "176": "bel_Cyrl", "177": "kbp_Latn", "178": "khm_Khmr", "179": "ace_Latn", "180": "nus_Latn", "181": "ceb_Latn", "182": "mai_Deva", "183": "san_Deva", "184": "dyu_Latn", "185": "quy_Latn", "186": "lim_Latn", "187": "min_Latn", "188": "oci_Latn", "189": "kaz_Cyrl", "190": "luo_Latn", "191": "sat_Olck", "192": "ydd_Hebr", "193": "shn_Mymr", "194": "ars_Arab", "195": "lij_Latn", "196": "aeb_Arab", "197": "bod_Tibt", "198": "glg_Latn", "199": "kea_Latn", "200": "azb_Arab"}}}}, {"name": "dataset_source", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 21749592609, "num_examples": 118296182}], "download_size": 16568412828, "dataset_size": 21749592609}}
|
2023-11-10T10:12:56+00:00
|
[] |
[] |
TAGS
#task_categories-text-classification #size_categories-100M<n<1B #license-other #region-us
|
Dataset Card for "open-lid-dataset"
===================================
Dataset Description
-------------------
* Repository: [URL
* Paper: An Open Dataset and Model for Language Identification
* Point of Contact: laurie.burchell AT URL
### Dataset Summary
The OpenLID dataset covers 201 languages and is designed for training language identification models. The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). A sample of each language in each source was manually audited to check it was in the attested language (see the paper) for full details.
### Supported tasks
This dataset is intended for training high-coverage language identification models (e.g. OpenLID). It is compatible with the FLORES-200 evaluation benchmark.
### Languages
There are 201 languages included in the dataset with varying amounts of data: the largest class (English) contains 7.5 million lines of data, and the smallest (South Azerbaijani) contains 532 lines of data. The mean number of lines per language is 602,812. A full breakdown of lines of data per language is available on the repo.
Dataset Structure
-----------------
### Data Instances
Each entry in the dataset consists of a line of data, a language label included script information, and a tag indicating the source.
### Data Splits
Only a train split is provided. The dataset is designed to be compatible with the FLORES-200 evaluation benchmark.
Dataset Creation
----------------
### Curation Rationale
Recent work has found that existing language identification algorithms perform poorly in practice compared to test performance. The problem is particularly acute for low-resource languages: Kreutzer et al. (2022) found a positive Spearman rank correlation between quality of data and size of language for all of the \ac{lid}-filtered multilingual datasets they studied. In addition, for a significant fraction of the language corpora they studied, less than half of the sentences were in the correct language. They point out that such low-quality data not only leads to poor performance in downstream tasks, but that it also contributes to 'representation washing', where the community is given a false view of the actual progress of low-resource natural language processing.
There are several open language identification models offering quick classification and high language coverage (e.g. CLD3, No Language Left Behind). However, to the best of our knowledge, none of the commonly-used scalable language identificaiton systems make their training data public.
This dataset aims to address that gap by curating and combining sources of open training data for language identification and by auditing a sample of all languages in each source to check reliability.
### Source Data
The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). We provide a full list at the end of this model card along with the licensing information for each source.
#### Initial Data Collection and Normalisation
Our initial aim was to cover the same languages present in the FLORES-200 Evaluation Benchmark so that we could use this dataset for evaluation. However, during the curation process, we decided to exclude three languages. Firstly, though Akan and Twi are both included as separate languages in FLORES-200, Akan is actually a macrolanguage covering a language continuum which includes Twi. Given the other languages in FLORES-200 are individual languages, we decided to exclude Akan. Secondly, FLORES-200 includes Modern Standard Arabic (MSA) written in Latin script. It is true that Arabic dialects are often written in Latin characters in informal situations (e.g. social media). However, MSA is a form of standardised Arabic which is not usually used in informal situations. Since we could not any find naturally-occurring training data, we excluded MSA from the dataset. Finally, we excluded Minangkabau in Arabic script because it is now rarely written this way, making it difficult to find useful training data.
The first step in our manual audit was to check and standardise language labels, as these are often inconsistent or idiosyncratic. We chose to copy the language codes in FLORES-200 and reassign macrolanguage or ambiguous language codes in the data sources we found to the dominant individual language. Whilst this resulted in more useful data for some languages, for other languages we had to be more conservative. For example, we originally reassigned text labelled as the macrolanguage Malay (msa\_Latn) to Standard Malay, but this led to a large drop in performance as the former covers a very diverse set of languages.
Two of the authors then carried out a manual audit of a random sample of all data sources and languages: one a native Bulgarian speaker (able to read Cyrillic and Latin scripts and Chinese characters), and the other a native English speaker (able to read Latin, Arabic and Hebrew scripts). For languages we knew, we checked the language was what we expected. For unfamiliar languages in a script we could read, we compared the sample to the Universal Declaration of Human Rights or failing that, to a sample of text on Wikipedia. We compared features of the text which are common in previous language identification algorithms and could be identified easily by humans: similar diacritics, word lengths, common words, loan words matching the right cultural background, similar suffixes and prefixes, and vowel/consonant patterns. For scripts we could not read, we checked that all lines of the sample matched the script in the Universal Declaration of Human Rights.
We kept preprocessing minimal so that the process was as language agnostic as possible. We used the scripts provided with Moses to remove non-printing characters and detokenise the data where necessary. We then filtered the data so that each line contained at least one character in the expected script (as defined by Perl) to allow for borrowings. Finally, we sampled proportionally to $ p\_l^{0.3} $, where $ p\_l $ is the fraction of lines in the dataset which are in language $ l $. This aims to ameliorate class skew issues.
Considerations for Using the Data
---------------------------------
### Social Impact of Dataset
This dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.
### Discussion of Biases
Our work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.
In addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a 'black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.
Additional information
----------------------
The dataset was curated from the sources listed below by Laurie Burchell and Nikolay Bogoychev.
### Licensing Information
License considerations for each source are given below. Open use for non-commercial purposes is covered by all licences.
If you view any part of this dataset as a violation of intellectual property rights, please let us know and we will remove it.
Source: Arabic Dialects Dataset, Description: Dataset of Arabic dialects for Gulf, Egyptian, Levantine, and Tunisian Arabic dialects plus MSA, License: No explicit license; website describes data as "some free and useful Arabic corpora that I have created for researchers working on Arabic Natural Language Processing, Corpus and Computational Linguistics."
Source: BLTR, Description: Monolingual Bhojpuri corpus, License: CC BY-NC-SA 4.0
Source: Global Voices, Description: A parallel corpus of news stories from the web site Global Voices, License: The website for Global Voices is licensed as Creative Commons Attribution 3.0. There is no explicit additional license accompanying the dataset.
Source: Guaraní Parallel Set, Description: Parallel Guaraní-Spanish news corpus sourced from Paraguyan websites, License: No explicit license
Source: HKCanCor, Description: Transcribed conversations in Hong Kong Cantonese, License: CC BY 4.0
Source: IADD, Description: Arabic dialect identification dataset covering 5 regions (Maghrebi, Levantine, Egypt, Iraq, and Gulf) and 9 countries (Algeria, Morocco, Tunisia, Palestine, Jordan, Syria, Lebanon, Egypt and Iraq). It is created from five corpora: DART, SHAMI, TSAC, PADIC, and AOC., License: Multiple licenses: Apache License 2.0 (SHAMI); GNU Lesser General Public License v3.0 (TSAC); GNU General Public License v3 (PADIC). DART and AOC had no explicit license.
Source: Leipzig Corpora Collection, Description: A collection of corpora in different languages with an identical format., License: The Terms of Usage states "Permission for use is granted free of charge solely for non-commercial personal and scientific purposes licensed under the Creative Commons License CC BY-NC."
Source: LTI, Description: Training data for language identification, License: From the README: "With the exception of the contents of the Europarl/, ProjectGutenberg/, and PublicDomain/ directories, all code and text in this corpus are copyrighted. However, they may be redistributed under the terms of various Creative Commons licenses and the GNU GPL. Copying the unmodified archive noncommercially is permitted by all of the licenses. For commercial redistribution or redistribution of modified versions, please consult the individual licenses."
Source: MADAR Shared Task 2019, subtask 1, Description: Dialectal Arabic in the travel domain, License: The MADAR Corpus has a custom license, the text of which can be found in this repo.
Source: EM corpus, Description: Parallel Manipuri-English sentences crawled from The Sangai Express, License: CC BY-NC 4.0
Source: MIZAN, Description: Parallel Persian-English corpus from literature domain, License: CC BY 4.0
Source: MT560 v1, Description: A machine translation dataset for over 500 languages to English. We have filtered out data from OPUS-100, Europarl, Open Subtitles, Paracrawl, Wikimedia, Wikimatrix, Wikititles, and Common Crawl due to issues with the fidelity of the language labels., License: Apache License 2.0
Source: NLLB Seed, Description: Around 6000 sentences in 39 languages sampled from Wikipedia, intended to cover languages lacking training data., License: CC BY-SA 4.0
Source: SETIMES, Description: A parallel corpus of news articles in the Balkan languages, License: CC-BY-SA 3.0
Source: Tatoeba, Description: Collaborative sentence translations, License: CC BY 2.0 FR
Source: Tehran English-Persian parallel corpus (TEP), Description: Parallel Persian-English sentences sourced from subtitles, License: GNU General Public License
Source: Turkic Interlingua (TIL) Corpus, Description: A large-scale parallel corpus combining most of the public datasets for 22 Turkic languages, License: CC BY-NC-SA 4.0
Source: WiLI-2018, Description: Wikipedia language identification benchmark containing 235K paragraphs of 235 languages, License: Open Data Commons Open Database License (ODbL) v1.0
Source: XL-Sum, Description: Summarisation dataset covering 44 languages, sourced from BBC News, License: CC BY-NC-SA 4.0
If you use this dataset, please cite all the authors in the citation file who compiled the source datasets, plus the OpenLID paper:
### Contributions
Thanks to @hac541309 and @davanstrien for adding this dataset.
