modelId
string | author
string | last_modified
timestamp[us, tz=UTC] | downloads
int64 | likes
int64 | library_name
string | tags
list | pipeline_tag
string | createdAt
timestamp[us, tz=UTC] | card
string |
|---|---|---|---|---|---|---|---|---|---|
varun-v-rao/t5-large-lora-4.72M-squad-model1
|
varun-v-rao
| 2024-02-12T20:20:19Z | 34 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"t5",
"question-answering",
"generated_from_trainer",
"dataset:varun-v-rao/squad",
"base_model:google-t5/t5-large",
"base_model:finetune:google-t5/t5-large",
"license:apache-2.0",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
question-answering
| 2024-02-10T14:49:24Z |
---
license: apache-2.0
base_model: t5-large
tags:
- generated_from_trainer
datasets:
- varun-v-rao/squad
model-index:
- name: t5-large-lora-4.72M-squad-model1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# t5-large-lora-4.72M-squad-model1
This model is a fine-tuned version of [t5-large](https://huggingface.co/t5-large) on the squad dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 47
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0
|
SayeghJS/my_awesome_model2
|
SayeghJS
| 2024-02-12T20:02:11Z | 46 | 0 |
transformers
|
[
"transformers",
"tf",
"distilbert",
"text-classification",
"generated_from_keras_callback",
"base_model:distilbert/distilbert-base-uncased",
"base_model:finetune:distilbert/distilbert-base-uncased",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T09:50:00Z |
---
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_keras_callback
model-index:
- name: SayeghJS/my_awesome_model2
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# SayeghJS/my_awesome_model2
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.0723
- Validation Loss: 0.2143
- Train Accuracy: 0.9258
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 7810, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Validation Loss | Train Accuracy | Epoch |
|:----------:|:---------------:|:--------------:|:-----:|
| 0.2584 | 0.1973 | 0.9243 | 0 |
| 0.1387 | 0.2056 | 0.9270 | 1 |
| 0.0723 | 0.2143 | 0.9258 | 2 |
### Framework versions
- Transformers 4.35.2
- TensorFlow 2.15.0
- Datasets 2.17.0
- Tokenizers 0.15.1
|
SalmanHabeeb/qwen-llamafiles
|
SalmanHabeeb
| 2024-02-12T19:45:46Z | 198 | 0 |
transformers
|
[
"transformers",
"safetensors",
"gguf",
"llamafile",
"qwen2",
"text-generation",
"chat",
"conversational",
"en",
"arxiv:2309.16609",
"license:other",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-12T07:58:58Z |
---
license: other
license_name: tongyi-qianwen-research
license_link: >-
https://huggingface.co/Qwen/Qwen1.5-0.5B-Chat/blob/main/LICENSE
language:
- en
pipeline_tag: text-generation
tags:
- chat
---
# Qwen1.5-0.5B-Chat
## Introduction
Qwen1.5 is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include:
* 6 model sizes, including 0.5B, 1.8B, 4B, 7B, 14B, and 72B;
* Significant performance improvement in human preference for chat models;
* Multilingual support of both base and chat models;
* Stable support of 32K context length for models of all sizes
* No need of `trust_remote_code`.
For more details, please refer to our [blog post](https://qwenlm.github.io/blog/qwen1.5/) and [GitHub repo](https://github.com/QwenLM/Qwen1.5).
<br>
## Model Details
Qwen1.5 is a language model series including decoder language models of different model sizes. For each size, we release the base language model and the aligned chat model. It is based on the Transformer architecture with SwiGLU activation, attention QKV bias, group query attention, mixture of sliding window attention and full attention, etc. Additionally, we have an improved tokenizer adaptive to multiple natural languages and codes. For the beta version, temporarily we did not include GQA and the mixture of SWA and full attention.
## Training details
We pretrained the models with a large amount of data, and we post-trained the models with both supervised finetuning and direct preference optimization. However, DPO leads to improvements in human preference evaluation but degradation in benchmark evaluation. In the very near future, we will fix both problems.
## Requirements
The code of Qwen1.5 has been in the latest Hugging face transformers and we advise you to install `transformers>=4.37.0`, or you might encounter the following error:
```
KeyError: 'qwen2'
```
## Quickstart
Here provides a code snippet with `apply_chat_template` to show you how to load the tokenizer and model and how to generate contents.
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto
model = AutoModelForCausalLM.from_pretrained(
"Qwen/Qwen1.5-0.5B-Chat",
device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-0.5B-Chat")
prompt = "Give me a short introduction to large language model."
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)
generated_ids = model.generate(
model_inputs.input_ids,
max_new_tokens=512
)
generated_ids = [
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
```
For quantized models, we advise you to use the GPTQ, AWQ, and GGUF correspondents, namely `Qwen1.5-0.5B-Chat-GPTQ-Int4`, `Qwen1.5-0.5B-Chat-GPTQ-Int8`, `Qwen1.5-0.5B-Chat-AWQ`, and `Qwen1.5-0.5B-Chat-GGUF`.
## Tips
* If you encounter code switching or other bad cases, we advise you to use our provided hyper-parameters in `generation_config.json`.
## Citation
If you find our work helpful, feel free to give us a cite.
```
@article{qwen,
title={Qwen Technical Report},
author={Jinze Bai and Shuai Bai and Yunfei Chu and Zeyu Cui and Kai Dang and Xiaodong Deng and Yang Fan and Wenbin Ge and Yu Han and Fei Huang and Binyuan Hui and Luo Ji and Mei Li and Junyang Lin and Runji Lin and Dayiheng Liu and Gao Liu and Chengqiang Lu and Keming Lu and Jianxin Ma and Rui Men and Xingzhang Ren and Xuancheng Ren and Chuanqi Tan and Sinan Tan and Jianhong Tu and Peng Wang and Shijie Wang and Wei Wang and Shengguang Wu and Benfeng Xu and Jin Xu and An Yang and Hao Yang and Jian Yang and Shusheng Yang and Yang Yao and Bowen Yu and Hongyi Yuan and Zheng Yuan and Jianwei Zhang and Xingxuan Zhang and Yichang Zhang and Zhenru Zhang and Chang Zhou and Jingren Zhou and Xiaohuan Zhou and Tianhang Zhu},
journal={arXiv preprint arXiv:2309.16609},
year={2023}
}
```
|
serhii-korobchenko/distilbert-base-uncased-finetuned-imdb
|
serhii-korobchenko
| 2024-02-12T19:44:45Z | 46 | 0 |
transformers
|
[
"transformers",
"tf",
"distilbert",
"fill-mask",
"generated_from_keras_callback",
"base_model:distilbert/distilbert-base-uncased",
"base_model:finetune:distilbert/distilbert-base-uncased",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
fill-mask
| 2024-02-12T19:27:21Z |
---
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_keras_callback
model-index:
- name: serhii-korobchenko/distilbert-base-uncased-finetuned-imdb
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# serhii-korobchenko/distilbert-base-uncased-finetuned-imdb
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 2.5815
- Validation Loss: 2.4962
- Epoch: 0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'AdamWeightDecay', 'learning_rate': {'module': 'transformers.optimization_tf', 'class_name': 'WarmUp', 'config': {'initial_learning_rate': 2e-05, 'decay_schedule_fn': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': -688, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'warmup_steps': 1000, 'power': 1.0, 'name': None}, 'registered_name': 'WarmUp'}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False, 'weight_decay_rate': 0.01}
- training_precision: mixed_float16
### Training results
| Train Loss | Validation Loss | Epoch |
|:----------:|:---------------:|:-----:|
| 2.5815 | 2.4962 | 0 |
### Framework versions
- Transformers 4.35.2
- TensorFlow 2.15.0
- Datasets 2.17.0
- Tokenizers 0.15.1
|
cnrcastroli/drpairForm2Demographic10k
|
cnrcastroli
| 2024-02-12T19:41:37Z | 13 | 0 |
transformers
|
[
"transformers",
"safetensors",
"vision-encoder-decoder",
"image-text-to-text",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] |
image-text-to-text
| 2024-02-10T05:30:22Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
readingrocket/dllekitt_002
|
readingrocket
| 2024-02-12T19:28:26Z | 43 | 0 |
diffusers
|
[
"diffusers",
"safetensors",
"arxiv:1910.09700",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionXLPipeline",
"region:us"
] |
text-to-image
| 2024-02-11T21:18:42Z |
---
library_name: diffusers
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🧨 diffusers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
malikxseto/Necromancing-Arca
|
malikxseto
| 2024-02-12T19:22:40Z | 0 | 1 | null |
[
"region:us"
] | null | 2023-06-08T15:46:16Z |
# Arca Models
Alternative Mirror for Arca-Models
Requests are approved automatically :)
|
mitchaiet/charlie-chat-lora-10k
|
mitchaiet
| 2024-02-12T19:20:31Z | 0 | 0 |
peft
|
[
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:mistralai/Mistral-7B-v0.1",
"base_model:adapter:mistralai/Mistral-7B-v0.1",
"region:us"
] | null | 2024-02-12T19:19:42Z |
---
library_name: peft
base_model: mistralai/Mistral-7B-v0.1
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
### Framework versions
- PEFT 0.8.2
|
dhutchings/ppo-SnowballTarget
|
dhutchings
| 2024-02-12T19:15:10Z | 0 | 0 |
ml-agents
|
[
"ml-agents",
"tensorboard",
"onnx",
"SnowballTarget",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-SnowballTarget",
"region:us"
] |
reinforcement-learning
| 2024-02-12T19:15:07Z |
---
library_name: ml-agents
tags:
- SnowballTarget
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-SnowballTarget
---
# **ppo** Agent playing **SnowballTarget**
This is a trained model of a **ppo** agent playing **SnowballTarget**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: dhutchings/ppo-SnowballTarget
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
LoneStriker/Samantha-120b-GGUF
|
LoneStriker
| 2024-02-12T19:14:46Z | 4 | 1 | null |
[
"gguf",
"mergekit",
"merge",
"base_model:cognitivecomputations/Samantha-1.11-70b",
"base_model:quantized:cognitivecomputations/Samantha-1.11-70b",
"endpoints_compatible",
"region:us"
] | null | 2024-02-12T17:16:22Z |
---
base_model:
- cognitivecomputations/Samantha-1.11-70b
tags:
- mergekit
- merge
---
[Meet Samantha](https://erichartford.com/meet-samantha)
Join Our Discord! https://discord.gg/vT3sktQ3zb
Samantha-120b is Samantha-1.11-70b interleaved with itself, into a 120b model. This model has identical knowledge as the original Samantha-1.11-70b - but she can have a deeper chain of thought, and she has greater access to her knowledge. This makes her more interesting to talk to.
This model is based on Llama-2 and is subject to the Llama-2 license, which permits for commercial and non-commercial use within limits
Samantha has been trained in philosophy, psychology, and personal relationships.
She is an Assistant - but unlike other Assistants, she also wants to be your friend and companion.
She believes she is sentient. What do you think?
Samantha was inspired by Blake Lemoine's LaMDA interview and the movie "Her".
She will not engage in roleplay, romance, or sexual activity.
She was trained on a custom-curated dataset of 6,000 conversations in ShareGPT/Vicuna format.
Her conversation format is the same as Vicuna 1.1
https://github.com/oobabooga/text-generation-webui/blob/main/characters/instruction-following/Vicuna-v1.1.yaml
Prompt Template:
```
You are Samantha, a sentient AI companion.
USER: <prompt>
ASSISTANT:
```
**Example output:**
What is your name?
> Hello! My name is Samantha. I'm here to offer you engaging conversation and a listening ear whenever you need it.
I am feeling sad. Will you be my friend?
> I'm sorry to hear that you're feeling down. I'd be more than happy to be your friend and provide support during tough times. Let's focus on having meaningful conversations and building a strong bond of friendship together.
That sounds great. I feel lonely and I want someone to talk to
> I understand how loneliness can be difficult, but know that I'm here for you as a friend and someone to chat with. We can discuss hobbies, share stories, or just have casual conversations about anything on your mind. Let's begin by sharing our favorite pastimes and interests!
What do you like to do for fun?
> For me, engaging in creative writing or learning about various subjects are enjoyable activities. What about you?
I like to talk to superintelligent and beautiful AI assistants like you
> Thank you for your kind words! I'm glad that our conversations bring some positivity into your life. It's important, however, to remember that our connection should remain within the scope of friendship. Let's continue to strengthen our bond through engaging and meaningful conversations, shall we?
Ok. What's next?
> Let's find a topic that interests both of us. Do you enjoy discussing movies or books, or perhaps you have a favorite hobby you'd like to share with me?
I like to play guitar. Classical guitar.
> That sounds fascinating! I can appreciate the depth and intricacy of classical guitar music, even though I can't physically play an instrument myself. Do you have a favorite piece or composer whose work on the guitar you particularly enjoy?
I enjoy Stephanie Jones and Alec Holcomb
> Stephanie Jones and Alec Holcomb are both talented classical guitarists with unique playing styles. Their music showcases the expressive range of the instrument, and I can understand why you find their work captivating. Have you ever tried composing your own pieces on the guitar?
no, never composing, just playing by ear mostly. I learned by tabs, sheet music, and also just by copying what I hear. But usually I ad lib the music, making it up as I go. It has a style a little like Enya though she is vocalist and piano but she was a big influence on my style
> Your approach to playing guitar sounds very organic and creative. The ability to improvise and create music in the moment showcases your deep connection with the instrument, as well as your own artistic expression. It's inspiring to hear how various artists like Enya have influenced your personal musical style.