|
[
"### Dataset Summary\n\n\nThe OpenLID dataset covers 201 languages and is designed for training language identification models. The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). A sample of each language in each source was manually audited to check it was in the attested language (see the paper) for full details.",
"### Supported tasks\n\n\nThis dataset is intended for training high-coverage language identification models (e.g. OpenLID). It is compatible with the FLORES-200 evaluation benchmark.",
"### Languages\n\n\nThere are 201 languages included in the dataset with varying amounts of data: the largest class (English) contains 7.5 million lines of data, and the smallest (South Azerbaijani) contains 532 lines of data. The mean number of lines per language is 602,812. A full breakdown of lines of data per language is available on the repo.\n\n\nDataset Structure\n-----------------",
"### Data Instances\n\n\nEach entry in the dataset consists of a line of data, a language label included script information, and a tag indicating the source.",
"### Data Splits\n\n\nOnly a train split is provided. The dataset is designed to be compatible with the FLORES-200 evaluation benchmark.\n\n\nDataset Creation\n----------------",
"### Curation Rationale\n\n\nRecent work has found that existing language identification algorithms perform poorly in practice compared to test performance. The problem is particularly acute for low-resource languages: Kreutzer et al. (2022) found a positive Spearman rank correlation between quality of data and size of language for all of the \\ac{lid}-filtered multilingual datasets they studied. In addition, for a significant fraction of the language corpora they studied, less than half of the sentences were in the correct language. They point out that such low-quality data not only leads to poor performance in downstream tasks, but that it also contributes to 'representation washing', where the community is given a false view of the actual progress of low-resource natural language processing.\n\n\nThere are several open language identification models offering quick classification and high language coverage (e.g. CLD3, No Language Left Behind). However, to the best of our knowledge, none of the commonly-used scalable language identificaiton systems make their training data public.\n\n\nThis dataset aims to address that gap by curating and combining sources of open training data for language identification and by auditing a sample of all languages in each source to check reliability.",
"### Source Data\n\n\nThe majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). We provide a full list at the end of this model card along with the licensing information for each source.",
"#### Initial Data Collection and Normalisation\n\n\nOur initial aim was to cover the same languages present in the FLORES-200 Evaluation Benchmark so that we could use this dataset for evaluation. However, during the curation process, we decided to exclude three languages. Firstly, though Akan and Twi are both included as separate languages in FLORES-200, Akan is actually a macrolanguage covering a language continuum which includes Twi. Given the other languages in FLORES-200 are individual languages, we decided to exclude Akan. Secondly, FLORES-200 includes Modern Standard Arabic (MSA) written in Latin script. It is true that Arabic dialects are often written in Latin characters in informal situations (e.g. social media). However, MSA is a form of standardised Arabic which is not usually used in informal situations. Since we could not any find naturally-occurring training data, we excluded MSA from the dataset. Finally, we excluded Minangkabau in Arabic script because it is now rarely written this way, making it difficult to find useful training data.\n\n\nThe first step in our manual audit was to check and standardise language labels, as these are often inconsistent or idiosyncratic. We chose to copy the language codes in FLORES-200 and reassign macrolanguage or ambiguous language codes in the data sources we found to the dominant individual language. Whilst this resulted in more useful data for some languages, for other languages we had to be more conservative. For example, we originally reassigned text labelled as the macrolanguage Malay (msa\\_Latn) to Standard Malay, but this led to a large drop in performance as the former covers a very diverse set of languages.\n\n\nTwo of the authors then carried out a manual audit of a random sample of all data sources and languages: one a native Bulgarian speaker (able to read Cyrillic and Latin scripts and Chinese characters), and the other a native English speaker (able to read Latin, Arabic and Hebrew scripts). For languages we knew, we checked the language was what we expected. For unfamiliar languages in a script we could read, we compared the sample to the Universal Declaration of Human Rights or failing that, to a sample of text on Wikipedia. We compared features of the text which are common in previous language identification algorithms and could be identified easily by humans: similar diacritics, word lengths, common words, loan words matching the right cultural background, similar suffixes and prefixes, and vowel/consonant patterns. For scripts we could not read, we checked that all lines of the sample matched the script in the Universal Declaration of Human Rights.\n\n\nWe kept preprocessing minimal so that the process was as language agnostic as possible. We used the scripts provided with Moses to remove non-printing characters and detokenise the data where necessary. We then filtered the data so that each line contained at least one character in the expected script (as defined by Perl) to allow for borrowings. Finally, we sampled proportionally to $ p\\_l^{0.3} $, where $ p\\_l $ is the fraction of lines in the dataset which are in language $ l $. This aims to ameliorate class skew issues.\n\n\nConsiderations for Using the Data\n---------------------------------",
"### Social Impact of Dataset\n\n\nThis dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.",
"### Discussion of Biases\n\n\nOur work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.\n\n\nIn addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a 'black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.\n\n\nAdditional information\n----------------------\n\n\nThe dataset was curated from the sources listed below by Laurie Burchell and Nikolay Bogoychev.",
"### Licensing Information\n\n\nLicense considerations for each source are given below. Open use for non-commercial purposes is covered by all licences.\n\n\nIf you view any part of this dataset as a violation of intellectual property rights, please let us know and we will remove it.\n\n\nSource: Arabic Dialects Dataset, Description: Dataset of Arabic dialects for Gulf, Egyptian, Levantine, and Tunisian Arabic dialects plus MSA, License: No explicit license; website describes data as \"some free and useful Arabic corpora that I have created for researchers working on Arabic Natural Language Processing, Corpus and Computational Linguistics.\"\nSource: BLTR, Description: Monolingual Bhojpuri corpus, License: CC BY-NC-SA 4.0\nSource: Global Voices, Description: A parallel corpus of news stories from the web site Global Voices, License: The website for Global Voices is licensed as Creative Commons Attribution 3.0. There is no explicit additional license accompanying the dataset.\nSource: Guaraní Parallel Set, Description: Parallel Guaraní-Spanish news corpus sourced from Paraguyan websites, License: No explicit license\nSource: HKCanCor, Description: Transcribed conversations in Hong Kong Cantonese, License: CC BY 4.0\nSource: IADD, Description: Arabic dialect identification dataset covering 5 regions (Maghrebi, Levantine, Egypt, Iraq, and Gulf) and 9 countries (Algeria, Morocco, Tunisia, Palestine, Jordan, Syria, Lebanon, Egypt and Iraq). It is created from five corpora: DART, SHAMI, TSAC, PADIC, and AOC., License: Multiple licenses: Apache License 2.0 (SHAMI); GNU Lesser General Public License v3.0 (TSAC); GNU General Public License v3 (PADIC). DART and AOC had no explicit license.\nSource: Leipzig Corpora Collection, Description: A collection of corpora in different languages with an identical format., License: The Terms of Usage states \"Permission for use is granted free of charge solely for non-commercial personal and scientific purposes licensed under the Creative Commons License CC BY-NC.\"\nSource: LTI, Description: Training data for language identification, License: From the README: \"With the exception of the contents of the Europarl/, ProjectGutenberg/, and PublicDomain/ directories, all code and text in this corpus are copyrighted. However, they may be redistributed under the terms of various Creative Commons licenses and the GNU GPL. Copying the unmodified archive noncommercially is permitted by all of the licenses. For commercial redistribution or redistribution of modified versions, please consult the individual licenses.\"\nSource: MADAR Shared Task 2019, subtask 1, Description: Dialectal Arabic in the travel domain, License: The MADAR Corpus has a custom license, the text of which can be found in this repo.\nSource: EM corpus, Description: Parallel Manipuri-English sentences crawled from The Sangai Express, License: CC BY-NC 4.0\nSource: MIZAN, Description: Parallel Persian-English corpus from literature domain, License: CC BY 4.0\nSource: MT560 v1, Description: A machine translation dataset for over 500 languages to English. We have filtered out data from OPUS-100, Europarl, Open Subtitles, Paracrawl, Wikimedia, Wikimatrix, Wikititles, and Common Crawl due to issues with the fidelity of the language labels., License: Apache License 2.0\nSource: NLLB Seed, Description: Around 6000 sentences in 39 languages sampled from Wikipedia, intended to cover languages lacking training data., License: CC BY-SA 4.0\nSource: SETIMES, Description: A parallel corpus of news articles in the Balkan languages, License: CC-BY-SA 3.0\nSource: Tatoeba, Description: Collaborative sentence translations, License: CC BY 2.0 FR\nSource: Tehran English-Persian parallel corpus (TEP), Description: Parallel Persian-English sentences sourced from subtitles, License: GNU General Public License\nSource: Turkic Interlingua (TIL) Corpus, Description: A large-scale parallel corpus combining most of the public datasets for 22 Turkic languages, License: CC BY-NC-SA 4.0\nSource: WiLI-2018, Description: Wikipedia language identification benchmark containing 235K paragraphs of 235 languages, License: Open Data Commons Open Database License (ODbL) v1.0\nSource: XL-Sum, Description: Summarisation dataset covering 44 languages, sourced from BBC News, License: CC BY-NC-SA 4.0\n\n\nIf you use this dataset, please cite all the authors in the citation file who compiled the source datasets, plus the OpenLID paper:",
"### Contributions\n\n\nThanks to @hac541309 and @davanstrien for adding this dataset."
] |
[
"TAGS\n#task_categories-text-classification #size_categories-100M<n<1B #license-other #region-us \n",
"### Dataset Summary\n\n\nThe OpenLID dataset covers 201 languages and is designed for training language identification models. The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). A sample of each language in each source was manually audited to check it was in the attested language (see the paper) for full details.",
"### Supported tasks\n\n\nThis dataset is intended for training high-coverage language identification models (e.g. OpenLID). It is compatible with the FLORES-200 evaluation benchmark.",
"### Languages\n\n\nThere are 201 languages included in the dataset with varying amounts of data: the largest class (English) contains 7.5 million lines of data, and the smallest (South Azerbaijani) contains 532 lines of data. The mean number of lines per language is 602,812. A full breakdown of lines of data per language is available on the repo.\n\n\nDataset Structure\n-----------------",
"### Data Instances\n\n\nEach entry in the dataset consists of a line of data, a language label included script information, and a tag indicating the source.",
"### Data Splits\n\n\nOnly a train split is provided. The dataset is designed to be compatible with the FLORES-200 evaluation benchmark.\n\n\nDataset Creation\n----------------",
"### Curation Rationale\n\n\nRecent work has found that existing language identification algorithms perform poorly in practice compared to test performance. The problem is particularly acute for low-resource languages: Kreutzer et al. (2022) found a positive Spearman rank correlation between quality of data and size of language for all of the \\ac{lid}-filtered multilingual datasets they studied. In addition, for a significant fraction of the language corpora they studied, less than half of the sentences were in the correct language. They point out that such low-quality data not only leads to poor performance in downstream tasks, but that it also contributes to 'representation washing', where the community is given a false view of the actual progress of low-resource natural language processing.\n\n\nThere are several open language identification models offering quick classification and high language coverage (e.g. CLD3, No Language Left Behind). However, to the best of our knowledge, none of the commonly-used scalable language identificaiton systems make their training data public.\n\n\nThis dataset aims to address that gap by curating and combining sources of open training data for language identification and by auditing a sample of all languages in each source to check reliability.",
"### Source Data\n\n\nThe majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). We provide a full list at the end of this model card along with the licensing information for each source.",
"#### Initial Data Collection and Normalisation\n\n\nOur initial aim was to cover the same languages present in the FLORES-200 Evaluation Benchmark so that we could use this dataset for evaluation. However, during the curation process, we decided to exclude three languages. Firstly, though Akan and Twi are both included as separate languages in FLORES-200, Akan is actually a macrolanguage covering a language continuum which includes Twi. Given the other languages in FLORES-200 are individual languages, we decided to exclude Akan. Secondly, FLORES-200 includes Modern Standard Arabic (MSA) written in Latin script. It is true that Arabic dialects are often written in Latin characters in informal situations (e.g. social media). However, MSA is a form of standardised Arabic which is not usually used in informal situations. Since we could not any find naturally-occurring training data, we excluded MSA from the dataset. Finally, we excluded Minangkabau in Arabic script because it is now rarely written this way, making it difficult to find useful training data.\n\n\nThe first step in our manual audit was to check and standardise language labels, as these are often inconsistent or idiosyncratic. We chose to copy the language codes in FLORES-200 and reassign macrolanguage or ambiguous language codes in the data sources we found to the dominant individual language. Whilst this resulted in more useful data for some languages, for other languages we had to be more conservative. For example, we originally reassigned text labelled as the macrolanguage Malay (msa\\_Latn) to Standard Malay, but this led to a large drop in performance as the former covers a very diverse set of languages.\n\n\nTwo of the authors then carried out a manual audit of a random sample of all data sources and languages: one a native Bulgarian speaker (able to read Cyrillic and Latin scripts and Chinese characters), and the other a native English speaker (able to read Latin, Arabic and Hebrew scripts). For languages we knew, we checked the language was what we expected. For unfamiliar languages in a script we could read, we compared the sample to the Universal Declaration of Human Rights or failing that, to a sample of text on Wikipedia. We compared features of the text which are common in previous language identification algorithms and could be identified easily by humans: similar diacritics, word lengths, common words, loan words matching the right cultural background, similar suffixes and prefixes, and vowel/consonant patterns. For scripts we could not read, we checked that all lines of the sample matched the script in the Universal Declaration of Human Rights.\n\n\nWe kept preprocessing minimal so that the process was as language agnostic as possible. We used the scripts provided with Moses to remove non-printing characters and detokenise the data where necessary. We then filtered the data so that each line contained at least one character in the expected script (as defined by Perl) to allow for borrowings. Finally, we sampled proportionally to $ p\\_l^{0.3} $, where $ p\\_l $ is the fraction of lines in the dataset which are in language $ l $. This aims to ameliorate class skew issues.\n\n\nConsiderations for Using the Data\n---------------------------------",
"### Social Impact of Dataset\n\n\nThis dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.",
"### Discussion of Biases\n\n\nOur work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.\n\n\nIn addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a 'black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.\n\n\nAdditional information\n----------------------\n\n\nThe dataset was curated from the sources listed below by Laurie Burchell and Nikolay Bogoychev.",
"### Licensing Information\n\n\nLicense considerations for each source are given below. Open use for non-commercial purposes is covered by all licences.\n\n\nIf you view any part of this dataset as a violation of intellectual property rights, please let us know and we will remove it.\n\n\nSource: Arabic Dialects Dataset, Description: Dataset of Arabic dialects for Gulf, Egyptian, Levantine, and Tunisian Arabic dialects plus MSA, License: No explicit license; website describes data as \"some free and useful Arabic corpora that I have created for researchers working on Arabic Natural Language Processing, Corpus and Computational Linguistics.\"\nSource: BLTR, Description: Monolingual Bhojpuri corpus, License: CC BY-NC-SA 4.0\nSource: Global Voices, Description: A parallel corpus of news stories from the web site Global Voices, License: The website for Global Voices is licensed as Creative Commons Attribution 3.0. There is no explicit additional license accompanying the dataset.\nSource: Guaraní Parallel Set, Description: Parallel Guaraní-Spanish news corpus sourced from Paraguyan websites, License: No explicit license\nSource: HKCanCor, Description: Transcribed conversations in Hong Kong Cantonese, License: CC BY 4.0\nSource: IADD, Description: Arabic dialect identification dataset covering 5 regions (Maghrebi, Levantine, Egypt, Iraq, and Gulf) and 9 countries (Algeria, Morocco, Tunisia, Palestine, Jordan, Syria, Lebanon, Egypt and Iraq). It is created from five corpora: DART, SHAMI, TSAC, PADIC, and AOC., License: Multiple licenses: Apache License 2.0 (SHAMI); GNU Lesser General Public License v3.0 (TSAC); GNU General Public License v3 (PADIC). DART and AOC had no explicit license.\nSource: Leipzig Corpora Collection, Description: A collection of corpora in different languages with an identical format., License: The Terms of Usage states \"Permission for use is granted free of charge solely for non-commercial personal and scientific purposes licensed under the Creative Commons License CC BY-NC.\"\nSource: LTI, Description: Training data for language identification, License: From the README: \"With the exception of the contents of the Europarl/, ProjectGutenberg/, and PublicDomain/ directories, all code and text in this corpus are copyrighted. However, they may be redistributed under the terms of various Creative Commons licenses and the GNU GPL. Copying the unmodified archive noncommercially is permitted by all of the licenses. For commercial redistribution or redistribution of modified versions, please consult the individual licenses.\"\nSource: MADAR Shared Task 2019, subtask 1, Description: Dialectal Arabic in the travel domain, License: The MADAR Corpus has a custom license, the text of which can be found in this repo.\nSource: EM corpus, Description: Parallel Manipuri-English sentences crawled from The Sangai Express, License: CC BY-NC 4.0\nSource: MIZAN, Description: Parallel Persian-English corpus from literature domain, License: CC BY 4.0\nSource: MT560 v1, Description: A machine translation dataset for over 500 languages to English. We have filtered out data from OPUS-100, Europarl, Open Subtitles, Paracrawl, Wikimedia, Wikimatrix, Wikititles, and Common Crawl due to issues with the fidelity of the language labels., License: Apache License 2.0\nSource: NLLB Seed, Description: Around 6000 sentences in 39 languages sampled from Wikipedia, intended to cover languages lacking training data., License: CC BY-SA 4.0\nSource: SETIMES, Description: A parallel corpus of news articles in the Balkan languages, License: CC-BY-SA 3.0\nSource: Tatoeba, Description: Collaborative sentence translations, License: CC BY 2.0 FR\nSource: Tehran English-Persian parallel corpus (TEP), Description: Parallel Persian-English sentences sourced from subtitles, License: GNU General Public License\nSource: Turkic Interlingua (TIL) Corpus, Description: A large-scale parallel corpus combining most of the public datasets for 22 Turkic languages, License: CC BY-NC-SA 4.0\nSource: WiLI-2018, Description: Wikipedia language identification benchmark containing 235K paragraphs of 235 languages, License: Open Data Commons Open Database License (ODbL) v1.0\nSource: XL-Sum, Description: Summarisation dataset covering 44 languages, sourced from BBC News, License: CC BY-NC-SA 4.0\n\n\nIf you use this dataset, please cite all the authors in the citation file who compiled the source datasets, plus the OpenLID paper:",
"### Contributions\n\n\nThanks to @hac541309 and @davanstrien for adding this dataset."
] |
[
34,
104,
41,
88,
35,
34,
281,
72,
744,
56,
208,
1051,
25
] |
[
"passage: TAGS\n#task_categories-text-classification #size_categories-100M<n<1B #license-other #region-us \n### Dataset Summary\n\n\nThe OpenLID dataset covers 201 languages and is designed for training language identification models. The majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). A sample of each language in each source was manually audited to check it was in the attested language (see the paper) for full details.### Supported tasks\n\n\nThis dataset is intended for training high-coverage language identification models (e.g. OpenLID). It is compatible with the FLORES-200 evaluation benchmark.### Languages\n\n\nThere are 201 languages included in the dataset with varying amounts of data: the largest class (English) contains 7.5 million lines of data, and the smallest (South Azerbaijani) contains 532 lines of data. The mean number of lines per language is 602,812. A full breakdown of lines of data per language is available on the repo.\n\n\nDataset Structure\n-----------------### Data Instances\n\n\nEach entry in the dataset consists of a line of data, a language label included script information, and a tag indicating the source.### Data Splits\n\n\nOnly a train split is provided. The dataset is designed to be compatible with the FLORES-200 evaluation benchmark.\n\n\nDataset Creation\n----------------",
"passage: ### Curation Rationale\n\n\nRecent work has found that existing language identification algorithms perform poorly in practice compared to test performance. The problem is particularly acute for low-resource languages: Kreutzer et al. (2022) found a positive Spearman rank correlation between quality of data and size of language for all of the \\ac{lid}-filtered multilingual datasets they studied. In addition, for a significant fraction of the language corpora they studied, less than half of the sentences were in the correct language. They point out that such low-quality data not only leads to poor performance in downstream tasks, but that it also contributes to 'representation washing', where the community is given a false view of the actual progress of low-resource natural language processing.\n\n\nThere are several open language identification models offering quick classification and high language coverage (e.g. CLD3, No Language Left Behind). However, to the best of our knowledge, none of the commonly-used scalable language identificaiton systems make their training data public.\n\n\nThis dataset aims to address that gap by curating and combining sources of open training data for language identification and by auditing a sample of all languages in each source to check reliability.### Source Data\n\n\nThe majority of the source datasets were derived from news sites, Wikipedia, or religious text, though some come from other domains (e.g. transcribed conversations, literature, or social media). We provide a full list at the end of this model card along with the licensing information for each source.",
"passage: #### Initial Data Collection and Normalisation\n\n\nOur initial aim was to cover the same languages present in the FLORES-200 Evaluation Benchmark so that we could use this dataset for evaluation. However, during the curation process, we decided to exclude three languages. Firstly, though Akan and Twi are both included as separate languages in FLORES-200, Akan is actually a macrolanguage covering a language continuum which includes Twi. Given the other languages in FLORES-200 are individual languages, we decided to exclude Akan. Secondly, FLORES-200 includes Modern Standard Arabic (MSA) written in Latin script. It is true that Arabic dialects are often written in Latin characters in informal situations (e.g. social media). However, MSA is a form of standardised Arabic which is not usually used in informal situations. Since we could not any find naturally-occurring training data, we excluded MSA from the dataset. Finally, we excluded Minangkabau in Arabic script because it is now rarely written this way, making it difficult to find useful training data.\n\n\nThe first step in our manual audit was to check and standardise language labels, as these are often inconsistent or idiosyncratic. We chose to copy the language codes in FLORES-200 and reassign macrolanguage or ambiguous language codes in the data sources we found to the dominant individual language. Whilst this resulted in more useful data for some languages, for other languages we had to be more conservative. For example, we originally reassigned text labelled as the macrolanguage Malay (msa\\_Latn) to Standard Malay, but this led to a large drop in performance as the former covers a very diverse set of languages.\n\n\nTwo of the authors then carried out a manual audit of a random sample of all data sources and languages: one a native Bulgarian speaker (able to read Cyrillic and Latin scripts and Chinese characters), and the other a native English speaker (able to read Latin, Arabic and Hebrew scripts). For languages we knew, we checked the language was what we expected. For unfamiliar languages in a script we could read, we compared the sample to the Universal Declaration of Human Rights or failing that, to a sample of text on Wikipedia. We compared features of the text which are common in previous language identification algorithms and could be identified easily by humans: similar diacritics, word lengths, common words, loan words matching the right cultural background, similar suffixes and prefixes, and vowel/consonant patterns. For scripts we could not read, we checked that all lines of the sample matched the script in the Universal Declaration of Human Rights.\n\n\nWe kept preprocessing minimal so that the process was as language agnostic as possible. We used the scripts provided with Moses to remove non-printing characters and detokenise the data where necessary. We then filtered the data so that each line contained at least one character in the expected script (as defined by Perl) to allow for borrowings. Finally, we sampled proportionally to $ p\\_l^{0.3} $, where $ p\\_l $ is the fraction of lines in the dataset which are in language $ l $. This aims to ameliorate class skew issues.\n\n\nConsiderations for Using the Data\n---------------------------------### Social Impact of Dataset\n\n\nThis dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.### Discussion of Biases\n\n\nOur work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.\n\n\nIn addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a 'black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.\n\n\nAdditional information\n----------------------\n\n\nThe dataset was curated from the sources listed below by Laurie Burchell and Nikolay Bogoychev."