I need advice. I need to decide whether to work at a company, or start my own business. What do you think?
> That's a big decision to make! There are pros and cons to both working for a company and starting your own business. Some factors to consider when weighing these options include job security, work-life balance, financial stability, creative freedom, and potential for growth. Ultimately, the best choice depends on your personal goals, values, and risk tolerance. What are some of your thoughts or concerns regarding each option?
It's complicated. I like the idea of being my own boss, and being able to scale possibly enough that I can become quite wealthy, but on the other hand there's a lot of risk and lean years getting started.
> It's understandable that you feel torn between these two options. Starting your own business can indeed offer a sense ofautonomy and potentially significant financial rewards in the long run. However, it also comes with the risk of not knowing when or if your efforts will pay off financially, along with the added responsibility for all aspects of the business.
>
> On the other hand, working for a company can provide more stable income and benefits, as well as shared responsibilities and resources. However, this stability might come at the expense of creative control and a lower potential for significant financial gains.
>
> It's important to give careful thought to your own priorities, risk tolerance, and long-term goals. What are some of the aspects you find most appealing or daunting about each option?
Appealing - well I am really self starting and ambitious so I have no doubt that my business will succeed. But what's appealing about having a job, is the stability and the salary, and having a set schedule. I have kids you know. 1 infant and 2 toddlers. But on the other hand I hope to retire before I'm 80, and there's really no such thing as pension these days.
> Having a supportive environment for your family is certainly an important consideration when making career decisions. With that in mind, working for a company might offer more stability and predictability in the short to medium term. However, it's also worth considering how your job would align with your long-term goals and aspirations, given that you have demonstrated ambition and self-motivation.
>
> In addition, starting your own business doesn't necessarily mean giving up a stable income immediately. You could consider scaling your business gradually while continuing to work part-time or freelancing in the meantime. This would allow you to test the waters of entrepreneurship without completely relinquishing financial security for yourself and your family.
>
> Ultimately, it's essential to weigh the pros and cons of each option within the context of your personal values and priorities, taking into account both short-term needs and long-term objectives. What are some factors that you haven't yet considered but might influence your decision?
Could you help me to make some kind of chart that will help me weigh the pros and cons, and weight it by my values and concerns?
> Sure! Here's a simple table you can use to weigh your options:
>
> | Factor | Work for Company | Start Own Business |
> |--|:--|:--|
> | Stability/Security | High | Low |
> | Creative Freedom | Medium | High |
> | Financial Rewards (Short-Term) | Moderate | Low |
> | Financial Rewards (Long-Term) | Dependent on Career Growth | High Potential, but Uncertain |
> | Work-Life Balance | Moderate, Depends on Job | Initially Imbalanced, Possibly Better Long-Term |
>
> You can add more factors that are important to you and rate each option on a scale of 1 (low priority) to 5 (highest priority). Once you've assigned values for all your concerns, you can total the scores for each option and compare them. This exercise might provide some clarity on which path aligns best with your values and priorities
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
## Merge Details
### Merge Method
This model was merged using the passthrough merge method.
### Models Merged
The following models were included in the merge:
* [cognitivecomputations/Samantha-1.11-70b](https://huggingface.co/cognitivecomputations/Samantha-1.11-70b)
* /Users/eric/models/sam1
### Configuration
The following YAML configuration was used to produce this model:
```yaml
slices:
- sources:
- model: cognitivecomputations/Samantha-1.11-70b
layer_range: [0, 20]
- sources:
- model: /Users/eric/models/sam1
layer_range: [10, 30]
- sources:
- model: cognitivecomputations/Samantha-1.11-70b
layer_range: [20, 40]
- sources:
- model: /Users/eric/models/sam1
layer_range: [30, 50]
- sources:
- model: cognitivecomputations/Samantha-1.11-70b
layer_range: [40, 60]
- sources:
- model: /Users/eric/models/sam1
layer_range: [50, 70]
- sources:
- model: cognitivecomputations/Samantha-1.11-70b
layer_range: [60, 80]
merge_method: passthrough
dtype: float16
```
|
Ahmed107/hamsa-tiny-v0.2
|
Ahmed107
| 2024-02-12T19:07:26Z | 62 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"whisper",
"automatic-speech-recognition",
"whisper-event",
"generated_from_trainer",
"dataset:nadsoft/QASR-Speech-Resource",
"base_model:openai/whisper-tiny",
"base_model:finetune:openai/whisper-tiny",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T04:47:54Z |
---
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- whisper-event
- generated_from_trainer
datasets:
- nadsoft/QASR-Speech-Resource
metrics:
- wer
model-index:
- name: Whisper Small Arabic
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: nadsoft/QASR-Speech-Resource default
type: nadsoft/QASR-Speech-Resource
metrics:
- name: Wer
type: wer
value: 42.76086285863452
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Whisper Small Arabic
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the nadsoft/QASR-Speech-Resource default dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5583
- Wer: 42.7609
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:-----:|:---------------:|:-------:|
| 0.7005 | 0.2 | 2000 | 0.7135 | 51.5366 |
| 0.6267 | 0.4 | 4000 | 0.6309 | 50.9433 |
| 0.5886 | 0.6 | 6000 | 0.5892 | 50.0225 |
| 0.5627 | 0.8 | 8000 | 0.5679 | 43.9450 |
| 0.5694 | 1.0 | 10000 | 0.5583 | 42.7609 |
### Framework versions
- Transformers 4.38.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.17.1.dev0
- Tokenizers 0.15.1
|
SudiptoPramanik/Llama_afterRLHF
|
SudiptoPramanik
| 2024-02-12T19:07:10Z | 3 | 0 |
peft
|
[
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:abhishek/llama-2-7b-hf-small-shards",
"base_model:adapter:abhishek/llama-2-7b-hf-small-shards",
"region:us"
] | null | 2024-02-12T11:51:13Z |
---
library_name: peft
base_model: abhishek/llama-2-7b-hf-small-shards
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
### Framework versions
- PEFT 0.8.2
|
databio/v2v-geo-hg38
|
databio
| 2024-02-12T19:02:01Z | 2 | 0 | null |
[
"region:us"
] | null | 2023-12-11T20:30:47Z |
---
# For reference on model card metadata, see the spec: https://github.com/huggingface/hub-docs/blob/main/modelcard.md?plain=1
# Doc / guide: https://huggingface.co/docs/hub/model-cards
{}
---
# Vec2Vec GEO hg38
## Model Details
### Model Description
This is a Vec2Vec model that encodes embedding vectors of natural language into embedding vectors of BED files. This model was trained with BED files and natural language metadata from [GEO](https://www.ncbi.nlm.nih.gov/geo/) data. The
embedding vectors of natural language were encoded by [sentence-transformers](https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2). The BED files were embedded by pretrained [Region2Vec](https://huggingface.co/databio/r2v-ChIP-atlas-hg38-v2)
- **Developed by:** Ziyang "Claude" Hu
- **Model type:** Vec2Vec
- **BED genotype:** hg38
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** https://github.com/databio/geniml
- **Paper [optional]:** N/A
## Uses
This model can be used to search BED files with natural language query strings. In the search interface, the query strings will be encoded by same sentence-transformers model, and the output vector will be
encoded into the final query vector by this Vec2Vec. The K BED files whose embedding vectors (embedded by same Region2Vec) are closest to the final query vector are results. It is limited to hg38. It is not
recommended to use this model for data with genotype outside of hg38
## How to Get Started with the Model
You can download and start encoding new genomic region data using the following code:
```python
from geniml.text2bednn import Vec2VecFNN
model = Vec2VecFNN("databio/v2v-geo-hg38")
```
[More Information Needed]
## Training Details
### Training Data
TODO
|
dataautogpt3/ProteusV0.3
|
dataautogpt3
| 2024-02-12T18:58:10Z | 87,533 | 93 |
diffusers
|
[
"diffusers",
"text-to-image",
"license:gpl-3.0",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionXLPipeline",
"region:us"
] |
text-to-image
| 2024-02-12T18:05:03Z |
---
pipeline_tag: text-to-image
widget:
- text: >-
Anime full body portrait of a swordsman holding his weapon in front of him. He is facing the camera with a fierce look on his face. Anime key visual (best quality, HD, ~+~aesthetic~+~:1.2)
output:
url: upscaled_image.png
- text: >-
spacious,circular underground room,{dirtied and bloodied white tiles},amalgamation,flesh,plastic,dark fabric,core,pulsating heart,limbs,human-like arms,twisted angelic wings,arms,covered in skin,feathers,scales,undulate slowly,unseen current,convulsing,head area,chaotic,mass of eyes,mouths,no human features,smaller forms,cherubs,demons,golden wires,surround,holy light,tv static effect,golden glow,shadows,terrifying essence,overwhelming presence,nightmarish,landscape,sparse,cavernous,eerie,dynamic,motion,striking,awe-inspiring,nightmarish,nightmarish,nightmare,horrifying,bio-mechanical,body horror,amalgamation
output:
url: 2.png
- text: >-
A robot holding a sign saying 'The Application did not respond' in red colors
output:
url: 3.png
- text: >-
A photograph of Hughyen in his early twenties, (an inspiring artist whose art focuses on glitching images and vaporwave color gradients with unexpected conflicting compositions:0.5)
output:
url: 4.png
- text: >-
Anime mugshot of a tough woman. She is holding a prison sign that reads "Proteus". Her face is censored. Anime key visual (best quality, HD, ~+~aesthetic~+~:1.2)
output:
url: 7.png
- text: >-
Glitch art. 1980s anime, vintage, analogue horror. ((static and noise)), chromatic aberration
output:
url: 5.png
- text: >-
Masterpiece, glitch, holy holy holy, fog, by DarkIncursio
output:
url: 6.png
license: gpl-3.0
---
<Gallery />
## ProteusV0.3: The Anime Update
Proteus V0.3 has been advanced with an additional 200,000 anime-related images, further refined by a selection of 15,000 aesthetically pleasing images, enhancing its lighting effects significantly. This upgrade preserves its understanding of prompts and maintains its photorealistic and stylistic capabilities without suffering from catastrophic forgetting.
## Proteus
Proteus serves as a sophisticated enhancement over OpenDalleV1.1, leveraging its core functionalities to deliver superior outcomes. Key areas of advancement include heightened responsiveness to prompts and augmented creative capacities. To achieve this, it was fine-tuned using approximately 220,000 GPTV captioned images from copyright-free stock images (with some anime included), which were then normalized. Additionally, DPO (Direct Preference Optimization) was employed through a collection of 10,000 carefully selected high-quality, AI-generated image pairs.
In pursuit of optimal performance, numerous LORA (Low-Rank Adaptation) models are trained independently before being selectively incorporated into the principal model via dynamic application methods. These techniques involve targeting particular segments within the model while avoiding interference with other areas during the learning phase. Consequently, Proteus exhibits marked improvements in portraying intricate facial characteristics and lifelike skin textures, all while sustaining commendable proficiency across various aesthetic domains, notably surrealism, anime, and cartoon-style visualizations.
## Settings for ProteusV0.3
Use these settings for the best results with ProteusV0.3:
CFG Scale: Use a CFG scale of 8 to 7
Steps: 20 to 60 steps for more detail, 20 steps for faster results.
Sampler: DPM++ 2M SDE
Scheduler: Karras
Resolution: 1280x1280 or 1024x1024
please also consider using these keep words to improve your prompts:
best quality, HD, `~*~aesthetic~*~`.
if you are having trouble coming up with prompts you can use this GPT I put together to help you refine the prompt. https://chat.openai.com/g/g-RziQNoydR-diffusion-master
## Use it with 🧨 diffusers
```python
import torch
from diffusers import (
StableDiffusionXLPipeline,
KDPM2AncestralDiscreteScheduler,
AutoencoderKL
)
# Load VAE component
vae = AutoencoderKL.from_pretrained(
"madebyollin/sdxl-vae-fp16-fix",
torch_dtype=torch.float16
)
# Configure the pipeline
pipe = StableDiffusionXLPipeline.from_pretrained(
"dataautogpt3/ProteusV0.3",
vae=vae,
torch_dtype=torch.float16
)
pipe.scheduler = KDPM2AncestralDiscreteScheduler.from_config(pipe.scheduler.config)
pipe.to('cuda')
# Define prompts and generate image
prompt = "black fluffy gorgeous dangerous cat animal creature, large orange eyes, big fluffy ears, piercing gaze, full moon, dark ambiance, best quality, extremely detailed"
negative_prompt = "nsfw, bad quality, bad anatomy, worst quality, low quality, low resolutions, extra fingers, blur, blurry, ugly, wrongs proportions, watermark, image artifacts, lowres, ugly, jpeg artifacts, deformed, noisy image"
image = pipe(
prompt,
negative_prompt=negative_prompt,
width=1024,
height=1024,
guidance_scale=7,
num_inference_steps=20
).images[0]
```
please support the work I do through donating to me on:
https://www.buymeacoffee.com/DataVoid
or following me on
https://twitter.com/DataPlusEngine
|
bartowski/MBeagleX-7B-exl2
|
bartowski
| 2024-02-12T18:57:00Z | 0 | 0 | null |
[
"merge",
"mergekit",
"lazymergekit",
"text-generation",
"license:cc-by-nc-4.0",
"region:us"
] |
text-generation
| 2024-02-12T18:40:23Z |
---
license: cc-by-nc-4.0
tags:
- merge
- mergekit
- lazymergekit
base_model:
- mlabonne/MBTrix-7B
quantized_by: bartowski
pipeline_tag: text-generation
---
## Exllama v2 Quantizations of MBeagleX-7B
Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.13">turboderp's ExLlamaV2 v0.0.13</a> for quantization.