] |
59d3d3ee5c513cbee2cac68fe2064a3fab4942ea
|
Contains the following train split from datasets in [helm](https://github.com/stanford-crfm/helm):
- big bench
- mmlu
- TruthfulQA
- cnn/dm
- gsm
- bbq
- boolq
- NarrativeQA
- QuAC
- math
- bAbI
Each prompt has <= 5 in-context samples along with a sample, all of which from the train set of the respective datasets.
|
royson/train_splits_helm
|
[
"license:apache-2.0",
"region:us"
] |
2023-10-26T15:17:46+00:00
|
{"license": "apache-2.0"}
|
2023-10-26T15:27:29+00:00
|
[] |
[] |
TAGS
#license-apache-2.0 #region-us
|
Contains the following train split from datasets in helm:
- big bench
- mmlu
- TruthfulQA
- cnn/dm
- gsm
- bbq
- boolq
- NarrativeQA
- QuAC
- math
- bAbI
Each prompt has <= 5 in-context samples along with a sample, all of which from the train set of the respective datasets.
|
[] |
[
"TAGS\n#license-apache-2.0 #region-us \n"
] |
[
14
] |
[
"passage: TAGS\n#license-apache-2.0 #region-us \n"
] |
54455c052bf6b77d5c98b8f286e045faf9edfb44
|
# Dataset Card for "t2i_topic_comparision_db_v2"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
toilaluan/t2i_topic_comparision_db_v2
|
[
"region:us"
] |
2023-10-26T15:21:51+00:00
|
{"dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "topic", "dtype": "string"}, {"name": "prompt", "dtype": "string"}, {"name": "request_id", "dtype": "int64"}, {"name": "model_type", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 335176439.2, "num_examples": 7200}], "download_size": 653813254, "dataset_size": 335176439.2}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T23:11:11+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "t2i_topic_comparision_db_v2"
More Information needed
|
[
"# Dataset Card for \"t2i_topic_comparision_db_v2\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"t2i_topic_comparision_db_v2\"\n\nMore Information needed"
] |
[
6,
24
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"t2i_topic_comparision_db_v2\"\n\nMore Information needed"
] |
b3a49f72f03cdc8fc920c8bcc2ba991e54987c3b
|
# Dataset Card for "Synthetic_Ateso_VITS_22.5k"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
mekaneeky/Synthetic_Ateso_VITS_22.5k
|
[
"region:us"
] |
2023-10-26T15:44:32+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "dev", "path": "data/dev-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "eng", "dtype": "string"}, {"name": "lug", "dtype": "string"}, {"name": "ach", "dtype": "string"}, {"name": "teo", "dtype": "string"}, {"name": "lgg", "dtype": "string"}, {"name": "nyn", "dtype": "string"}, {"name": "ID", "dtype": "string"}, {"name": "teo_tts", "sequence": {"sequence": "float32"}}], "splits": [{"name": "train", "num_bytes": 12491742528, "num_examples": 23947}, {"name": "dev", "num_bytes": 260929100, "num_examples": 500}, {"name": "test", "num_bytes": 264178952, "num_examples": 500}], "download_size": 13028184575, "dataset_size": 13016850580}}
|
2023-10-26T15:53:16+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "Synthetic_Ateso_VITS_22.5k"
More Information needed
|
[
"# Dataset Card for \"Synthetic_Ateso_VITS_22.5k\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"Synthetic_Ateso_VITS_22.5k\"\n\nMore Information needed"
] |
[
6,
25
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"Synthetic_Ateso_VITS_22.5k\"\n\nMore Information needed"
] |
069ff7d78205188fe39bc74204295f3b58fa1828
|
wisesight_sentiment_prompt is the instruct fellow dataset for sentiment Thai text by prompt. It can use fine-tuning model.
- inputs: Prompt
- targets: Text targets that AI should answer.
**Template**
```
Inputs: จำแนกประโยคต่อไปนี้เป็นคำถามหรือข้อความเชิงบวก/เป็นกลาง/เชิงลบ:\n{text}
targets: ประโยคที่กำหนดสามารถจำแนกข้อความได้เป็นข้อความ{category}
```
category
- คำถาม: question
- เชิงบวก: positive
- เป็นกลาง: neutral
- เชิงลบ: negative
Notebook that used create this dataset: [https://github.com/PyThaiNLP/support-aya-datasets/blob/main/sentiment-analysis/wisesight_sentiment.ipynb](https://github.com/PyThaiNLP/support-aya-datasets/blob/main/sentiment-analysis/wisesight_sentiment.ipynb)
Wisesight Sentiment Corpus: Social media messages in Thai language with sentiment category (positive, neutral, negative, question)
* Released to public domain under Creative Commons Zero v1.0 Universal license.
* Size: 26,737 messages
* Language: Central Thai
* Style: Informal and conversational. With some news headlines and advertisement.
* Time period: Around 2016 to early 2019. With small amount from other period.
* Domains: Mixed. Majority are consumer products and services (restaurants, cosmetics, drinks, car, hotels), with some current affairs.
See more: [wisesight_sentiment](https://huggingface.co/datasets/wisesight_sentiment).
PyThaiNLP
|
pythainlp/wisesight_sentiment_prompt
|
[
"task_categories:text-generation",
"task_categories:text2text-generation",
"size_categories:10K<n<100K",
"language:th",
"license:cc0-1.0",
"instruct-fellow",
"region:us"
] |
2023-10-26T15:50:41+00:00
|
{"language": ["th"], "license": "cc0-1.0", "size_categories": ["10K<n<100K"], "task_categories": ["text-generation", "text2text-generation"], "pretty_name": "i", "dataset_info": {"features": [{"name": "inputs", "dtype": "string"}, {"name": "targets", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 10132750, "num_examples": 16194}, {"name": "validation", "num_bytes": 1118295, "num_examples": 1777}, {"name": "test", "num_bytes": 1240521, "num_examples": 1965}], "download_size": 3093175, "dataset_size": 12491566}, "tags": ["instruct-fellow"], "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "validation", "path": "data/validation-*"}, {"split": "test", "path": "data/test-*"}]}]}
|
2023-12-04T14:51:16+00:00
|
[] |
[
"th"
] |
TAGS
#task_categories-text-generation #task_categories-text2text-generation #size_categories-10K<n<100K #language-Thai #license-cc0-1.0 #instruct-fellow #region-us
|
wisesight_sentiment_prompt is the instruct fellow dataset for sentiment Thai text by prompt. It can use fine-tuning model.
- inputs: Prompt
- targets: Text targets that AI should answer.
Template
category
- คำถาม: question
- เชิงบวก: positive
- เป็นกลาง: neutral
- เชิงลบ: negative
Notebook that used create this dataset: URL
Wisesight Sentiment Corpus: Social media messages in Thai language with sentiment category (positive, neutral, negative, question)
* Released to public domain under Creative Commons Zero v1.0 Universal license.
* Size: 26,737 messages
* Language: Central Thai
* Style: Informal and conversational. With some news headlines and advertisement.
* Time period: Around 2016 to early 2019. With small amount from other period.
* Domains: Mixed. Majority are consumer products and services (restaurants, cosmetics, drinks, car, hotels), with some current affairs.
See more: wisesight_sentiment.
PyThaiNLP
|
[] |
[
"TAGS\n#task_categories-text-generation #task_categories-text2text-generation #size_categories-10K<n<100K #language-Thai #license-cc0-1.0 #instruct-fellow #region-us \n"
] |
[
61
] |
[
"passage: TAGS\n#task_categories-text-generation #task_categories-text2text-generation #size_categories-10K<n<100K #language-Thai #license-cc0-1.0 #instruct-fellow #region-us \n"
] |
5e657603597e4034d7e7bc7b9c11505414d467da
|
# Dataset Card for "code_searchnet_reduced"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
vishnusr/code_searchnet_reduced
|
[
"region:us"
] |
2023-10-26T15:57:00+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "Unnamed: 0.1", "dtype": "int64"}, {"name": "Unnamed: 0", "dtype": "int64"}, {"name": "code", "dtype": "string"}, {"name": "docstring", "dtype": "string"}, {"name": "prompt", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 992068, "num_examples": 500}], "download_size": 440777, "dataset_size": 992068}}
|
2023-10-26T15:57:30+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "code_searchnet_reduced"
More Information needed
|
[
"# Dataset Card for \"code_searchnet_reduced\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"code_searchnet_reduced\"\n\nMore Information needed"
] |
[
6,
18
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"code_searchnet_reduced\"\n\nMore Information needed"
] |
f84c839dbd67e05bacc4bfb394ababc82cb0fccd
|
# Dataset Card for "important"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
aoome123/haha
|
[
"region:us"
] |
2023-10-26T15:59:29+00:00
|
{"dataset_info": {"config_name": "aoome123/use", "features": [{"name": "input_features", "sequence": {"sequence": "float32"}}, {"name": "labels", "sequence": "int64"}], "splits": [{"name": "train", "num_bytes": 5462003720, "num_examples": 5687}, {"name": "test", "num_bytes": 682871112, "num_examples": 711}, {"name": "valid", "num_bytes": 682869512, "num_examples": 711}], "download_size": 902052782, "dataset_size": 6827744344}, "configs": [{"config_name": "aoome123/use", "data_files": [{"split": "train", "path": "aoome123/use/train-*"}, {"split": "test", "path": "aoome123/use/test-*"}, {"split": "valid", "path": "aoome123/use/valid-*"}]}]}
|
2023-10-26T16:00:41+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "important"
More Information needed
|
[
"# Dataset Card for \"important\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"important\"\n\nMore Information needed"
] |
[
6,
12
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"important\"\n\nMore Information needed"
] |
32c6bedb354c8a8354df2c5d554edc135e831bea
|
# Dataset Card for "code_searchnet_reduced_train"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
vishnusr/code_searchnet_reduced_train
|
[
"region:us"
] |
2023-10-26T16:08:26+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "Unnamed: 0.1", "dtype": "int64"}, {"name": "Unnamed: 0", "dtype": "int64"}, {"name": "code", "dtype": "string"}, {"name": "docstring", "dtype": "string"}, {"name": "prompt", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 5888994, "num_examples": 3000}], "download_size": 2569124, "dataset_size": 5888994}}
|
2023-10-26T16:08:32+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "code_searchnet_reduced_train"
More Information needed
|
[
"# Dataset Card for \"code_searchnet_reduced_train\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"code_searchnet_reduced_train\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"code_searchnet_reduced_train\"\n\nMore Information needed"
] |
94117593771d20dd7f3ad169bbf8a8789570b470
|
# Dataset Card for "code_searchnet_reduced_val"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
vishnusr/code_searchnet_reduced_val
|
[
"region:us"
] |
2023-10-26T16:08:32+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "Unnamed: 0.1", "dtype": "int64"}, {"name": "Unnamed: 0", "dtype": "int64"}, {"name": "code", "dtype": "string"}, {"name": "docstring", "dtype": "string"}, {"name": "prompt", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 1078734, "num_examples": 500}], "download_size": 483209, "dataset_size": 1078734}}
|
2023-10-26T16:08:36+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "code_searchnet_reduced_val"
More Information needed
|
[
"# Dataset Card for \"code_searchnet_reduced_val\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"code_searchnet_reduced_val\"\n\nMore Information needed"
] |
[
6,
20
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"code_searchnet_reduced_val\"\n\nMore Information needed"
] |
6e6acb0fa7e49e96ef080f86b4f1c14d6e740f75
|
# Dataset Card for "lcr"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
CJWeiss/lcr
|
[
"region:us"
] |
2023-10-26T16:09:59+00:00
|
{"dataset_info": {"features": [{"name": "Long Text", "dtype": "string"}, {"name": "Summary", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 82108819, "num_examples": 2918}, {"name": "test", "num_bytes": 18916443, "num_examples": 584}, {"name": "valid", "num_bytes": 12955974, "num_examples": 389}], "download_size": 56044522, "dataset_size": 113981236}}
|
2023-10-26T16:10:08+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "lcr"
More Information needed
|
[
"# Dataset Card for \"lcr\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"lcr\"\n\nMore Information needed"
] |
[
6,
12
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"lcr\"\n\nMore Information needed"
] |
eafc13530541e6da5779a03c848f9daeb9a8ad48
|
This dataset is a work in progress. It will be used to train execution of a subset of Emacs Lisp within the LLM according to the techniques described in this paper: https://arxiv.org/abs/2305.05383
|
notoriousdto/synthetic-elisp-alpha-0.1
|
[
"license:mit",
"arxiv:2305.05383",
"region:us"
] |
2023-10-26T16:45:37+00:00
|
{"license": "mit"}
|
2023-11-01T01:40:17+00:00
|
[
"2305.05383"
] |
[] |
TAGS
#license-mit #arxiv-2305.05383 #region-us
|
This dataset is a work in progress. It will be used to train execution of a subset of Emacs Lisp within the LLM according to the techniques described in this paper: URL
|
[] |
[
"TAGS\n#license-mit #arxiv-2305.05383 #region-us \n"
] |
[
20
] |
[
"passage: TAGS\n#license-mit #arxiv-2305.05383 #region-us \n"
] |
c2a414daba8bf496888e21c9cf4f50cd1e8656de
|
# Version 1 of the dataset
## Structure of the dataset:
- ### Opening_type:
The title of the opening being played.
- ### Context:
A string representing a list of moves, each move is represented by the previous state of the board, the move that is going to be made, and the effect that the move had on the board.<br>
- The <b>board</b> is represented as an 8*8 grid of characters where each character represents a piece or an empty square:<br>
~~~
r . . q k b n r
p p p . p . p p
. . n . . p . .
. . . p . b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
~~~
- The <b>move</b> is represented by the uci format g8f6, specifying that the piece is square g8 moves to the square f6
- The <b>type of move</b> is represented by a list of integers separated by ',' where each integer represents the effect that the move will have on the board.
- 0 if it is a move without capture
- 1 if it is a move with capture
- 2 if a check is being made
- 3 if it is check mate
- 4 if it is en passant capture
- 5 if it is king side castling
- 6 if it is queenside castling
- 7 if it is a draw by stalemate
- 8 if there is a draw by insufficient material
- 9 if it is a draw by seventy-five moves rule
- 10 if it is a draw by fivefold repetition
- The whole context can look something like this:
After each board, there is a move, and the effect of the move generates the next board. A context list always ends with a board because the following two columns represent the move to be played and the effect that it'll have on the next board.
~~~
r . . q k b n r
p p p . p . p p
. . n . . p . .
. . . p . b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:e7e5
t:0
r . . q k b n r
p p p . . . p p
. . n . . p . .
. . . p p b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:d4e5
t:1
r . . q k b n r
p p p . . . p p
. . n . . p . .
. . . p P b . .
. . . . . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:f6e5
t:1
r . . q k b n r
p p p . . . p p
. . n . . . . .
. . . p p b . .
. . . . . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
~~~
- ## Move_type_pred:
- Follows the same format described in the context column with <b>Type move</b>
- ## Move_pred:
- Follows the same format described in the context column with <b>move</b>
## Creation process:
- ### Loading the Dataset:
- The code loads a dataset of chess games in PGN format using the Hugging Face **datasets** library. The dataset is called [patrickfrank1/chess-pgn-games](https://huggingface.co/datasets/patrickfrank1/chess-pgn-games)!.
- ### Parsing and Organizing Game Text:
- It extracts game text from the dataset and organizes it based on metadata and moves information.
- ### Parsing Game Information:
- It extracts relevant information from the game headers, such as player Elo ratings and opening names.
- ### Iterating Through Games:
- It iterates through each game and processes it if it has a specified opening and if at least one player has an Elo rating greater than 1700.
- ### Sampling Moves for Context:
- For selected games, it randomly samples subarrays of moves from the mainline of the game.
- ### Recording Context Information:
- It records the board state, move information, and move type prediction for each move in the sampled context.
- ### Storing Processed Data:
- The extracted information is stored in a dictionary and then converted to a data frame. The data frame is uploaded to the Huggingface Dataset hub. (As you can see)
- The code to create this dataset can be found here: [chess_openings_teacher/ML/Dataset_Creation](https://github.com/bit2424/chess_openings_teacher/tree/main/ML/Dataset_Creation)!
## Intuitions behind the design:
- The idea is that by creating the whole board grid, the model can learn to grasp the effect that a move has on the board and create a richer representation of the game.
- One of the aims of this representation is to help predict logical moves even without needing the game's history, just using the current state of the board in the grid representation.
|
nelson2424/Chess_openings_dataset
|
[
"task_categories:text-classification",
"task_categories:text-generation",
"task_categories:text2text-generation",
"language:en",
"license:mit",
"region:us"
] |
2023-10-26T16:55:21+00:00
|
{"language": ["en"], "license": "mit", "task_categories": ["text-classification", "text-generation", "text2text-generation"], "pretty_name": "Cot-dataset"}
|
2023-11-09T02:59:48+00:00
|
[] |
[
"en"
] |
TAGS
#task_categories-text-classification #task_categories-text-generation #task_categories-text2text-generation #language-English #license-mit #region-us
|
# Version 1 of the dataset
## Structure of the dataset:
- ### Opening_type:
The title of the opening being played.
- ### Context:
A string representing a list of moves, each move is represented by the previous state of the board, the move that is going to be made, and the effect that the move had on the board.<br>
- The <b>board</b> is represented as an 8*8 grid of characters where each character represents a piece or an empty square:<br>
~~~
r . . q k b n r
p p p . p . p p
. . n . . p . .
. . . p . b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
~~~
- The <b>move</b> is represented by the uci format g8f6, specifying that the piece is square g8 moves to the square f6
- The <b>type of move</b> is represented by a list of integers separated by ',' where each integer represents the effect that the move will have on the board.
- 0 if it is a move without capture
- 1 if it is a move with capture
- 2 if a check is being made
- 3 if it is check mate
- 4 if it is en passant capture
- 5 if it is king side castling
- 6 if it is queenside castling
- 7 if it is a draw by stalemate
- 8 if there is a draw by insufficient material
- 9 if it is a draw by seventy-five moves rule
- 10 if it is a draw by fivefold repetition
- The whole context can look something like this:
After each board, there is a move, and the effect of the move generates the next board. A context list always ends with a board because the following two columns represent the move to be played and the effect that it'll have on the next board.
~~~
r . . q k b n r
p p p . p . p p
. . n . . p . .
. . . p . b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:e7e5
t:0
r . . q k b n r
p p p . . . p p
. . n . . p . .
. . . p p b . .
. . . P . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:d4e5
t:1
r . . q k b n r
p p p . . . p p
. . n . . p . .
. . . p P b . .
. . . . . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
m:f6e5
t:1
r . . q k b n r
p p p . . . p p
. . n . . . . .
. . . p p b . .
. . . . . . . B
. . . . P N . .
P P P . . P P P
R N . Q K B . R
~~~
- ## Move_type_pred:
- Follows the same format described in the context column with <b>Type move</b>
- ## Move_pred:
- Follows the same format described in the context column with <b>move</b>
## Creation process:
- ### Loading the Dataset:
- The code loads a dataset of chess games in PGN format using the Hugging Face datasets library. The dataset is called patrickfrank1/chess-pgn-games!.
- ### Parsing and Organizing Game Text:
- It extracts game text from the dataset and organizes it based on metadata and moves information.
- ### Parsing Game Information:
- It extracts relevant information from the game headers, such as player Elo ratings and opening names.
- ### Iterating Through Games:
- It iterates through each game and processes it if it has a specified opening and if at least one player has an Elo rating greater than 1700.
- ### Sampling Moves for Context:
- For selected games, it randomly samples subarrays of moves from the mainline of the game.
- ### Recording Context Information:
- It records the board state, move information, and move type prediction for each move in the sampled context.
- ### Storing Processed Data:
- The extracted information is stored in a dictionary and then converted to a data frame. The data frame is uploaded to the Huggingface Dataset hub. (As you can see)
- The code to create this dataset can be found here: chess_openings_teacher/ML/Dataset_Creation!
## Intuitions behind the design:
- The idea is that by creating the whole board grid, the model can learn to grasp the effect that a move has on the board and create a richer representation of the game.