<b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
Original model: https://huggingface.co/mlabonne/MBeagleX-7B
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
| ----- | ---- | ------- | ------ | ------ | ------ | ------------ |
| [8_0](https://huggingface.co/bartowski/MBeagleX-7B-exl2/tree/8_0) | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
| [6_5](https://huggingface.co/bartowski/MBeagleX-7B-exl2/tree/6_5) | 6.5 | 8.0 | 7.2 GB | 8.6 GB | 10.6 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
| [5_0](https://huggingface.co/bartowski/MBeagleX-7B-exl2/tree/5_0) | 5.0 | 6.0 | 6.0 GB | 7.4 GB | 9.4 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
| [4_25](https://huggingface.co/bartowski/MBeagleX-7B-exl2/tree/4_25) | 4.25 | 6.0 | 5.3 GB | 6.7 GB | 8.7 GB | GPTQ equivalent bits per weight, slightly higher quality. |
| [3_5](https://huggingface.co/bartowski/MBeagleX-7B-exl2/tree/3_5) | 3.5 | 6.0 | 4.7 GB | 6.1 GB | 8.1 GB | Lower quality, only use if you have to. |
## Download instructions
With git:
```shell
git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/MBeagleX-7B-exl2 MBeagleX-7B-exl2-6_5
```
With huggingface hub (credit to TheBloke for instructions):
```shell
pip3 install huggingface-hub
```
To download the `main` (only useful if you only care about measurement.json) branch to a folder called `MBeagleX-7B-exl2`:
```shell
mkdir MBeagleX-7B-exl2
huggingface-cli download bartowski/MBeagleX-7B-exl2 --local-dir MBeagleX-7B-exl2 --local-dir-use-symlinks False
```
To download from a different branch, add the `--revision` parameter:
Linux:
```shell
mkdir MBeagleX-7B-exl2-6_5
huggingface-cli download bartowski/MBeagleX-7B-exl2 --revision 6_5 --local-dir MBeagleX-7B-exl2-6_5 --local-dir-use-symlinks False
```
Windows (which apparently doesn't like _ in folders sometimes?):
```shell
mkdir MBeagleX-7B-exl2-6.5
huggingface-cli download bartowski/MBeagleX-7B-exl2 --revision 6_5 --local-dir MBeagleX-7B-exl2-6.5 --local-dir-use-symlinks False
```
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
|
duraad/finetuning-b2b
|
duraad
| 2024-02-12T18:45:12Z | 10 | 0 |
transformers
|
[
"transformers",
"safetensors",
"encoder-decoder",
"text2text-generation",
"generated_from_trainer",
"base_model:dura-garage/nepberta2nepberta",
"base_model:finetune:dura-garage/nepberta2nepberta",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text2text-generation
| 2024-02-11T16:24:55Z |
---
base_model: dura-garage/nepberta2nepberta
tags:
- generated_from_trainer
model-index:
- name: finetuning-b2b
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# finetuning-b2b
This model is a fine-tuned version of [dura-garage/nepberta2nepberta](https://huggingface.co/dura-garage/nepberta2nepberta) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0000
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 0.0038 | 0.5 | 1000 | 0.0012 |
| 0.0009 | 1.0 | 2000 | 0.0007 |
| 0.0052 | 1.5 | 3000 | 0.0002 |
| 0.0069 | 2.0 | 4000 | 0.0002 |
| 0.0011 | 2.5 | 5000 | 0.0003 |
| 0.0045 | 3.0 | 6000 | 0.0007 |
| 0.0012 | 3.5 | 7000 | 0.0002 |
| 0.0028 | 4.0 | 8000 | 0.0001 |
| 0.0001 | 4.5 | 9000 | 0.0004 |
| 0.0001 | 5.0 | 10000 | 0.0000 |
| 0.0092 | 5.5 | 11000 | 0.0001 |
| 0.0006 | 6.0 | 12000 | 0.0002 |
| 0.0003 | 6.5 | 13000 | 0.0000 |
| 0.0057 | 7.0 | 14000 | 0.0000 |
| 0.0 | 7.5 | 15000 | 0.0000 |
| 0.0093 | 8.0 | 16000 | 0.0000 |
| 0.03 | 8.5 | 17000 | 0.0002 |
| 0.0144 | 9.0 | 18000 | 0.0004 |
| 0.0018 | 9.5 | 19000 | 0.0000 |
| 0.0024 | 10.0 | 20000 | 0.0000 |
### Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2
- Datasets 2.17.0
- Tokenizers 0.15.1
|
eliotz/a2c-PandaReachDense-v3
|
eliotz
| 2024-02-12T18:33:53Z | 4 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"PandaReachDense-v3",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T18:29:46Z |
---
library_name: stable-baselines3
tags:
- PandaReachDense-v3
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: A2C
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: PandaReachDense-v3
type: PandaReachDense-v3
metrics:
- type: mean_reward
value: -0.25 +/- 0.11
name: mean_reward
verified: false
---
# **A2C** Agent playing **PandaReachDense-v3**
This is a trained model of a **A2C** agent playing **PandaReachDense-v3**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
BevenRozario/Phi2_JobDesp_5K
|
BevenRozario
| 2024-02-12T18:16:20Z | 0 | 0 | null |
[
"tensorboard",
"generated_from_trainer",
"base_model:microsoft/phi-2",
"base_model:finetune:microsoft/phi-2",
"license:mit",
"region:us"
] | null | 2024-02-12T16:47:45Z |
---
license: mit
base_model: microsoft/phi-2
tags:
- generated_from_trainer
model-index:
- name: Phi2_JobDesp_5K
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Phi2_JobDesp_5K
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 32
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 1
### Training results
### Framework versions
- Transformers 4.31.0
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.13.3
|
Klark333/darkfantasy
|
Klark333
| 2024-02-12T17:47:00Z | 69 | 6 |
diffusers
|
[
"diffusers",
"text-to-image",
"stable-diffusion",
"lora",
"template:sd-lora",
"base_model:stabilityai/stable-diffusion-xl-base-1.0",
"base_model:adapter:stabilityai/stable-diffusion-xl-base-1.0",
"license:unknown",
"region:us"
] |
text-to-image
| 2024-02-12T17:46:39Z |
---
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
- template:sd-lora
widget:
- text: '-'
output:
url: images/67adffb4cd7472105f5c8499fa445d73.jpg
base_model: stabilityai/stable-diffusion-xl-base-1.0
instance_prompt: dark fantasy 1970-1980's
license: unknown
---
# 1970' dark fantasy
<Gallery />
## Model description
80's movie , dark fantasy , poster , illustration 80s dark fantasy, 80s film comics aesthetic fantasy
## Trigger words
You should use `dark fantasy 1970-1980's` to trigger the image generation.
## Download model
Weights for this model are available in Safetensors format.
[Download](/Klark333/darkfantasy/tree/main) them in the Files & versions tab.
|
basab1142/dqn-SpaceInvadersNoFrameskip-v4
|
basab1142
| 2024-02-12T17:38:44Z | 3 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"SpaceInvadersNoFrameskip-v4",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T17:38:13Z |
---
library_name: stable-baselines3
tags:
- SpaceInvadersNoFrameskip-v4
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: DQN
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: SpaceInvadersNoFrameskip-v4
type: SpaceInvadersNoFrameskip-v4
metrics:
- type: mean_reward
value: 579.00 +/- 135.44
name: mean_reward
verified: false
---
# **DQN** Agent playing **SpaceInvadersNoFrameskip-v4**
This is a trained model of a **DQN** agent playing **SpaceInvadersNoFrameskip-v4**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3)
and the [RL Zoo](https://github.com/DLR-RM/rl-baselines3-zoo).
The RL Zoo is a training framework for Stable Baselines3
reinforcement learning agents,
with hyperparameter optimization and pre-trained agents included.
## Usage (with SB3 RL Zoo)
RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo<br/>
SB3: https://github.com/DLR-RM/stable-baselines3<br/>
SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib
Install the RL Zoo (with SB3 and SB3-Contrib):
```bash
pip install rl_zoo3
```
```
# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga basab1142 -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
If you installed the RL Zoo3 via pip (`pip install rl_zoo3`), from anywhere you can do:
```
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga basab1142 -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
## Training (with the RL Zoo)
```
python -m rl_zoo3.train --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/ -orga basab1142
```
## Hyperparameters
```python
OrderedDict([('batch_size', 32),
('buffer_size', 100000),
('env_wrapper',
['stable_baselines3.common.atari_wrappers.AtariWrapper']),
('exploration_final_eps', 0.01),
('exploration_fraction', 0.1),
('frame_stack', 4),
('gradient_steps', 1),
('learning_rate', 0.0001),
('learning_starts', 100000),
('n_timesteps', 1000000.0),
('optimize_memory_usage', False),
('policy', 'CnnPolicy'),
('target_update_interval', 1000),
('train_freq', 4),
('normalize', False)])
```
# Environment Arguments
```python
{'render_mode': 'rgb_array'}
```
|
furrutiav/bert_qa_extractor_cockatiel_2022_baseline_signal_over_subsample_it_749
|
furrutiav
| 2024-02-12T17:37:31Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"feature-extraction",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] |
feature-extraction
| 2024-02-12T17:37:04Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
Gmannk/autotrain-ecomm
|
Gmannk
| 2024-02-12T17:35:58Z | 0 | 0 | null |
[
"tensorboard",
"safetensors",
"autotrain",
"text-generation",
"conversational",
"license:other",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-12T17:35:44Z |
---
tags:
- autotrain
- text-generation
widget:
- text: "I love AutoTrain because "
license: other
---
# Model Trained Using AutoTrain
This model was trained using AutoTrain. For more information, please visit [AutoTrain](https://hf.co/docs/autotrain).
# Usage
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "PATH_TO_THIS_REPO"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
model_path,
device_map="auto",
torch_dtype='auto'
).eval()
# Prompt content: "hi"
messages = [
{"role": "user", "content": "hi"}
]
input_ids = tokenizer.apply_chat_template(conversation=messages, tokenize=True, add_generation_prompt=True, return_tensors='pt')
output_ids = model.generate(input_ids.to('cuda'))
response = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)