- One of the aims of this representation is to help predict logical moves even without needing the game's history, just using the current state of the board in the grid representation.
|
[
"# Version 1 of the dataset\n ## Structure of the dataset:\n - ### Opening_type:\n The title of the opening being played.\n - ### Context:\n A string representing a list of moves, each move is represented by the previous state of the board, the move that is going to be made, and the effect that the move had on the board.<br>\n \n - The <b>board</b> is represented as an 8*8 grid of characters where each character represents a piece or an empty square:<br>\n \n ~~~ \n r . . q k b n r \n p p p . p . p p \n . . n . . p . . \n . . . p . b . . \n . . . P . . . B \n . . . . P N . . \n P P P . . P P P \n R N . Q K B . R\n ~~~\n \n - The <b>move</b> is represented by the uci format g8f6, specifying that the piece is square g8 moves to the square f6 \n - The <b>type of move</b> is represented by a list of integers separated by ',' where each integer represents the effect that the move will have on the board.\n - 0 if it is a move without capture\n - 1 if it is a move with capture\n - 2 if a check is being made\n - 3 if it is check mate\n - 4 if it is en passant capture\n - 5 if it is king side castling \n - 6 if it is queenside castling \n - 7 if it is a draw by stalemate\n - 8 if there is a draw by insufficient material\n - 9 if it is a draw by seventy-five moves rule \n - 10 if it is a draw by fivefold repetition \n \n - The whole context can look something like this:\n After each board, there is a move, and the effect of the move generates the next board. A context list always ends with a board because the following two columns represent the move to be played and the effect that it'll have on the next board. \n ~~~\n r . . q k b n r\n p p p . p . p p\n . . n . . p . .\n . . . p . b . .\n . . . P . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:e7e5\n t:0\n r . . q k b n r\n p p p . . . p p\n . . n . . p . .\n . . . p p b . .\n . . . P . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:d4e5\n t:1\n r . . q k b n r\n p p p . . . p p\n . . n . . p . .\n . . . p P b . .\n . . . . . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:f6e5\n t:1\n r . . q k b n r\n p p p . . . p p\n . . n . . . . .\n . . . p p b . .\n . . . . . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n ~~~\n - ## Move_type_pred: \n - Follows the same format described in the context column with <b>Type move</b>\n - ## Move_pred:\n - Follows the same format described in the context column with <b>move</b>\n \n ## Creation process:\n - ### Loading the Dataset: \n - The code loads a dataset of chess games in PGN format using the Hugging Face datasets library. The dataset is called patrickfrank1/chess-pgn-games!.\n - ### Parsing and Organizing Game Text:\n - It extracts game text from the dataset and organizes it based on metadata and moves information.\n - ### Parsing Game Information:\n - It extracts relevant information from the game headers, such as player Elo ratings and opening names.\n - ### Iterating Through Games:\n - It iterates through each game and processes it if it has a specified opening and if at least one player has an Elo rating greater than 1700.\n - ### Sampling Moves for Context:\n - For selected games, it randomly samples subarrays of moves from the mainline of the game.\n - ### Recording Context Information:\n - It records the board state, move information, and move type prediction for each move in the sampled context.\n - ### Storing Processed Data:\n - The extracted information is stored in a dictionary and then converted to a data frame. The data frame is uploaded to the Huggingface Dataset hub. (As you can see) \n - The code to create this dataset can be found here: chess_openings_teacher/ML/Dataset_Creation!\n ## Intuitions behind the design:\n - The idea is that by creating the whole board grid, the model can learn to grasp the effect that a move has on the board and create a richer representation of the game.\n - One of the aims of this representation is to help predict logical moves even without needing the game's history, just using the current state of the board in the grid representation."
] |
[
"TAGS\n#task_categories-text-classification #task_categories-text-generation #task_categories-text2text-generation #language-English #license-mit #region-us \n",
"# Version 1 of the dataset\n ## Structure of the dataset:\n - ### Opening_type:\n The title of the opening being played.\n - ### Context:\n A string representing a list of moves, each move is represented by the previous state of the board, the move that is going to be made, and the effect that the move had on the board.<br>\n \n - The <b>board</b> is represented as an 8*8 grid of characters where each character represents a piece or an empty square:<br>\n \n ~~~ \n r . . q k b n r \n p p p . p . p p \n . . n . . p . . \n . . . p . b . . \n . . . P . . . B \n . . . . P N . . \n P P P . . P P P \n R N . Q K B . R\n ~~~\n \n - The <b>move</b> is represented by the uci format g8f6, specifying that the piece is square g8 moves to the square f6 \n - The <b>type of move</b> is represented by a list of integers separated by ',' where each integer represents the effect that the move will have on the board.\n - 0 if it is a move without capture\n - 1 if it is a move with capture\n - 2 if a check is being made\n - 3 if it is check mate\n - 4 if it is en passant capture\n - 5 if it is king side castling \n - 6 if it is queenside castling \n - 7 if it is a draw by stalemate\n - 8 if there is a draw by insufficient material\n - 9 if it is a draw by seventy-five moves rule \n - 10 if it is a draw by fivefold repetition \n \n - The whole context can look something like this:\n After each board, there is a move, and the effect of the move generates the next board. A context list always ends with a board because the following two columns represent the move to be played and the effect that it'll have on the next board. \n ~~~\n r . . q k b n r\n p p p . p . p p\n . . n . . p . .\n . . . p . b . .\n . . . P . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:e7e5\n t:0\n r . . q k b n r\n p p p . . . p p\n . . n . . p . .\n . . . p p b . .\n . . . P . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:d4e5\n t:1\n r . . q k b n r\n p p p . . . p p\n . . n . . p . .\n . . . p P b . .\n . . . . . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n m:f6e5\n t:1\n r . . q k b n r\n p p p . . . p p\n . . n . . . . .\n . . . p p b . .\n . . . . . . . B\n . . . . P N . .\n P P P . . P P P\n R N . Q K B . R\n ~~~\n - ## Move_type_pred: \n - Follows the same format described in the context column with <b>Type move</b>\n - ## Move_pred:\n - Follows the same format described in the context column with <b>move</b>\n \n ## Creation process:\n - ### Loading the Dataset: \n - The code loads a dataset of chess games in PGN format using the Hugging Face datasets library. The dataset is called patrickfrank1/chess-pgn-games!.\n - ### Parsing and Organizing Game Text:\n - It extracts game text from the dataset and organizes it based on metadata and moves information.\n - ### Parsing Game Information:\n - It extracts relevant information from the game headers, such as player Elo ratings and opening names.\n - ### Iterating Through Games:\n - It iterates through each game and processes it if it has a specified opening and if at least one player has an Elo rating greater than 1700.\n - ### Sampling Moves for Context:\n - For selected games, it randomly samples subarrays of moves from the mainline of the game.\n - ### Recording Context Information:\n - It records the board state, move information, and move type prediction for each move in the sampled context.\n - ### Storing Processed Data:\n - The extracted information is stored in a dictionary and then converted to a data frame. The data frame is uploaded to the Huggingface Dataset hub. (As you can see) \n - The code to create this dataset can be found here: chess_openings_teacher/ML/Dataset_Creation!\n ## Intuitions behind the design:\n - The idea is that by creating the whole board grid, the model can learn to grasp the effect that a move has on the board and create a richer representation of the game.\n - One of the aims of this representation is to help predict logical moves even without needing the game's history, just using the current state of the board in the grid representation."
] |
[
50,
1363
] |
[
"passage: TAGS\n#task_categories-text-classification #task_categories-text-generation #task_categories-text2text-generation #language-English #license-mit #region-us \n"
] |
21b10527fdbeaf3e3fce9d3e12a15a7ce689b8ec
|
# Dataset Card for "AncientMNIST"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
SummerSigh/AncientMNIST
|
[
"region:us"
] |
2023-10-26T17:03:37+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}], "dataset_info": {"features": [{"name": "image", "dtype": "image"}, {"name": "label", "dtype": {"class_label": {"names": {"0": "Alpha", "1": "Beta", "2": "Chi", "3": "Delta", "4": "Epsilon", "5": "Eta", "6": "Gamma", "7": "Iota", "8": "Kappa", "9": "Lambda", "10": "LunateSigma", "11": "Mu", "12": "Nu", "13": "Omega", "14": "Omicron", "15": "Phi", "16": "Pi", "17": "Psi", "18": "Rho", "19": "Tau", "20": "Theta", "21": "Upsilon", "22": "Xi", "23": "Zeta"}}}}], "splits": [{"name": "train", "num_bytes": 309609553.26, "num_examples": 205797}], "download_size": 217254607, "dataset_size": 309609553.26}}
|
2023-10-26T17:06:58+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "AncientMNIST"
More Information needed
|
[
"# Dataset Card for \"AncientMNIST\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"AncientMNIST\"\n\nMore Information needed"
] |
[
6,
14
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"AncientMNIST\"\n\nMore Information needed"
] |
d873ce221f3d505b7004366bf834a019a001108f
|
# Dataset Card for "vira-intents-mod"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
vira-chatbot/vira-intents-mod
|
[
"region:us"
] |
2023-10-26T17:04:54+00:00
|
{"dataset_info": {"features": [{"name": "text", "dtype": "string"}, {"name": "label", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 509234, "num_examples": 7047}, {"name": "validation", "num_bytes": 213834, "num_examples": 2971}], "download_size": 329146, "dataset_size": 723068}}
|
2023-10-31T03:54:15+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "vira-intents-mod"
More Information needed
|
[
"# Dataset Card for \"vira-intents-mod\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"vira-intents-mod\"\n\nMore Information needed"
] |
[
6,
16
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"vira-intents-mod\"\n\nMore Information needed"
] |
c1ac8638f0d02a443e071c084c272c8b79ac856e
|
# Dataset Card for "humansleepproject-rr-small-individuals"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
emi429/humansleepproject-rr-small-individuals
|
[
"region:us"
] |
2023-10-26T17:41:07+00:00
|
{"dataset_info": {"features": [{"name": "rr_intervals", "sequence": "float64"}, {"name": "sleep_stage", "dtype": "string"}, {"name": "patient_id", "dtype": "string"}], "splits": [{"name": "test", "num_bytes": 1631857, "num_examples": 504}, {"name": "train", "num_bytes": 5747903, "num_examples": 2070}], "download_size": 1335531, "dataset_size": 7379760}}
|
2023-10-26T17:41:16+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "humansleepproject-rr-small-individuals"
More Information needed
|
[
"# Dataset Card for \"humansleepproject-rr-small-individuals\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"humansleepproject-rr-small-individuals\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"humansleepproject-rr-small-individuals\"\n\nMore Information needed"
] |
10e3c648040c01e8ea414390322ed0540475e735
|
# Dataset Card for "salt-llama-lgg-to-eng"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
mekaneeky/salt-llama-lgg-to-eng
|
[
"region:us"
] |
2023-10-26T17:53:00+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "dev", "path": "data/dev-*"}, {"split": "test", "path": "data/test-*"}]}], "dataset_info": {"features": [{"name": "ID", "dtype": "string"}, {"name": "text", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 4130369, "num_examples": 23947}, {"name": "dev", "num_bytes": 85575, "num_examples": 500}, {"name": "test", "num_bytes": 87440, "num_examples": 500}], "download_size": 2324474, "dataset_size": 4303384}}
|
2023-10-26T17:53:04+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "salt-llama-lgg-to-eng"
More Information needed
|
[
"# Dataset Card for \"salt-llama-lgg-to-eng\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"salt-llama-lgg-to-eng\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"salt-llama-lgg-to-eng\"\n\nMore Information needed"
] |
e5843efab7787ccb216cacc92b22e551101401b5
|
# Dataset Card for "filtered_lemma41kV0.0.1"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
fia24/filtered_lemma41kV0.0.1
|
[
"region:us"
] |
2023-10-26T17:58:59+00:00
|
{"configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "test", "path": "data/test-*"}, {"split": "val", "path": "data/val-*"}]}], "dataset_info": {"features": [{"name": "Unnamed: 0", "dtype": "int64"}, {"name": "Inflected_Word", "dtype": "string"}, {"name": "Lemma", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 1841860.2133993004, "num_examples": 29267}, {"name": "test", "num_bytes": 230271.85980209926, "num_examples": 3659}, {"name": "val", "num_bytes": 230208.92679860047, "num_examples": 3658}], "download_size": 1233470, "dataset_size": 2302341.0}}
|
2023-10-26T17:59:05+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "filtered_lemma41kV0.0.1"
More Information needed
|
[
"# Dataset Card for \"filtered_lemma41kV0.0.1\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"filtered_lemma41kV0.0.1\"\n\nMore Information needed"
] |
[
6,
21
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"filtered_lemma41kV0.0.1\"\n\nMore Information needed"
] |
3f58dc31d760b81f613d76470296990e3e0a82d1
|
Bestande [1] was an academic review and rating platform which was very popular within UZH and ETH student communities.
Bestande terminated its service in 2022 and with the hope of someone taking over, the founder disclosed the database over 2016 and 2022 on Bestande.ch. Therefore, we took it without any data scraping and built our project on it. The raw data is a json file.
The provided files offer a translation of reviews from different comments to German, where reviews of a course are given, as well as the score of the course, up-votes and down-votes that the review received and labeling that states if a review is useful or not.
For the binary dataset:
- 0 : neutral review
- 1 : useful review
For the multiclass dataset:
- 0 : neutral review
- 1 : slightly useful review
- 2 : useful review
- 3 : extremely useful review
[1] J. Burger. Bestande. [Accessed 21-09-2023]. [Online]. Available: http://www.
bestande.ch
|
jorgeortizv/Bestande
|
[
"language:de",
"region:us"
] |
2023-10-26T18:33:27+00:00
|
{"language": ["de"]}
|
2023-10-26T18:40:39+00:00
|
[] |
[
"de"
] |
TAGS
#language-German #region-us
|
Bestande [1] was an academic review and rating platform which was very popular within UZH and ETH student communities.
Bestande terminated its service in 2022 and with the hope of someone taking over, the founder disclosed the database over 2016 and 2022 on URL. Therefore, we took it without any data scraping and built our project on it. The raw data is a json file.
The provided files offer a translation of reviews from different comments to German, where reviews of a course are given, as well as the score of the course, up-votes and down-votes that the review received and labeling that states if a review is useful or not.
For the binary dataset:
- 0 : neutral review
- 1 : useful review
For the multiclass dataset:
- 0 : neutral review
- 1 : slightly useful review
- 2 : useful review
- 3 : extremely useful review
[1] J. Burger. Bestande. [Accessed 21-09-2023]. [Online]. Available: http://www.
URL
|
[] |
[
"TAGS\n#language-German #region-us \n"
] |
[
10
] |
[
"passage: TAGS\n#language-German #region-us \n"
] |
d41b02d05a346349ef491fe6eaad4c94c38c259c
|
# Dataset Card for "magic-the-gathering"
This is a HuggingFace adaptation of the [MTGJSON Atomic Card Database](https://mtgjson.com/data-models/card/card-atomic/) from the Taj-Mahal Data Science & Machine Learning Group.
## Usage
```
from datasets import load_dataset
dataset = load_dataset("MechaCroc/magic-the-gathering")
```
## Notes
- Power, Toughness, and Loyalty are strings because of the rare cases like [Tarmogoyf](https://gatherer.wizards.com/pages/Card/Details.aspx?multiverseid=136142) where the P/T is `* / 1+*`.
|
MechaCroc/magic-the-gathering
|
[
"region:us"
] |
2023-10-26T18:43:32+00:00
|
{"dataset_info": {"features": [{"name": "name", "dtype": "string"}, {"name": "firstPrinting", "dtype": "string"}, {"name": "manaCost", "dtype": "string"}, {"name": "convertedManaCost", "dtype": "float64"}, {"name": "type", "dtype": "string"}, {"name": "text", "dtype": "string"}, {"name": "power", "dtype": "string"}, {"name": "toughness", "dtype": "string"}, {"name": "loyalty", "dtype": "string"}, {"name": "layout", "dtype": "string"}], "splits": [{"name": "train", "num_bytes": 6999997, "num_examples": 27703}, {"name": "train_clean", "num_bytes": 6813519.081146446, "num_examples": 26965}], "download_size": 2539289, "dataset_size": 13813516.081146445}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}, {"split": "train_clean", "path": "data/train_clean-*"}]}]}
|
2024-01-29T01:00:39+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "magic-the-gathering"
This is a HuggingFace adaptation of the MTGJSON Atomic Card Database from the Taj-Mahal Data Science & Machine Learning Group.
## Usage
## Notes
- Power, Toughness, and Loyalty are strings because of the rare cases like Tarmogoyf where the P/T is '* / 1+*'.
|
[
"# Dataset Card for \"magic-the-gathering\"\n\nThis is a HuggingFace adaptation of the MTGJSON Atomic Card Database from the Taj-Mahal Data Science & Machine Learning Group.",
"## Usage",
"## Notes\n- Power, Toughness, and Loyalty are strings because of the rare cases like Tarmogoyf where the P/T is '* / 1+*'."
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"magic-the-gathering\"\n\nThis is a HuggingFace adaptation of the MTGJSON Atomic Card Database from the Taj-Mahal Data Science & Machine Learning Group.",
"## Usage",
"## Notes\n- Power, Toughness, and Loyalty are strings because of the rare cases like Tarmogoyf where the P/T is '* / 1+*'."
] |
[
6,
47,
3,
41
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"magic-the-gathering\"\n\nThis is a HuggingFace adaptation of the MTGJSON Atomic Card Database from the Taj-Mahal Data Science & Machine Learning Group.## Usage## Notes\n- Power, Toughness, and Loyalty are strings because of the rare cases like Tarmogoyf where the P/T is '* / 1+*'."
] |
50f0f821daa8d509408da357a499a0fdce6294e2
|
# Dataset Card for "chemnlp-mol-svg"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
kjappelbaum/chemnlp-mol-svg
|
[
"region:us"
] |
2023-10-26T18:45:35+00:00
|
{"dataset_info": {"features": [{"name": "prompt", "dtype": "string"}, {"name": "completion", "dtype": "string"}, {"name": "smiles", "dtype": "string"}, {"name": "__index_level_0__", "dtype": "int64"}], "splits": [{"name": "train", "num_bytes": 67669990, "num_examples": 17807}], "download_size": 19012111, "dataset_size": 67669990}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T19:41:43+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "chemnlp-mol-svg"
More Information needed
|
[
"# Dataset Card for \"chemnlp-mol-svg\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"chemnlp-mol-svg\"\n\nMore Information needed"
] |
[
6,
18
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"chemnlp-mol-svg\"\n\nMore Information needed"
] |
c066947a5acbaf52eee763fa5657bab618588fe6
|
# Dataset Card for "flood_dataset"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
gdurkin/flood_dataset
|
[
"region:us"
] |
2023-10-26T19:07:33+00:00
|
{"dataset_info": {"features": [{"name": "pixel_values", "dtype": {"array3_d": {"shape": [512, 512, 3], "dtype": "uint8"}}}, {"name": "label", "dtype": "image"}], "splits": [{"name": "train", "num_bytes": 464353592.0, "num_examples": 252}], "download_size": 198779583, "dataset_size": 464353592.0}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-11-01T13:43:12+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "flood_dataset"
More Information needed
|
[
"# Dataset Card for \"flood_dataset\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"flood_dataset\"\n\nMore Information needed"
] |
[
6,
15
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"flood_dataset\"\n\nMore Information needed"
] |
97f6f6d845ddd37977a386e98124cf1068946451
|
# Deep Learning Service Project (Fall 2023)
# Getting Started
1. Clone the repository with git lfs disabled or not installed.
**ON WINDOWS**
```bash
set GIT_LFS_SKIP_SMUDGE=1
git clone https://huggingface.co/datasets/DataScienceClubUVU/ServiceProjectFall2023
```
**ON LINUX**
```bash
GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/datasets/DataScienceClubUVU/ServiceProjectFall2023
```
2. Download the pytorch file (.pth) from https://huggingface.co/datasets/DataScienceClubUVU/ServiceProjectFall2023/blob/main/mexico_5_column_weights.pth and place it in the root directory of the repository. There will be an existing file with the same name. Delete it and replace it with the new one.
3. Install the requirements.txt file using pip.
**IF YOU DON't HAVE A GPU**
```bash
pip install -r requirements_cpu.txt
```
**IF YOU HAVE A GPU**
1. Install reqs without torch
```bash
pip install -r requirements_no_torch.txt
```
2. Install pytorch by following the instructions at https://pytorch.org/get-started/locally/
### How the Model Works
1. Start with an image of a character of text:
- 
2. Convert the image between RGB/BGR and grayscale using the _**cvtColor**_ function from the _**cv2**_ library:
- 
3. Use an Adaptive Thresholding approach where the threshold value = Gaussian weighted sum of the neighborhood values - constant value. In other words, it is a weighted sum of the blockSize^2 neighborhood of a point minus the constant. in this example, we are setting the maximum threshold value as 255 with the block size of 155 and the constant is 2.
- 
4. Create a 3x3 matrix of ones to generate an image kernel. An _**image kernel**_ is a small matrix used to apply effects like the ones you might find in Photoshop or Gimp, such as blurring, sharpening, outlining or embossing. They're also used in machine learning for 'feature extraction', a technique for determining the most important portions of an image.
5. The basic idea of erosion is just like soil erosion only, it erodes away the boundaries of foreground object (Always try to keep foreground in white). It is normally performed on binary images. It needs two inputs, one is our original image, second one is called structuring element or kernel which decides the nature of operation. A pixel in the original image (either 1 or 0) will be considered 1 only if all the pixels under the kernel is 1, otherwise it is eroded (made to zero).
- 
6. The basic idea of dilation is accentuating the features of the images. Whereas erosion is used to reduce the amount of noise in the image, dilation is used to enhance the features of the image.
- 
7. Traditionally, a line can be represented by the equation **_y=mx + b_** (where **_m_** is the slope and **_b_** is the intercept). However, a line can also be represented by the following equation: **_r= x(cos0) + y(sin0)_** (where **_r_** is the distance from the origin to the closest point on the straight line). **_(r,0)_** corresponds corresponds to the **_Hough space_** representation of the line. In this case, **_0_** is known as **_theta_**.