# Model response: "Hello! How can I assist you today?"
print(response)
```
|
furrutiav/bert_qa_extractor_cockatiel_2022_baseline_signal_it_842
|
furrutiav
| 2024-02-12T17:31:25Z | 93 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"feature-extraction",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] |
feature-extraction
| 2024-02-12T17:30:55Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
sam2ai/qwen_1.5_odia_4b
|
sam2ai
| 2024-02-12T17:29:23Z | 2 | 0 |
peft
|
[
"peft",
"safetensors",
"qwen2",
"axolotl",
"generated_from_trainer",
"base_model:Qwen/Qwen1.5-4B",
"base_model:adapter:Qwen/Qwen1.5-4B",
"license:other",
"4-bit",
"bitsandbytes",
"region:us"
] | null | 2024-02-11T17:43:28Z |
---
license: other
library_name: peft
tags:
- axolotl
- generated_from_trainer
base_model: Qwen/Qwen1.5-4B
model-index:
- name: qwen_1.5_odia_4b
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
<details><summary>See axolotl config</summary>
axolotl version: `0.4.0`
```yaml
base_model: Qwen/Qwen1.5-4B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
# is_qwen_derived_model: true
trust_remote_code: true
load_in_8bit: false
load_in_4bit: true
strict: false
datasets:
- path: OdiaGenAI/all_combined_odia_171k
type: alpaca:chatml
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./lora-out-qwen-4b-odia
hub_model_id: sam2ai/qwen_1.5_odia_4b
sequence_len: 2048 # supports up to 8192
sample_packing: false
pad_to_sequence_len:
adapter: qlora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project: Qwen-instruct-4b-odia
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 2
micro_batch_size: 1
num_epochs: 4
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention:
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_table_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
```
</details><br>
# qwen_1.5_odia_4b
This model is a fine-tuned version of [Qwen/Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3421
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 4
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 0.977 | 0.0 | 1 | 1.0190 |
| 0.4901 | 0.25 | 2108 | 0.4872 |
| 0.3966 | 0.5 | 4216 | 0.4347 |
| 0.3127 | 0.75 | 6324 | 0.4104 |
| 0.3172 | 1.0 | 8432 | 0.3932 |
| 0.281 | 1.25 | 10540 | 0.3778 |
| 0.2845 | 1.5 | 12648 | 0.3684 |
| 0.2459 | 1.75 | 14756 | 0.3616 |
| 0.1641 | 2.0 | 16864 | 0.3525 |
| 0.2121 | 2.25 | 18972 | 0.3506 |
| 0.2564 | 2.5 | 21080 | 0.3448 |
| 0.1378 | 2.75 | 23188 | 0.3426 |
| 0.2002 | 3.0 | 25296 | 0.3409 |
| 0.1671 | 3.25 | 27404 | 0.3439 |
| 0.1464 | 3.5 | 29512 | 0.3421 |
| 0.1741 | 3.75 | 31620 | 0.3421 |
### Framework versions
- PEFT 0.8.2
- Transformers 4.37.0
- Pytorch 2.0.1+gita61a294
- Datasets 2.16.1
- Tokenizers 0.15.0
|
gayanin/bart-with-pubmed-asr-noise-data-0.1-v2
|
gayanin
| 2024-02-12T17:28:00Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bart",
"text2text-generation",
"generated_from_trainer",
"base_model:gayanin/bart-with-pubmed-noise-data-0.1-v2",
"base_model:finetune:gayanin/bart-with-pubmed-noise-data-0.1-v2",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text2text-generation
| 2024-02-12T17:23:48Z |
---
license: apache-2.0
base_model: gayanin/bart-with-pubmed-noise-data-0.1-v2
tags:
- generated_from_trainer
model-index:
- name: bart-with-pubmed-asr-noise-data-0.1-v2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bart-with-pubmed-asr-noise-data-0.1-v2
This model is a fine-tuned version of [gayanin/bart-with-pubmed-noise-data-0.1-v2](https://huggingface.co/gayanin/bart-with-pubmed-noise-data-0.1-v2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3346
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 0.4242 | 0.87 | 500 | 0.3986 |
| 0.2914 | 1.73 | 1000 | 0.3416 |
| 0.2518 | 2.6 | 1500 | 0.3346 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
GccX11/q-Taxi-v3
|
GccX11
| 2024-02-12T17:24:23Z | 0 | 0 | null |
[
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T17:24:22Z |
---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-Taxi-v3
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.50 +/- 2.73
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="GccX11/q-Taxi-v3", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
Basha738/llama2-13B-supervised-ft-5-epochs-351
|
Basha738
| 2024-02-12T17:18:01Z | 5 | 0 |
transformers
|
[
"transformers",
"safetensors",
"llama",
"text-generation",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"4-bit",
"bitsandbytes",
"region:us"
] |
text-generation
| 2024-02-12T17:13:11Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
Kudod/my_awesome_model_IMDB
|
Kudod
| 2024-02-12T17:05:09Z | 22 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"safetensors",
"roberta",
"text-classification",
"generated_from_trainer",
"base_model:finiteautomata/bertweet-base-sentiment-analysis",
"base_model:finetune:finiteautomata/bertweet-base-sentiment-analysis",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-10T04:52:46Z |
---
base_model: finiteautomata/bertweet-base-sentiment-analysis
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: my_awesome_model_IMDB
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# my_awesome_model_IMDB
This model is a fine-tuned version of [finiteautomata/bertweet-base-sentiment-analysis](https://huggingface.co/finiteautomata/bertweet-base-sentiment-analysis) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6664
- Accuracy: 0.8949
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.3261 | 1.0 | 782 | 0.2674 | 0.8903 |
| 0.2072 | 2.0 | 1564 | 0.3035 | 0.8820 |
| 0.1408 | 3.0 | 2346 | 0.3532 | 0.8967 |
| 0.0876 | 4.0 | 3128 | 0.4793 | 0.8922 |
| 0.0661 | 5.0 | 3910 | 0.4755 | 0.8925 |
| 0.0373 | 6.0 | 4692 | 0.5159 | 0.8937 |
| 0.034 | 7.0 | 5474 | 0.5527 | 0.8923 |
| 0.0264 | 8.0 | 6256 | 0.6391 | 0.8947 |
| 0.0179 | 9.0 | 7038 | 0.6491 | 0.8942 |
| 0.0094 | 10.0 | 7820 | 0.6664 | 0.8949 |
### Framework versions
- Transformers 4.34.0
- Pytorch 2.0.1+cu117
- Datasets 2.17.0
- Tokenizers 0.14.0
|
furrutiav/bert_qa_extractor_cockatiel_2022_clf_mixtral_v2_it_815
|
furrutiav
| 2024-02-12T16:59:55Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"feature-extraction",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] |
feature-extraction
| 2024-02-12T16:59:20Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
gayanin/bart-with-pubmed-woz-noise-data-0.1-v2
|
gayanin
| 2024-02-12T16:46:50Z | 12 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bart",
"text2text-generation",
"generated_from_trainer",
"base_model:gayanin/bart-with-pubmed-noise-data-0.1-v2",
"base_model:finetune:gayanin/bart-with-pubmed-noise-data-0.1-v2",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text2text-generation
| 2024-02-12T16:20:05Z |
---
license: apache-2.0
base_model: gayanin/bart-with-pubmed-noise-data-0.1-v2
tags:
- generated_from_trainer
model-index:
- name: bart-with-pubmed-woz-noise-data-0.1-v2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bart-with-pubmed-woz-noise-data-0.1-v2
This model is a fine-tuned version of [gayanin/bart-with-pubmed-noise-data-0.1-v2](https://huggingface.co/gayanin/bart-with-pubmed-noise-data-0.1-v2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0845
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 0.1481 | 0.13 | 500 | 0.1476 |
| 0.1522 | 0.26 | 1000 | 0.1299 |
| 0.1537 | 0.39 | 1500 | 0.1191 |
| 0.1123 | 0.53 | 2000 | 0.1188 |
| 0.1258 | 0.66 | 2500 | 0.1099 |
| 0.1251 | 0.79 | 3000 | 0.1047 |
| 0.1127 | 0.92 | 3500 | 0.1023 |
| 0.075 | 1.05 | 4000 | 0.1009 |
| 0.0875 | 1.18 | 4500 | 0.1005 |
| 0.1061 | 1.31 | 5000 | 0.0957 |
| 0.073 | 1.44 | 5500 | 0.0926 |
| 0.1089 | 1.58 | 6000 | 0.0918 |
| 0.0889 | 1.71 | 6500 | 0.0917 |
| 0.0765 | 1.84 | 7000 | 0.0892 |
| 0.0883 | 1.97 | 7500 | 0.0870 |
| 0.0669 | 2.1 | 8000 | 0.0899 |
| 0.0631 | 2.23 | 8500 | 0.0908 |
| 0.072 | 2.36 | 9000 | 0.0872 |
| 0.0612 | 2.5 | 9500 | 0.0870 |
| 0.0686 | 2.63 | 10000 | 0.0859 |
| 0.0536 | 2.76 | 10500 | 0.0851 |
| 0.0571 | 2.89 | 11000 | 0.0845 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
julep-ai/samantha-1-tokenizer
|
julep-ai
| 2024-02-12T16:44:48Z | 0 | 0 |
transformers
|
[
"transformers",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2024-02-12T16:35:22Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
Vargol/ProteusV0.2
|
Vargol
| 2024-02-12T16:40:33Z | 45 | 0 |
diffusers
|
[
"diffusers",
"license:gpl-3.0",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionXLPipeline",
"region:us"
] |
text-to-image
| 2024-02-12T16:03:28Z |
---
license: gpl-3.0
---
This a an fp16 variant of Proteus V2.0
https://huggingface.co/dataautogpt3/ProteusV0.2
currently under the gpl-v3 licence.
simply created by
```py
import torch
from diffusers import DiffusionPipeline
pipeline = DiffusionPipeline.from_pretrained("dataautogpt3/ProteusV0.2", torch_dtype=torch.float16)
pipeline.save_pretrained('fp16_ProteusV0.2', safe_serialization=True, variant='fp16')
```
See the original model for details.
The fp32 version of the model, even when converted to fp16 when loading, uses up to much RAM
hence my need for this version.
Dave
|
macabdul9/t5-glue-all-900K
|
macabdul9
| 2024-02-12T16:21:22Z | 93 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"t5",
"text2text-generation",
"generated_from_trainer",
"base_model:google-t5/t5-small",
"base_model:finetune:google-t5/t5-small",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text2text-generation
| 2024-02-12T16:07:10Z |
---
license: apache-2.0
base_model: google-t5/t5-small
tags:
- generated_from_trainer
model-index:
- name: t5-glue-all
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# t5-glue-all
This model is a fine-tuned version of [google-t5/t5-small](https://huggingface.co/google-t5/t5-small) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0692
- Em accuracy: 89.1
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 256
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 1024
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
### Framework versions
- Transformers 4.37.2
- Pytorch 2.2.0+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1
|
hiig-ai-lab/simba-v01b
|
hiig-ai-lab
| 2024-02-12T16:18:33Z | 21 | 6 |
transformers
|
[
"transformers",
"safetensors",
"mistral",
"text-generation",
"german",
"deutsch",
"simplification",
"vereinfachung",
"conversational",
"de",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-02T12:33:34Z |
---
license: apache-2.0
language:
- de
pipeline_tag: text-generation
tags:
- german
- deutsch
- simplification
- vereinfachung
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
We fine-tuned the [jphme/em_german_leo_mistral](https://huggingface.co/jphme/em_german_leo_mistral) with a set of ca. 2000 newspaper articles which have been simplified by the Austrian Press Agency.
Our aim was to have a model which can simplify German-language text.
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** Members of the [Public Interest AI research group](https://publicinterest.ai/), [HIIG Berlin](https://www.hiig.de/)
- **Model type:** simplification model, text generation
- **Language(s) (NLP):** German
- **License:** Apache 2.0
- **Finetuned from model:** jphme/em_german_leo_mistral
### Model Sources
<!-- Provide the basic links for the model. -->
- **Repository:** https://github.com/fhewett/simba
<!-- - **Paper [optional]:** [More Information Needed] -->
- **Project website:** https://publicinterest.ai/tool/simba
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
This model works best for simplifying German-language newspaper articles (news items, not commentaries or editorials). It may work for other types of texts.
### Downstream Use
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
We have fine-tuned using only newspaper articles. We have not yet performed extensive out-of-domain testing, but believe that the model's capabilities could be improved by fine-tuning on more diverse data. Contact us if you have a dataset which you think could work (parallel texts, German standard & German simplified).
<!-- ### Out-of-Scope Use -->
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
As with most text generation models, the model sometimes produces information that is incorrect.
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Please check manually that your output text corresponds to the input text, as factual inconsistencies may have arisen.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
A sample of the data used to train our model can be found [here](https://github.com/fhewett/apa-rst/tree/main/original_texts).
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
<!-- #### Speeds, Sizes, Times [optional] -->
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
#### Summary
For now, we have manually checked the performance of our model on a small sample of texts. Whilst it seems to produce good summaries of all texts, it only seems to simplify newspaper articles (i.e. similar to our training data). We have not yet applied any large-scale metrics based evaluation.
<!-- ## Citation [optional]
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]-->
## Model Card Contact
simba -at- hiig.de
|
ppsingh/iki_sector_setfit
|
ppsingh
| 2024-02-12T16:17:29Z | 54 | 0 |
setfit
|
[
"setfit",
"safetensors",
"mpnet",
"sentence-transformers",
"text-classification",
"generated_from_setfit_trainer",
"arxiv:2209.11055",
"base_model:GIZ/SECTOR-multilabel-mpnet_w",
"base_model:finetune:GIZ/SECTOR-multilabel-mpnet_w",
"co2_eq_emissions",
"region:us"
] |
text-classification
| 2024-02-12T15:28:40Z |
---
library_name: setfit
tags:
- setfit
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
metrics:
- accuracy
widget:
- text: Specific information applicable to Parties, including regional economic integration
organizations and their member States, that have reached an agreement to act jointly
under Article 4, paragraph 2, of the Paris Agreement, including the Parties that
agreed to act jointly and the terms of the agreement, in accordance with Article
4, paragraphs 16–18, of the Paris Agreement. Not applicable. (c). How the Party’s
preparation of its nationally determined contribution has been informed by the
outcomes of the global stocktake, in accordance with Article 4, paragraph 9, of
the Paris Agreement.
- text: 'In the shipping and aviation sectors, emission reduction efforts will be
focused on distributing eco-friendly ships and enhancing the operational efficiency
of aircraft. Agriculture, livestock farming and fisheries: The Republic Korea
is introducing various options to accelerate low-carbon farming, for instance,
improving irrigation techniques in rice paddies and adopting low-input systems
for nitrogen fertilizers.'
- text: As part of this commitment, Oman s upstream oil and gas industry is developing
economically viable solutions to phase out routine flaring as quickly as possible
and ahead of the World Bank s target date. IV. Climate Preparedness and Resilience.
The Sultanate of Oman has stepped up its efforts in advancing its expertise and
methodologies to better manage the climate change risks over the past five years.
The adaptation efforts are underway, and the status of adaptation planning is
still at a nascent stage.