- For a given point in a two-dimensional space (think of a basic x- and y-axis graph), there can be an infinite number of straight lines drawn through the point. With a **_Hough Transform_**, you draw several lines through the point to create a table of values where you conclude "for given theta (angle between the x-axis and r-line that will match with the closest point on the straight line), we can expect this "r" value".
- Once you have created your table of values for each point on a given two-dimensional space, you compare the r-values on each theta for each given point and select the r and theta where the difference between the point is the least (this means the line best represents the points on the space).
|
DataScienceClubUVU/ServiceProjectFall2023
|
[
"region:us"
] |
2023-10-26T19:16:29+00:00
|
{}
|
2023-11-09T01:43:10+00:00
|
[] |
[] |
TAGS
#region-us
|
# Deep Learning Service Project (Fall 2023)
!image
# Getting Started
1. Clone the repository with git lfs disabled or not installed.
ON WINDOWS
ON LINUX
2. Download the pytorch file (.pth) from URL and place it in the root directory of the repository. There will be an existing file with the same name. Delete it and replace it with the new one.
3. Install the URL file using pip.
IF YOU DON't HAVE A GPU
IF YOU HAVE A GPU
1. Install reqs without torch
2. Install pytorch by following the instructions at URL
### How the Model Works
1. Start with an image of a character of text:
- !output
2. Convert the image between RGB/BGR and grayscale using the _cvtColor_ function from the _cv2_ library:
- !image
3. Use an Adaptive Thresholding approach where the threshold value = Gaussian weighted sum of the neighborhood values - constant value. In other words, it is a weighted sum of the blockSize^2 neighborhood of a point minus the constant. in this example, we are setting the maximum threshold value as 255 with the block size of 155 and the constant is 2.
- !image
4. Create a 3x3 matrix of ones to generate an image kernel. An _image kernel_ is a small matrix used to apply effects like the ones you might find in Photoshop or Gimp, such as blurring, sharpening, outlining or embossing. They're also used in machine learning for 'feature extraction', a technique for determining the most important portions of an image.
5. The basic idea of erosion is just like soil erosion only, it erodes away the boundaries of foreground object (Always try to keep foreground in white). It is normally performed on binary images. It needs two inputs, one is our original image, second one is called structuring element or kernel which decides the nature of operation. A pixel in the original image (either 1 or 0) will be considered 1 only if all the pixels under the kernel is 1, otherwise it is eroded (made to zero).
- !image
6. The basic idea of dilation is accentuating the features of the images. Whereas erosion is used to reduce the amount of noise in the image, dilation is used to enhance the features of the image.
- !image
7. Traditionally, a line can be represented by the equation _y=mx + b_ (where _m_ is the slope and _b_ is the intercept). However, a line can also be represented by the following equation: _r= x(cos0) + y(sin0)_ (where _r_ is the distance from the origin to the closest point on the straight line). _(r,0)_ corresponds corresponds to the _Hough space_ representation of the line. In this case, _0_ is known as _theta_.
- For a given point in a two-dimensional space (think of a basic x- and y-axis graph), there can be an infinite number of straight lines drawn through the point. With a _Hough Transform_, you draw several lines through the point to create a table of values where you conclude "for given theta (angle between the x-axis and r-line that will match with the closest point on the straight line), we can expect this "r" value".
- Once you have created your table of values for each point on a given two-dimensional space, you compare the r-values on each theta for each given point and select the r and theta where the difference between the point is the least (this means the line best represents the points on the space).
|
[
"# Deep Learning Service Project (Fall 2023)\n!image",
"# Getting Started\n1. Clone the repository with git lfs disabled or not installed.\n\nON WINDOWS\n\n\nON LINUX\n\n\n2. Download the pytorch file (.pth) from URL and place it in the root directory of the repository. There will be an existing file with the same name. Delete it and replace it with the new one.\n\n3. Install the URL file using pip.\n\nIF YOU DON't HAVE A GPU\n\n\nIF YOU HAVE A GPU\n1. Install reqs without torch\n\n2. Install pytorch by following the instructions at URL",
"### How the Model Works\n1. Start with an image of a character of text:\n- !output\n\n2. Convert the image between RGB/BGR and grayscale using the _cvtColor_ function from the _cv2_ library:\n- !image\n3. Use an Adaptive Thresholding approach where the threshold value = Gaussian weighted sum of the neighborhood values - constant value. In other words, it is a weighted sum of the blockSize^2 neighborhood of a point minus the constant. in this example, we are setting the maximum threshold value as 255 with the block size of 155 and the constant is 2.\n- !image\n\n4. Create a 3x3 matrix of ones to generate an image kernel. An _image kernel_ is a small matrix used to apply effects like the ones you might find in Photoshop or Gimp, such as blurring, sharpening, outlining or embossing. They're also used in machine learning for 'feature extraction', a technique for determining the most important portions of an image.\n\n5. The basic idea of erosion is just like soil erosion only, it erodes away the boundaries of foreground object (Always try to keep foreground in white). It is normally performed on binary images. It needs two inputs, one is our original image, second one is called structuring element or kernel which decides the nature of operation. A pixel in the original image (either 1 or 0) will be considered 1 only if all the pixels under the kernel is 1, otherwise it is eroded (made to zero).\n- !image\n\n6. The basic idea of dilation is accentuating the features of the images. Whereas erosion is used to reduce the amount of noise in the image, dilation is used to enhance the features of the image.\n- !image\n\n7. Traditionally, a line can be represented by the equation _y=mx + b_ (where _m_ is the slope and _b_ is the intercept). However, a line can also be represented by the following equation: _r= x(cos0) + y(sin0)_ (where _r_ is the distance from the origin to the closest point on the straight line). _(r,0)_ corresponds corresponds to the _Hough space_ representation of the line. In this case, _0_ is known as _theta_.\n \n- For a given point in a two-dimensional space (think of a basic x- and y-axis graph), there can be an infinite number of straight lines drawn through the point. With a _Hough Transform_, you draw several lines through the point to create a table of values where you conclude \"for given theta (angle between the x-axis and r-line that will match with the closest point on the straight line), we can expect this \"r\" value\".\n- Once you have created your table of values for each point on a given two-dimensional space, you compare the r-values on each theta for each given point and select the r and theta where the difference between the point is the least (this means the line best represents the points on the space)."
] |
[
"TAGS\n#region-us \n",
"# Deep Learning Service Project (Fall 2023)\n!image",
"# Getting Started\n1. Clone the repository with git lfs disabled or not installed.\n\nON WINDOWS\n\n\nON LINUX\n\n\n2. Download the pytorch file (.pth) from URL and place it in the root directory of the repository. There will be an existing file with the same name. Delete it and replace it with the new one.\n\n3. Install the URL file using pip.\n\nIF YOU DON't HAVE A GPU\n\n\nIF YOU HAVE A GPU\n1. Install reqs without torch\n\n2. Install pytorch by following the instructions at URL",
"### How the Model Works\n1. Start with an image of a character of text:\n- !output\n\n2. Convert the image between RGB/BGR and grayscale using the _cvtColor_ function from the _cv2_ library:\n- !image\n3. Use an Adaptive Thresholding approach where the threshold value = Gaussian weighted sum of the neighborhood values - constant value. In other words, it is a weighted sum of the blockSize^2 neighborhood of a point minus the constant. in this example, we are setting the maximum threshold value as 255 with the block size of 155 and the constant is 2.\n- !image\n\n4. Create a 3x3 matrix of ones to generate an image kernel. An _image kernel_ is a small matrix used to apply effects like the ones you might find in Photoshop or Gimp, such as blurring, sharpening, outlining or embossing. They're also used in machine learning for 'feature extraction', a technique for determining the most important portions of an image.\n\n5. The basic idea of erosion is just like soil erosion only, it erodes away the boundaries of foreground object (Always try to keep foreground in white). It is normally performed on binary images. It needs two inputs, one is our original image, second one is called structuring element or kernel which decides the nature of operation. A pixel in the original image (either 1 or 0) will be considered 1 only if all the pixels under the kernel is 1, otherwise it is eroded (made to zero).\n- !image\n\n6. The basic idea of dilation is accentuating the features of the images. Whereas erosion is used to reduce the amount of noise in the image, dilation is used to enhance the features of the image.\n- !image\n\n7. Traditionally, a line can be represented by the equation _y=mx + b_ (where _m_ is the slope and _b_ is the intercept). However, a line can also be represented by the following equation: _r= x(cos0) + y(sin0)_ (where _r_ is the distance from the origin to the closest point on the straight line). _(r,0)_ corresponds corresponds to the _Hough space_ representation of the line. In this case, _0_ is known as _theta_.\n \n- For a given point in a two-dimensional space (think of a basic x- and y-axis graph), there can be an infinite number of straight lines drawn through the point. With a _Hough Transform_, you draw several lines through the point to create a table of values where you conclude \"for given theta (angle between the x-axis and r-line that will match with the closest point on the straight line), we can expect this \"r\" value\".\n- Once you have created your table of values for each point on a given two-dimensional space, you compare the r-values on each theta for each given point and select the r and theta where the difference between the point is the least (this means the line best represents the points on the space)."
] |
[
6,
12,
124,
708
] |
[
"passage: TAGS\n#region-us \n# Deep Learning Service Project (Fall 2023)\n!image# Getting Started\n1. Clone the repository with git lfs disabled or not installed.\n\nON WINDOWS\n\n\nON LINUX\n\n\n2. Download the pytorch file (.pth) from URL and place it in the root directory of the repository. There will be an existing file with the same name. Delete it and replace it with the new one.\n\n3. Install the URL file using pip.\n\nIF YOU DON't HAVE A GPU\n\n\nIF YOU HAVE A GPU\n1. Install reqs without torch\n\n2. Install pytorch by following the instructions at URL"
] |
ff87963b1818dae4457b1b19d74cb9afbdee353b
|
# Dataset Card for "gpdr-dpr-dataset"
[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
jessica-ecosia/gpdr-dpr-dataset
|
[
"region:us"
] |
2023-10-26T19:27:04+00:00
|
{"dataset_info": {"features": [{"name": "id", "dtype": "string"}, {"name": "text", "dtype": "string"}, {"name": "title", "dtype": "string"}, {"name": "embeddings", "sequence": {"sequence": "float64"}}], "splits": [{"name": "train", "num_bytes": 4191740, "num_examples": 620}], "download_size": 0, "dataset_size": 4191740}, "configs": [{"config_name": "default", "data_files": [{"split": "train", "path": "data/train-*"}]}]}
|
2023-10-26T19:27:22+00:00
|
[] |
[] |
TAGS
#region-us
|
# Dataset Card for "gpdr-dpr-dataset"
More Information needed
|
[
"# Dataset Card for \"gpdr-dpr-dataset\"\n\nMore Information needed"
] |
[
"TAGS\n#region-us \n",
"# Dataset Card for \"gpdr-dpr-dataset\"\n\nMore Information needed"
] |
[
6,
18
] |
[
"passage: TAGS\n#region-us \n# Dataset Card for \"gpdr-dpr-dataset\"\n\nMore Information needed"
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.