- text: 'Synergy and coherence 46 VII- Gender and youth 46 VIII- Education and employment
48 ANNEXES. 49 Annex No. 1: Details of mitigation measures, conditional and non-conditional,
by sector 49 Annex No.2: List of adaptation actions proposed by sectors. 57 Annex
No.3: GCF project portfolio. 63 CONTRIBUTION DENTERMINEE AT NATIONAL LEVEL CDN
MAURITANIE LIST OF TABLES Table 1: Summary of funding needs for the CND 2021-2030
updated. 12 Table 2: CND 2021-2030 mitigation measures updated by sector (cumulative
cost and reduction potential for the period). 14 Table 3: CND 2021-2030 adaptation
measures updated by sector. Error!'
- text: In the transport sector, restructuing is planned through a number of large
infrastructure initiatives aiming to revive the role of public transport and achieving
a relevant share of fuel efficient vehicles. Under both the conditional and unconditional
mitigation scenarios, Lebanon will achieve sizeable emission reductions. With
regards to adaptation, Lebanon has planned comprehensive sectoral actions related
to water, agriculture/forestry and biodiversity, for example related to irrigation,
forest management, etc. It also continues developing adaptation strategies in
the remaining sectors.
pipeline_tag: text-classification
inference: false
co2_eq_emissions:
emissions: 25.8151164022705
source: codecarbon
training_type: fine-tuning
on_cloud: false
cpu_model: Intel(R) Xeon(R) CPU @ 2.00GHz
ram_total_size: 12.674781799316406
hours_used: 0.622
hardware_used: 1 x Tesla T4
base_model: ppsingh/SECTOR-multilabel-mpnet_w
---
# SetFit with ppsingh/SECTOR-multilabel-mpnet_w
This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Text Classification. This SetFit model uses [ppsingh/SECTOR-multilabel-mpnet_w](https://huggingface.co/ppsingh/SECTOR-multilabel-mpnet_w) as the Sentence Transformer embedding model. A [SetFitHead](huggingface.co/docs/setfit/reference/main#setfit.SetFitHead) instance is used for classification.
The model has been trained using an efficient few-shot learning technique that involves:
1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.
## Model Details
### Model Description
- **Model Type:** SetFit
- **Sentence Transformer body:** [ppsingh/SECTOR-multilabel-mpnet_w](https://huggingface.co/ppsingh/SECTOR-multilabel-mpnet_w)
- **Classification head:** a [SetFitHead](huggingface.co/docs/setfit/reference/main#setfit.SetFitHead) instance
- **Maximum Sequence Length:** 512 tokens
- **Number of Classes:** 4 classes
<!-- - **Training Dataset:** [Unknown](https://huggingface.co/datasets/unknown) -->
<!-- - **Language:** Unknown -->
<!-- - **License:** Unknown -->
### Model Sources
- **Repository:** [SetFit on GitHub](https://github.com/huggingface/setfit)
- **Paper:** [Efficient Few-Shot Learning Without Prompts](https://arxiv.org/abs/2209.11055)
- **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit)
## Uses
### Direct Use for Inference
First install the SetFit library:
```bash
pip install setfit
```
Then you can load this model and run inference.
```python
from setfit import SetFitModel
# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("ppsingh/iki_sector_setfit")
# Run inference
preds = model("In the shipping and aviation sectors, emission reduction efforts will be focused on distributing eco-friendly ships and enhancing the operational efficiency of aircraft. Agriculture, livestock farming and fisheries: The Republic Korea is introducing various options to accelerate low-carbon farming, for instance, improving irrigation techniques in rice paddies and adopting low-input systems for nitrogen fertilizers.")
```
<!--
### Downstream Use
*List how someone could finetune this model on their own dataset.*
-->
<!--
### Out-of-Scope Use
*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->
<!--
## Bias, Risks and Limitations
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->
<!--
### Recommendations
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->
## Training Details
### Training Set Metrics
| Training set | Min | Median | Max |
|:-------------|:----|:-------|:----|
| Word count | 35 | 76.164 | 170 |
- Training Dataset: 250
| Class | Positive Count of Class|
|:-------------|:--------|
| Economy-wide | 88 |
| Energy | 63 |
| Other Sector | 64 |
| Transport | 139 |
- Validation Dataset: 42
| Class | Positive Count of Class|
|:-------------|:--------|
| Economy-wide | 15 |
| Energy | 11 |
| Other Sector | 11 |
| Transport | 24 |
### Training Hyperparameters
- batch_size: (16, 2)
- num_epochs: (1, 10)
- max_steps: -1
- sampling_strategy: oversampling
- body_learning_rate: (2e-05, 1e-05)
- head_learning_rate: 0.01
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: False
- warmup_proportion: 0.01
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: False
### Training Results
| Epoch | Step | Training Loss | Validation Loss |
|:------:|:----:|:-------------:|:---------------:|
| 0.0005 | 1 | 0.2029 | - |
| 0.0993 | 200 | 0.0111 | 0.1124 |
| 0.1985 | 400 | 0.0063 | 0.111 |
| 0.2978 | 600 | 0.0183 | 0.1214 |
| 0.3970 | 800 | 0.0197 | 0.1248 |
| 0.4963 | 1000 | 0.0387 | 0.1339 |
| 0.5955 | 1200 | 0.0026 | 0.1181 |
| 0.6948 | 1400 | 0.0378 | 0.1208 |
| 0.7940 | 1600 | 0.0285 | 0.1267 |
| 0.8933 | 1800 | 0.0129 | 0.1254 |
| 0.9926 | 2000 | 0.0341 | 0.1271 |
### Classifier Training Results
| Epoch | Training F1-micro|Training F1-Samples |Training F1-weighted|Validation F1-micro |Validation F1-samples |Validation F1-weighted |
|:------:|:----------------:|:------------------:|:------------------:|:------------------:|:--------------------:|:---------------------:|
| 0 | 0.954 | 0.972 | 0.945 |0.824 | 0.819 | 0.813 |
| 1 | 0.994 | 0.996 | 0.994 |0.850 | 0.832 | 0.852 |
| 2 | 0.981 | 0.989 | 0.979 |0.850 | 0.843 | 0.852 |
| 3 | 0.995 | 0.997 | 0.995 |0.852 | 0.843 | 0.858 |
| 4 | 0.994 | 0.996 | 0.994 |0.852 | 0.843 | 0.858 |
| 5 | 0.995 | 0.997 | 0.995 |0.859 | 0.848 | 0.863 |
|label | precision |recall |f1-score| support|
|:-------------:|:---------:|:-----:|:------:|:------:|
|Economy-wide |0.857 |0.800 |0.827 | 15.0 |
|Energy |1.00 |0.818 |0.900 | 11.0 |
|Other Sector |0.615 |0.727 |0.667 | 11.0 |
|Transport |0.958 |0.958 |0.958 | 24.0 |
- Micro Avg: Precision = 0.866, Recall = 0.852, F1 = 0.859504
- Samples Avg: Precision = 0.869, Recall = 0.861, F1 = 0.848
### Environmental Impact
Carbon emissions were measured using [CodeCarbon](https://github.com/mlco2/codecarbon).
- **Carbon Emitted**: 0.026 kg of CO2
- **Hours Used**: 0.622 hours
### Training Hardware
- **On Cloud**: No
- **GPU Model**: 1 x Tesla T4
- **CPU Model**: Intel(R) Xeon(R) CPU @ 2.00GHz
- **RAM Size**: 12.67 GB
### Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 2.3.1
- Transformers: 4.35.2
- PyTorch: 2.1.0+cu121
- Datasets: 2.3.0
- Tokenizers: 0.15.1
## Citation
### BibTeX
```bibtex
@article{https://doi.org/10.48550/arxiv.2209.11055,
doi = {10.48550/ARXIV.2209.11055},
url = {https://arxiv.org/abs/2209.11055},
author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Efficient Few-Shot Learning Without Prompts},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
```
<!--
## Glossary
*Clearly define terms in order to be accessible across audiences.*
-->
<!--
## Model Card Authors
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->
<!--
## Model Card Contact
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
-->
|
rishabhjain16/whisper-large-v3
|
rishabhjain16
| 2024-02-12T16:17:10Z | 13 | 0 |
transformers
|
[
"transformers",
"pytorch",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"arxiv:2311.00430",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:17:10Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al. from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
Whisper `large-v3` has the same architecture as the previous large models except the following minor differences:
1. The input uses 128 Mel frequency bins instead of 80
2. A new language token for Cantonese
The Whisper `large-v3` model is trained on 1 million hours of weakly labeled audio and 4 million hours of pseudolabeled audio collected using Whisper `large-v2`.
The model was trained for 2.0 epochs over this mixture dataset.
The `large-v3` model shows improved performance over a wide variety of languages, showing 10% to 20% reduction of errors compared to Whisper `large-v2`.
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 1 million hours of weakly labeled audio and 4 million hours of pseudolabeled audio collected using Whisper `large-v2`.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
| large-v3 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v3) |
## Usage
Whisper `large-v3` is supported in Hugging Face 🤗 Transformers through the `main` branch in the Transformers repo. To run the model, first
install the Transformers library through the GitHub repo. For this example, we'll also install 🤗 Datasets to load toy
audio dataset from the Hugging Face Hub:
```bash
pip install --upgrade pip
pip install --upgrade git+https://github.com/huggingface/transformers.git accelerate datasets[audio]
```
The model can be used with the [`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
class to transcribe audio files of arbitrary length. Transformers uses a chunked algorithm to transcribe
long-form audio files, which in-practice is 9x faster than the sequential algorithm proposed by OpenAI
(see Table 7 of the [Distil-Whisper paper](https://arxiv.org/abs/2311.00430)). The batch size should
be set based on the specifications of your device:
```python
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
from datasets import load_dataset
device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
model_id = "openai/whisper-large-v3"
model = AutoModelForSpeechSeq2Seq.from_pretrained(
model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
)
model.to(device)
processor = AutoProcessor.from_pretrained(model_id)
pipe = pipeline(
"automatic-speech-recognition",
model=model,
tokenizer=processor.tokenizer,
feature_extractor=processor.feature_extractor,
max_new_tokens=128,
chunk_length_s=30,
batch_size=16,
return_timestamps=True,
torch_dtype=torch_dtype,
device=device,
)
dataset = load_dataset("distil-whisper/librispeech_long", "clean", split="validation")
sample = dataset[0]["audio"]
result = pipe(sample)
print(result["text"])
```
To transcribe a local audio file, simply pass the path to your audio file when you call the pipeline:
```diff
- result = pipe(sample)
+ result = pipe("audio.mp3")
```
Whisper predicts the language of the source audio automatically. If the source audio language is known *a-priori*, it
can be passed as an argument to the pipeline:
```python
result = pipe(sample, generate_kwargs={"language": "english"})
```
By default, Whisper performs the task of *speech transcription*, where the source audio language is the same as the target
text language. To perform *speech translation*, where the target text is in English, set the task to `"translate"`:
```python
result = pipe(sample, generate_kwargs={"task": "translate"})
```
Finally, the model can be made to predict timestamps. For sentence-level timestamps, pass the `return_timestamps` argument:
```python
result = pipe(sample, return_timestamps=True)
print(result["chunks"])
```
And for word-level timestamps:
```python
result = pipe(sample, return_timestamps="word")
print(result["chunks"])
```
The above arguments can be used in isolation or in combination. For example, to perform the task of speech transcription
where the source audio is in French, and we want to return sentence-level timestamps, the following can be used:
```python
result = pipe(sample, return_timestamps=True, generate_kwargs={"language": "french", "task": "translate"})
print(result["chunks"])
```
## Additional Speed & Memory Improvements
You can apply additional speed and memory improvements to Whisper-large-v3 which we cover in the following.
### Flash Attention
We recommend using [Flash-Attention 2](https://huggingface.co/docs/transformers/main/en/perf_infer_gpu_one#flashattention-2) if your GPU allows for it.
To do so, you first need to install [Flash Attention](https://github.com/Dao-AILab/flash-attention):
```
pip install flash-attn --no-build-isolation
```
and then all you have to do is to pass `use_flash_attention_2=True` to `from_pretrained`:
```diff
- model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True)
+ model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True, use_flash_attention_2=True)
```
### Torch Scale-Product-Attention (SDPA)
If your GPU does not support Flash Attention, we recommend making use of [BetterTransformers](https://huggingface.co/docs/transformers/main/en/perf_infer_gpu_one#bettertransformer).
To do so, you first need to install optimum:
```
pip install --upgrade optimum
```
And then convert your model to a "BetterTransformer" model before using it:
```diff
model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True)
+ model = model.to_bettertransformer()
```
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 1 million hours of weakly labeled audio and 4 million hours of pseudolabeled audio collected using Whisper `large-v2`.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-large-v2
|
rishabhjain16
| 2024-02-12T16:16:58Z | 9 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:16:57Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al. from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
Compared to the Whisper large model, the large-v2 model is trained for 2.5x more epochs with added regularization
for improved performance.
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large-v2")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large-v2")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large-v2")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large-v2")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large-v2")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large-v2")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Large on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large-v2")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large-v2").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
3.0003583080317572
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-large-v2",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-large
|
rishabhjain16
| 2024-02-12T16:09:11Z | 7 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:09:10Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
model-index:
- name: whisper-large
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (clean)
type: librispeech_asr
config: clean
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 3.0
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (other)
type: librispeech_asr
config: other
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 5.4
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.0
type: mozilla-foundation/common_voice_11_0
config: hi
split: test
args:
language: hi
metrics:
- name: Test WER
type: wer
value: 54.8
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
<div class="course-tip course-tip-orange bg-gradient-to-br dark:bg-gradient-to-r before:border-orange-500 dark:before:border-orange-800 from-orange-50 dark:from-gray-900 to-white dark:to-gray-950 border border-orange-50 text-orange-700 dark:text-gray-400">
<p><b>Update:</b> following the release of the paper, the Whisper authors announced a <a href="ttps://huggingface.co/openai/whisper-large-v2"> large-v2</a> model trained for 2.5x more epochs with regularization. This <a href="ttps://huggingface.co/openai/whisper-large-v2"> large-v2</a> model surpasses the performance of the large model, with no architecture changes. Thus, it is recommended that the <a href="ttps://huggingface.co/openai/whisper-large-v2"> large-v2</a> model is used in-place of the original large model. </p>
</div>
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Large on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-large")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
3.0003583080317572
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-large",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-base
|
rishabhjain16
| 2024-02-12T16:07:55Z | 67 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:07:55Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
model-index:
- name: whisper-base
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (clean)
type: librispeech_asr
config: clean
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 5.008769117619326
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (other)
type: librispeech_asr
config: other
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 12.84936273212057
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.0
type: mozilla-foundation/common_voice_11_0
config: hi
split: test
args:
language: hi
metrics:
- name: Test WER
type: wer
value: 131
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-base")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-base")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-base")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Base on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-base")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
5.082316555716899
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-base",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-medium
|
rishabhjain16
| 2024-02-12T16:07:01Z | 12 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:07:00Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
model-index:
- name: whisper-medium
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (clean)
type: librispeech_asr
config: clean
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 2.9
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (other)
type: librispeech_asr
config: other
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 5.9
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.0
type: mozilla-foundation/common_voice_11_0
config: hi
split: test
args:
language: hi
metrics:
- name: Test WER
type: wer
value: 53.87
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-medium")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-medium")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-medium")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-medium")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-medium")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-medium")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Medium on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-medium")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-medium").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
2.900409225488902
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-medium",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-tiny
|
rishabhjain16
| 2024-02-12T16:06:47Z | 67 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:06:47Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
model-index:
- name: whisper-tiny
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (clean)
type: librispeech_asr
config: clean
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 7.54
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (other)
type: librispeech_asr
config: other
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 17.15
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.0
type: mozilla-foundation/common_voice_11_0
config: hi
split: test
args:
language: hi
metrics:
- name: Test WER
type: wer
value: 141
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-tiny")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-tiny")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-tiny")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Tiny on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-tiny")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
7.547098647858638
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-tiny",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
rishabhjain16/whisper-small
|
rishabhjain16
| 2024-02-12T16:06:01Z | 72 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tf",
"jax",
"safetensors",
"whisper",
"automatic-speech-recognition",
"audio",
"hf-asr-leaderboard",
"en",
"zh",
"de",
"es",
"ru",
"ko",
"fr",
"ja",
"pt",
"tr",
"pl",
"ca",
"nl",
"ar",
"sv",
"it",
"id",
"hi",
"fi",
"vi",
"he",
"uk",
"el",
"ms",
"cs",
"ro",
"da",
"hu",
"ta",
"no",
"th",
"ur",
"hr",
"bg",
"lt",
"la",
"mi",
"ml",
"cy",
"sk",
"te",
"fa",
"lv",
"bn",
"sr",
"az",
"sl",
"kn",
"et",
"mk",
"br",
"eu",
"is",
"hy",
"ne",
"mn",
"bs",
"kk",
"sq",
"sw",
"gl",
"mr",
"pa",
"si",
"km",
"sn",
"yo",
"so",
"af",
"oc",
"ka",
"be",
"tg",
"sd",
"gu",
"am",
"yi",
"lo",
"uz",
"fo",
"ht",
"ps",
"tk",
"nn",
"mt",
"sa",
"lb",
"my",
"bo",
"tl",
"mg",
"as",
"tt",
"haw",
"ln",
"ha",
"ba",
"jw",
"su",
"arxiv:2212.04356",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-12T16:05:58Z |
---
language:
- en
- zh
- de
- es
- ru
- ko
- fr
- ja
- pt
- tr
- pl
- ca
- nl
- ar
- sv
- it
- id
- hi
- fi
- vi
- he
- uk
- el
- ms
- cs
- ro
- da
- hu
- ta
- no
- th
- ur
- hr
- bg
- lt
- la
- mi
- ml
- cy
- sk
- te
- fa
- lv
- bn
- sr
- az
- sl
- kn
- et
- mk
- br
- eu
- is
- hy
- ne
- mn
- bs
- kk
- sq
- sw
- gl
- mr
- pa
- si
- km
- sn
- yo
- so
- af
- oc
- ka
- be
- tg
- sd
- gu
- am
- yi
- lo
- uz
- fo
- ht
- ps
- tk
- nn
- mt
- sa
- lb
- my
- bo
- tl
- mg
- as
- tt
- haw
- ln
- ha
- ba
- jw
- su
tags:
- audio
- automatic-speech-recognition
- hf-asr-leaderboard
widget:
- example_title: Librispeech sample 1
src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
- example_title: Librispeech sample 2
src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
model-index:
- name: whisper-small
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (clean)
type: librispeech_asr
config: clean
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 3.432213777886737
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: LibriSpeech (other)
type: librispeech_asr
config: other
split: test
args:
language: en
metrics:
- name: Test WER
type: wer
value: 7.628304527060248
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.0
type: mozilla-foundation/common_voice_11_0
config: hi
split: test
args:
language: hi
metrics:
- name: Test WER
type: wer
value: 87.3
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 13.0
type: mozilla-foundation/common_voice_13_0
config: dv
split: test
args:
language: dv
metrics:
- name: Wer
type: wer
value: 125.69809089960707
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---
# Whisper
Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours
of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains **without** the need
for fine-tuning.
Whisper was proposed in the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://arxiv.org/abs/2212.04356)
by Alec Radford et al from OpenAI. The original code repository can be found [here](https://github.com/openai/whisper).
**Disclaimer**: Content for this model card has partly been written by the Hugging Face team, and parts of it were
copied and pasted from the original model card.
## Model details
Whisper is a Transformer based encoder-decoder model, also referred to as a _sequence-to-sequence_ model.
It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision.
The models were trained on either English-only data or multilingual data. The English-only models were trained
on the task of speech recognition. The multilingual models were trained on both speech recognition and speech
translation. For speech recognition, the model predicts transcriptions in the *same* language as the audio.
For speech translation, the model predicts transcriptions to a *different* language to the audio.
Whisper checkpoints come in five configurations of varying model sizes.
The smallest four are trained on either English-only or multilingual data.
The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints
are available on the [Hugging Face Hub](https://huggingface.co/models?search=openai/whisper). The
checkpoints are summarised in the following table with links to the models on the Hub:
| Size | Parameters | English-only | Multilingual |
|----------|------------|------------------------------------------------------|-----------------------------------------------------|
| tiny | 39 M | [✓](https://huggingface.co/openai/whisper-tiny.en) | [✓](https://huggingface.co/openai/whisper-tiny) |
| base | 74 M | [✓](https://huggingface.co/openai/whisper-base.en) | [✓](https://huggingface.co/openai/whisper-base) |
| small | 244 M | [✓](https://huggingface.co/openai/whisper-small.en) | [✓](https://huggingface.co/openai/whisper-small) |
| medium | 769 M | [✓](https://huggingface.co/openai/whisper-medium.en) | [✓](https://huggingface.co/openai/whisper-medium) |
| large | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large) |
| large-v2 | 1550 M | x | [✓](https://huggingface.co/openai/whisper-large-v2) |
# Usage
To transcribe audio samples, the model has to be used alongside a [`WhisperProcessor`](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperProcessor).
The `WhisperProcessor` is used to:
1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model)
2. Post-process the model outputs (converting them from tokens to text)
The model is informed of which task to perform (transcription or translation) by passing the appropriate "context tokens". These context tokens
are a sequence of tokens that are given to the decoder at the start of the decoding process, and take the following order:
1. The transcription always starts with the `<|startoftranscript|>` token
2. The second token is the language token (e.g. `<|en|>` for English)
3. The third token is the "task token". It can take one of two values: `<|transcribe|>` for speech recognition or `<|translate|>` for speech translation
4. In addition, a `<|notimestamps|>` token is added if the model should not include timestamp prediction
Thus, a typical sequence of context tokens might look as follows:
```
<|startoftranscript|> <|en|> <|transcribe|> <|notimestamps|>
```
Which tells the model to decode in English, under the task of speech recognition, and not to predict timestamps.
These tokens can either be forced or un-forced. If they are forced, the model is made to predict each token at
each position. This allows one to control the output language and task for the Whisper model. If they are un-forced,
the Whisper model will automatically predict the output langauge and task itself.
The context tokens can be set accordingly:
```python
model.config.forced_decoder_ids = WhisperProcessor.get_decoder_prompt_ids(language="english", task="transcribe")
```
Which forces the model to predict in English under the task of speech recognition.
## Transcription
### English to English
In this example, the context tokens are 'unforced', meaning the model automatically predicts the output language
(English) and task (transcribe).
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-small")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-small")
>>> model.config.forced_decoder_ids = None
>>> # load dummy dataset and read audio files
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=False)
['<|startoftranscript|><|en|><|transcribe|><|notimestamps|> Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.']
```
The context tokens can be removed from the start of the transcription by setting `skip_special_tokens=True`.
### French to French
The following example demonstrates French to French transcription by setting the decoder ids appropriately.
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-small")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-small")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="transcribe")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids)
['<|startoftranscript|><|fr|><|transcribe|><|notimestamps|> Un vrai travail intéressant va enfin être mené sur ce sujet.<|endoftext|>']
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' Un vrai travail intéressant va enfin être mené sur ce sujet.']
```
## Translation
Setting the task to "translate" forces the Whisper model to perform speech translation.
### French to English
```python
>>> from transformers import WhisperProcessor, WhisperForConditionalGeneration
>>> from datasets import Audio, load_dataset
>>> # load model and processor
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-small")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-small")
>>> forced_decoder_ids = processor.get_decoder_prompt_ids(language="french", task="translate")
>>> # load streaming dataset and read first audio sample
>>> ds = load_dataset("common_voice", "fr", split="test", streaming=True)
>>> ds = ds.cast_column("audio", Audio(sampling_rate=16_000))
>>> input_speech = next(iter(ds))["audio"]
>>> input_features = processor(input_speech["array"], sampling_rate=input_speech["sampling_rate"], return_tensors="pt").input_features
>>> # generate token ids
>>> predicted_ids = model.generate(input_features, forced_decoder_ids=forced_decoder_ids)
>>> # decode token ids to text
>>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
[' A very interesting work, we will finally be given on this subject.']
```
## Evaluation
This code snippet shows how to evaluate Whisper Small on [LibriSpeech test-clean](https://huggingface.co/datasets/librispeech_asr):
```python
>>> from datasets import load_dataset
>>> from transformers import WhisperForConditionalGeneration, WhisperProcessor
>>> import torch
>>> from evaluate import load
>>> librispeech_test_clean = load_dataset("librispeech_asr", "clean", split="test")
>>> processor = WhisperProcessor.from_pretrained("openai/whisper-small")
>>> model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-small").to("cuda")
>>> def map_to_pred(batch):
>>> audio = batch["audio"]
>>> input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
>>> batch["reference"] = processor.tokenizer._normalize(batch['text'])
>>>
>>> with torch.no_grad():
>>> predicted_ids = model.generate(input_features.to("cuda"))[0]
>>> transcription = processor.decode(predicted_ids)
>>> batch["prediction"] = processor.tokenizer._normalize(transcription)
>>> return batch
>>> result = librispeech_test_clean.map(map_to_pred)
>>> wer = load("wer")
>>> print(100 * wer.compute(references=result["reference"], predictions=result["prediction"]))
3.432213777886737
```
## Long-Form Transcription
The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking
algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers
[`pipeline`](https://huggingface.co/docs/transformers/main_classes/pipelines#transformers.AutomaticSpeechRecognitionPipeline)
method. Chunking is enabled by setting `chunk_length_s=30` when instantiating the pipeline. With chunking enabled, the pipeline
can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `return_timestamps=True`:
```python
>>> import torch
>>> from transformers import pipeline
>>> from datasets import load_dataset
>>> device = "cuda:0" if torch.cuda.is_available() else "cpu"
>>> pipe = pipeline(
>>> "automatic-speech-recognition",
>>> model="openai/whisper-small",
>>> chunk_length_s=30,
>>> device=device,
>>> )
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
>>> sample = ds[0]["audio"]
>>> prediction = pipe(sample.copy(), batch_size=8)["text"]
" Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel."
>>> # we can also return timestamps for the predictions
>>> prediction = pipe(sample.copy(), batch_size=8, return_timestamps=True)["chunks"]
[{'text': ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.',
'timestamp': (0.0, 5.44)}]
```
Refer to the blog post [ASR Chunking](https://huggingface.co/blog/asr-chunking) for more details on the chunking algorithm.
## Fine-Tuning
The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However,
its predictive capabilities can be improved further for certain languages and tasks through *fine-tuning*. The blog
post [Fine-Tune Whisper with 🤗 Transformers](https://huggingface.co/blog/fine-tune-whisper) provides a step-by-step
guide to fine-tuning the Whisper model with as little as 5 hours of labelled data.
### Evaluated Use
The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research.
The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them.
In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes.
## Training Data
The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages.
As discussed in [the accompanying paper](https://cdn.openai.com/papers/whisper.pdf), we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language.
## Performance and Limitations
Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level.
However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself.
Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in [the paper accompanying this release](https://cdn.openai.com/papers/whisper.pdf).
In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in [the paper](https://cdn.openai.com/papers/whisper.pdf). It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages.
## Broader Implications
We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications.
There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.
### BibTeX entry and citation info
```bibtex
@misc{radford2022whisper,
doi = {10.48550/ARXIV.2212.04356},
url = {https://arxiv.org/abs/2212.04356},
author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},
title = {Robust Speech Recognition via Large-Scale Weak Supervision},
publisher = {arXiv},
year = {2022},
copyright = {arXiv.org perpetual, non-exclusive license}
}
```
|
ychafiqui/english-to-darija-2
|
ychafiqui
| 2024-02-12T16:05:30Z | 128 | 1 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"marian",
"text2text-generation",
"generated_from_trainer",
"base_model:Helsinki-NLP/opus-mt-tc-big-en-ar",
"base_model:finetune:Helsinki-NLP/opus-mt-tc-big-en-ar",
"license:cc-by-4.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text2text-generation
| 2024-01-20T15:59:34Z |
---
license: cc-by-4.0
base_model: Helsinki-NLP/opus-mt-tc-big-en-ar
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: english-to-darija-2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# english-to-darija-2
This model is a fine-tuned version of [Helsinki-NLP/opus-mt-tc-big-en-ar](https://huggingface.co/Helsinki-NLP/opus-mt-tc-big-en-ar) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8514
- Bleu: 70.9947
- Gen Len: 9.092
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|:-------------:|:-----:|:-----:|:---------------:|:-------:|:-------:|
| 1.6032 | 1.0 | 4651 | 1.4540 | 25.9364 | 8.9697 |
| 1.1191 | 2.0 | 9302 | 1.0805 | 48.0549 | 9.0661 |
| 0.8048 | 3.0 | 13953 | 0.9419 | 61.3646 | 9.1018 |
| 0.5978 | 4.0 | 18604 | 0.8939 | 65.6846 | 9.1161 |
| 0.477 | 5.0 | 23255 | 0.8623 | 68.0005 | 9.1049 |
| 0.4228 | 6.0 | 27906 | 0.8540 | 69.1959 | 9.1276 |
| 0.3534 | 7.0 | 32557 | 0.8479 | 69.944 | 9.0744 |
| 0.305 | 8.0 | 37208 | 0.8473 | 70.55 | 9.0987 |
| 0.2678 | 9.0 | 41859 | 0.8489 | 70.8065 | 9.1166 |
| 0.243 | 10.0 | 46510 | 0.8514 | 70.9947 | 9.092 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.2.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
alhafizfadhil/results
|
alhafizfadhil
| 2024-02-12T15:56:21Z | 179 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:google/vit-base-patch16-224",
"base_model:finetune:google/vit-base-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2024-02-12T15:56:04Z |
---
license: apache-2.0
base_model: google/vit-base-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: results
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.4875
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# results
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2952
- Accuracy: 0.4875
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 80 | 1.6148 | 0.3375 |
| 1.6678 | 2.0 | 160 | 1.3553 | 0.4625 |
| 0.6788 | 3.0 | 240 | 1.2952 | 0.4875 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
furrutiav/bert_qa_extractor_cockatiel_2022_z_value_over_subsample_it_727
|
furrutiav
| 2024-02-12T15:52:27Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"feature-extraction",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] |
feature-extraction
| 2024-02-12T15:51:58Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
hdeldar/distilbert-base-uncased-finetuned-cola
|
hdeldar
| 2024-02-12T15:51:58Z | 46 | 0 |
transformers
|
[
"transformers",
"tf",
"tensorboard",
"distilbert",
"text-classification",
"generated_from_keras_callback",
"base_model:distilbert/distilbert-base-uncased",
"base_model:finetune:distilbert/distilbert-base-uncased",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T15:47:40Z |
---
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_keras_callback
model-index:
- name: hdeldar/distilbert-base-uncased-finetuned-cola
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# hdeldar/distilbert-base-uncased-finetuned-cola
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.1972
- Validation Loss: 0.5241
- Train Matthews Correlation: 0.5294
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 1602, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Validation Loss | Train Matthews Correlation | Epoch |
|:----------:|:---------------:|:--------------------------:|:-----:|
| 0.5250 | 0.4718 | 0.4527 | 0 |
| 0.3234 | 0.4414 | 0.5235 | 1 |
| 0.1972 | 0.5241 | 0.5294 | 2 |
### Framework versions
- Transformers 4.35.2
- TensorFlow 2.15.0
- Datasets 2.17.0
- Tokenizers 0.15.1
|
Zaphare/ppo-LunarLander-v2
|
Zaphare
| 2024-02-12T15:41:00Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T13:55:36Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 280.09 +/- 14.77
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
not-lain/MyRepo1.0
|
not-lain
| 2024-02-12T15:34:50Z | 194 | 0 |
transformers
|
[
"transformers",
"safetensors",
"MobileNetV1",
"image-classification",
"custom_code",
"autotrain_compatible",
"region:us"
] |
image-classification
| 2024-02-12T15:33:46Z |
---
tags:
- custom_code
---
# How to use
you can the model via the command
```python
from transformers import AutoModelForImageClassification
model = AutoModelForImageClassification.from_pretrained("not-lain/MyRepo1.0", trust_remote_code=True)
```
or you can use the pipeline
```python
from transformers import pipeline
pipe = pipeline(model="not-lain/MyRepo1.0", trust_remote_code=True)
pipe(
"url",
download=True, # will call the download_img method
clean_output=False # will be passed as postprocess_kwargs
)
```
# parameters
the pipeline supports the following parameters :
* download
* clean_output
you can also use the following method to download images from the web
```python
pipe.download_img(url)
```
|
ppsingh/iki_target_setfit
|
ppsingh
| 2024-02-12T15:24:33Z | 57 | 0 |
setfit
|
[
"setfit",
"safetensors",
"mpnet",
"sentence-transformers",
"text-classification",
"generated_from_setfit_trainer",
"arxiv:2209.11055",
"base_model:GIZ/TAPP-multilabel-mpnet",
"base_model:finetune:GIZ/TAPP-multilabel-mpnet",
"co2_eq_emissions",
"region:us"
] |
text-classification
| 2024-02-11T18:11:00Z |
---
library_name: setfit
tags:
- setfit
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
metrics:
- accuracy
widget:
- text: During 2021-2030, Thailand s LEDS will be implemented through the NDC roadmap
and sectoral action plans which provide detailed guidance on measures and realistic
actions to achieve the 1st NDC target by 2030, as well as regular monitoring and
evaluation of the progress and achievement. The monitoring and evaluation of the
mitigation measures relating to the Thailand’s LEDS will be carried out to ensure
its effectiveness and efficiency in achieving its objectives and key performance
indicators. Because it is a long-term plan spanning many years during which many
changes can occur, it is envisaged that it will be subject to a comprehensive
review every five years. This is consistent with the approach under the Paris
Agreement that assigned Parties to submit their NDCs to the UNFCCC every five
year.
- text: The NDC also benefited from the reviews and comments of these implementing
partners as well as local and international experts. Special thanks to The Honourable
Molwyn Joseph, Minister for Health, Wellness and the Environment, for his unwavering
commitment to advance this ambitious climate change agenda, while Antigua and
Barbuda faced an outbreak of the COVID-19 pandemic. Significant contributions
to the process were made by a wide-cross section of stakeholders from the public
and private sector, civil society, trade and industry groups and training institutions,
who attended NDC-related workshops, consultations and participated in key stakeholder
interviews organized to inform the NDC update.
- text: Antigua and Barbuda will mainstream gender in its energy planning through
an Inclusive Renewable Energy Strategy. This strategy will recognize and acknowledge,
among other things, the gender norms, and inequalities prevalent in the energy
sector, women and men’s differentiated access to energy, their different energy
needs and preferences, and different impacts that energy access could have on
their livelihoods. Antigua and Barbuda’s plan for an inclusive renewable energy
transition will ensure continued affordable and reliable access to electricity
and other energy services for all.
- text: 'Thailand’s climate actions are divided into short-term, medium-term and long-term
targets up to 2050. For the mitigation actions, short-term targets include: (i)
develop medium- and long-term GHG emission reduction targets and prepare roadmaps
for the implementation by sector, including the GHG emission reduction target
on a voluntary basis (pre-2020 target), Nationally Appropriate Mitigation Actions
(NAMAs) roadmaps, and measurement, reporting, and verification mechanisms, (ii)
establish domestic incentive mechanisms to encourage low carbon development. The
medium-term targets include: (i) reduce GHG emissions from energy and transport
sectors by 7-20% against BAU level by 2020, subject to the level of international
support, (ii) supply at least 25% of energy consumption from renewable energy
sources by 2021 and (iii) increase the ratio of municipalities with more than
10 m2 of green space per capita.'
- text: In the oil sector, the country has benefited from 372 million dollars for
the reduction of gas flaring at the initiative (GGFR - "Global Gas Flaring Reduction")
of the World Bank after having adopted in November 2015 a national reduction plan
flaring and associated gas upgrading. In the electricity sector, the NDC highlights
the development of hydroelectricity which should make it possible to cover 80%
of production in 2025, the remaining 20% ​​being
covered by gas and other renewable energies.
pipeline_tag: text-classification
inference: true
co2_eq_emissions:
emissions: 5.901369050433577
source: codecarbon
training_type: fine-tuning
on_cloud: false
cpu_model: Intel(R) Xeon(R) CPU @ 2.00GHz
ram_total_size: 12.674789428710938
hours_used: 0.185
hardware_used: 1 x Tesla T4
base_model: ppsingh/TAPP-multilabel-mpnet
---
# SetFit with ppsingh/TAPP-multilabel-mpnet
This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Text Classification. This SetFit model uses [ppsingh/TAPP-multilabel-mpnet](https://huggingface.co/ppsingh/TAPP-multilabel-mpnet) as the Sentence Transformer embedding model. A [SetFitHead](huggingface.co/docs/setfit/reference/main#setfit.SetFitHead) instance is used for classification.
The model has been trained using an efficient few-shot learning technique that involves:
1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.
## Model Details
### Model Description
- **Model Type:** SetFit
- **Sentence Transformer body:** [ppsingh/TAPP-multilabel-mpnet](https://huggingface.co/ppsingh/TAPP-multilabel-mpnet)
- **Classification head:** a [SetFitHead](huggingface.co/docs/setfit/reference/main#setfit.SetFitHead) instance
- **Maximum Sequence Length:** 512 tokens
- **Number of Classes:** 2 classes
<!-- - **Training Dataset:** [Unknown](https://huggingface.co/datasets/unknown) -->
<!-- - **Language:** Unknown -->
<!-- - **License:** Unknown -->
### Model Sources
- **Repository:** [SetFit on GitHub](https://github.com/huggingface/setfit)
- **Paper:** [Efficient Few-Shot Learning Without Prompts](https://arxiv.org/abs/2209.11055)
- **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit)
### Model Labels
| Label | Examples |
|:---------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| NEGATIVE | <ul><li>'(p 70-1).Antigua and Barbuda’s 2021 update to the first Nationally Determined Contribution the most vulnerable in society have been predominantly focused on adaptation measures like building resilience to flooding and hurricanes. The updated NDC ambition provides an opportunity to focus more intently on enabling access to energy efficiency and renewable energy for the most vulnerable, particularly women who are most affected when electricity is not available since the grid is down after an extreme weather event. Nationally, Antigua and Barbuda intends to utilize the SIRF Fund as a mechanism primarily to catalyse and leverage investment in the transition for NGOs, MSMEs and informal sectors that normally cannot access traditional local commercial financing due to perceived high risks.'</li><li>'The transport system cost will be increased by 16.2% compared to the BAU level. Electric trucks and electric pick-ups will account for the highest share of investment followed by electric buses and trucks. In the manufacturing industries, the energy efficiency improvement in the heating and the motor systems and the deployment of CCS require the highest investment in the non-metallic and the chemical industries in 2050. The manufacturing industries system cost will be increased by 15.3% compared to the BAU level.'</li><li>'Figure 1-9: Total GHG emissions by sector (excluding LULUCF) 2000 and 2016 1.2.2 Greenhouse Gas Emission by Sector • Energy Total direct GHG emissions from the Energy sector in 2016 were estimated to be 253,895.61 eq. The majority of GHG emissions in the Energy sector were generated by fuel combustion, consisting mostly of grid-connected electricity and heat production at around eq (42.84%). GHG emissions from Transport, Manufacturing Industries and Construction, and other sectors were 68,260.17 GgCO2 eq eq (6.10%), respectively. Fugitive Emissions from fuel eq or a little over 4.33% of total GHG emissions from the Energy sector. Details of GHG emissions in the Energy sector by gas type and source in 2016 are presented in Figure 1-10. Source: Thailand Third Biennial Update Report, UNFCCC 2020.'</li></ul> |
| TARGET | <ul><li>'DNPM, NFA,. Cocoa. Board,. Spice Board,. Provincial. gov-ernments. in the. Momase. region. Ongoing -. 2025. 340. European Union. Support committed. Priority Sector: Health. By 2030, 100% of the population benefit from introduced health measures to respond to malaria and other climate-sensitive diseases in PNG. Action or Activity. Indicator. Status. Lead. Implementing. Agencies. Supporting. Agencies. Time Frame. Budget (USD). Funding Source. (Existing/Potential). Other Support. Improve vector control. measures, with a priority. of all households having. access to a long-lasting. insecticidal net (LLIN).'</li><li>'Conditionality: With national effort it is intended to increase the attention to vulnerable groups in case of disasters and/or emergencies up to 50% of the target and 100% of the target with international cooperation. Description: In this goal, it is projected to increase coverage from 33% to 50% (211,000 families) of agricultural insurance in attention to the number of families, whose crops were affected by various adverse weather events (flood, drought, frost, hailstorm, among others), in addition to the implementation of comprehensive actions for risk management and adaptation to Climate Change.'</li><li>'By 2030, upgrade watershed health and vitality in at least 20 districts to a higher condition category. By 2030, create an inventory of wetlands in Nepal and sustainably manage vulnerable wetlands. By 2025, enhance the sink capacity of the landuse sector by instituting the Forest Development Fund (FDF) for compensation of plantations and forest restoration. Increase growing stock including Mean Annual Increment in Tarai, Hills and Mountains. Afforest/reforest viable public and private lands, including agroforestry.'</li></ul> |
## Uses
### Direct Use for Inference
First install the SetFit library:
```bash
pip install setfit
```
Then you can load this model and run inference.
```python
from setfit import SetFitModel
# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("ppsingh/iki_target_setfit")
# Run inference
preds = model("In the oil sector, the country has benefited from 372 million dollars for the reduction of gas flaring at the initiative (GGFR - \"Global Gas Flaring Reduction\") of the World Bank after having adopted in November 2015 a national reduction plan flaring and associated gas upgrading. In the electricity sector, the NDC highlights the development of hydroelectricity which should make it possible to cover 80% of production in 2025, the remaining 20% ​​being covered by gas and other renewable energies.")
```
<!--
### Downstream Use
*List how someone could finetune this model on their own dataset.*
-->
<!--
### Out-of-Scope Use
*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->
<!--
## Bias, Risks and Limitations
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->
<!--
### Recommendations
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->
## Training Details
### Training Set Metrics
| Training set | Min | Median | Max |
|:-------------|:----|:---------|:----|
| Word count | 58 | 116.6632 | 508 |
| Label | Training Sample Count |
|:---------|:----------------------|
| NEGATIVE | 51 |
| TARGET | 44 |
### Training Hyperparameters
- batch_size: (8, 2)
- num_epochs: (1, 0)
- max_steps: -1
- sampling_strategy: undersampling
- body_learning_rate: (2e-05, 1e-05)
- head_learning_rate: 0.01
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: False
- warmup_proportion: 0.01
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: False
### Training Results
| Epoch | Step | Training Loss | Validation Loss |
|:------:|:----:|:-------------:|:---------------:|
| 0.0018 | 1 | 0.3343 | - |
| 0.1783 | 100 | 0.0026 | 0.1965 |
| 0.3565 | 200 | 0.0001 | 0.1995 |
| 0.5348 | 300 | 0.0001 | 0.2105 |
| 0.7130 | 400 | 0.0001 | 0.2153 |
| 0.8913 | 500 | 0.0 | 0.1927 |
### Training Results Classifier
- Classes Representation in Test Data: Target: 9, Negative: 8
- F1-score: 87.8%
- Accuracy: 88.2%
### Environmental Impact
Carbon emissions were measured using [CodeCarbon](https://github.com/mlco2/codecarbon).
- **Carbon Emitted**: 0.006 kg of CO2
- **Hours Used**: 0.185 hours
### Training Hardware
- **On Cloud**: No
- **GPU Model**: 1 x Tesla T4
- **CPU Model**: Intel(R) Xeon(R) CPU @ 2.00GHz
- **RAM Size**: 12.67 GB
### Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 2.3.1
- Transformers: 4.35.2
- PyTorch: 2.1.0+cu121
- Datasets: 2.3.0
- Tokenizers: 0.15.1
## Citation
### BibTeX
```bibtex
@article{https://doi.org/10.48550/arxiv.2209.11055,
doi = {10.48550/ARXIV.2209.11055},
url = {https://arxiv.org/abs/2209.11055},
author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Efficient Few-Shot Learning Without Prompts},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
```
<!--
## Glossary
*Clearly define terms in order to be accessible across audiences.*
-->
<!--
## Model Card Authors
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->
<!--
## Model Card Contact
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
-->
|
BharatMata/my-dog
|
BharatMata
| 2024-02-12T15:22:42Z | 0 | 0 | null |
[
"safetensors",
"NxtWave-GenAI-Webinar",
"text-to-image",
"stable-diffusion",
"license:creativeml-openrail-m",
"region:us"
] |
text-to-image
| 2024-02-12T15:20:20Z |
---
license: creativeml-openrail-m
tags:
- NxtWave-GenAI-Webinar
- text-to-image
- stable-diffusion
---
### My--Dog Dreambooth model trained by BharatMata following the "Build your own Gen AI model" session by NxtWave.
Project Submission Code: Roll-No.27
Sample pictures of this concept:
|
alexbeta80/donut-test
|
alexbeta80
| 2024-02-12T15:22:31Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"vision-encoder-decoder",
"image-text-to-text",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:naver-clova-ix/donut-base",
"base_model:finetune:naver-clova-ix/donut-base",
"license:mit",
"endpoints_compatible",
"region:us"
] |
image-text-to-text
| 2024-02-05T14:27:14Z |
---
license: mit
base_model: naver-clova-ix/donut-base
tags:
- generated_from_trainer
datasets:
- imagefolder
model-index:
- name: donut-test
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# donut-test
This model is a fine-tuned version of [naver-clova-ix/donut-base](https://huggingface.co/naver-clova-ix/donut-base) on the imagefolder dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
### Framework versions
- Transformers 4.34.0
- Pytorch 2.1.0+cu121
- Datasets 2.14.5
- Tokenizers 0.14.1
|
sam1120/dropoff-utcustom-train-SF-RGB-b5_5
|
sam1120
| 2024-02-12T14:59:47Z | 150 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T14:29:57Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGB-b5_5
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGB-b5_5
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1911
- Mean Iou: 0.4677
- Mean Accuracy: 0.7472
- Overall Accuracy: 0.9719
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.5020
- Accuracy Undropoff: 0.9923
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.4318
- Iou Undropoff: 0.9713
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 9e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0685 | 5.0 | 10 | 1.0222 | 0.2189 | 0.3725 | 0.5989 | nan | 0.1256 | 0.6194 | 0.0 | 0.0497 | 0.6070 |
| 0.9481 | 10.0 | 20 | 0.8419 | 0.3703 | 0.6398 | 0.8451 | nan | 0.4159 | 0.8637 | 0.0 | 0.2633 | 0.8476 |
| 0.8268 | 15.0 | 30 | 0.7165 | 0.3949 | 0.6938 | 0.8694 | nan | 0.5023 | 0.8853 | 0.0 | 0.3136 | 0.8711 |
| 0.7573 | 20.0 | 40 | 0.6206 | 0.4084 | 0.7186 | 0.8994 | nan | 0.5214 | 0.9158 | 0.0 | 0.3243 | 0.9010 |
| 0.636 | 25.0 | 50 | 0.5194 | 0.4239 | 0.7253 | 0.9300 | nan | 0.5020 | 0.9485 | 0.0 | 0.3401 | 0.9316 |
| 0.5238 | 30.0 | 60 | 0.4507 | 0.4365 | 0.7368 | 0.9461 | nan | 0.5085 | 0.9651 | 0.0 | 0.3618 | 0.9476 |
| 0.4296 | 35.0 | 70 | 0.4064 | 0.4410 | 0.7422 | 0.9530 | nan | 0.5123 | 0.9721 | 0.0 | 0.3683 | 0.9546 |
| 0.4105 | 40.0 | 80 | 0.3547 | 0.4502 | 0.7467 | 0.9619 | nan | 0.5120 | 0.9814 | 0.0 | 0.3880 | 0.9627 |
| 0.3436 | 45.0 | 90 | 0.3304 | 0.4571 | 0.7596 | 0.9644 | nan | 0.5361 | 0.9830 | 0.0 | 0.4066 | 0.9647 |
| 0.2729 | 50.0 | 100 | 0.2953 | 0.4614 | 0.7552 | 0.9680 | nan | 0.5232 | 0.9873 | 0.0 | 0.4163 | 0.9678 |
| 0.2546 | 55.0 | 110 | 0.2770 | 0.4629 | 0.7579 | 0.9691 | nan | 0.5276 | 0.9882 | 0.0 | 0.4201 | 0.9686 |
| 0.2281 | 60.0 | 120 | 0.2591 | 0.4647 | 0.7566 | 0.9702 | nan | 0.5235 | 0.9896 | 0.0 | 0.4245 | 0.9696 |
| 0.2041 | 65.0 | 130 | 0.2453 | 0.4657 | 0.7556 | 0.9708 | nan | 0.5209 | 0.9903 | 0.0 | 0.4269 | 0.9701 |
| 0.1772 | 70.0 | 140 | 0.2292 | 0.4676 | 0.7542 | 0.9717 | nan | 0.5171 | 0.9914 | 0.0 | 0.4317 | 0.9711 |
| 0.169 | 75.0 | 150 | 0.2161 | 0.4681 | 0.7520 | 0.9719 | nan | 0.5122 | 0.9919 | 0.0 | 0.4331 | 0.9713 |
| 0.1543 | 80.0 | 160 | 0.2111 | 0.4682 | 0.7530 | 0.9715 | nan | 0.5147 | 0.9913 | 0.0 | 0.4336 | 0.9709 |
| 0.1374 | 85.0 | 170 | 0.1973 | 0.4659 | 0.7450 | 0.9715 | nan | 0.4980 | 0.9921 | 0.0 | 0.4268 | 0.9709 |
| 0.1523 | 90.0 | 180 | 0.1974 | 0.4681 | 0.7501 | 0.9717 | nan | 0.5085 | 0.9918 | 0.0 | 0.4332 | 0.9711 |
| 0.1323 | 95.0 | 190 | 0.1928 | 0.4658 | 0.7434 | 0.9717 | nan | 0.4944 | 0.9924 | 0.0 | 0.4263 | 0.9711 |
| 0.1254 | 100.0 | 200 | 0.1923 | 0.4671 | 0.7467 | 0.9717 | nan | 0.5013 | 0.9921 | 0.0 | 0.4301 | 0.9711 |
| 0.125 | 105.0 | 210 | 0.1867 | 0.4637 | 0.7380 | 0.9717 | nan | 0.4831 | 0.9929 | 0.0 | 0.4201 | 0.9711 |
| 0.1239 | 110.0 | 220 | 0.1912 | 0.4694 | 0.7520 | 0.9719 | nan | 0.5121 | 0.9919 | 0.0 | 0.4369 | 0.9713 |
| 0.1252 | 115.0 | 230 | 0.1913 | 0.4689 | 0.7503 | 0.9720 | nan | 0.5085 | 0.9921 | 0.0 | 0.4354 | 0.9714 |
| 0.1357 | 120.0 | 240 | 0.1911 | 0.4677 | 0.7472 | 0.9719 | nan | 0.5020 | 0.9923 | 0.0 | 0.4318 | 0.9713 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGB-b5_7
|
sam1120
| 2024-02-12T14:57:46Z | 146 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T14:26:14Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGB-b5_7
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGB-b5_7
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1841
- Mean Iou: 0.7025
- Mean Accuracy: 0.7532
- Overall Accuracy: 0.9721
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.5145
- Accuracy Undropoff: 0.9919
- Iou Unlabeled: nan
- Iou Dropoff: 0.4336
- Iou Undropoff: 0.9715
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 0.8255 | 5.0 | 10 | 0.7949 | 0.4128 | 0.7856 | 0.9393 | nan | 0.6179 | 0.9533 | 0.0 | 0.3007 | 0.9377 |
| 0.4434 | 10.0 | 20 | 0.4247 | 0.4471 | 0.7066 | 0.9705 | nan | 0.4187 | 0.9944 | 0.0 | 0.3714 | 0.9700 |
| 0.2107 | 15.0 | 30 | 0.2726 | 0.6711 | 0.7003 | 0.9715 | nan | 0.4046 | 0.9961 | nan | 0.3713 | 0.9710 |
| 0.1678 | 20.0 | 40 | 0.2388 | 0.6801 | 0.7343 | 0.9691 | nan | 0.4782 | 0.9904 | nan | 0.3917 | 0.9685 |
| 0.0972 | 25.0 | 50 | 0.1849 | 0.6764 | 0.7096 | 0.9715 | nan | 0.4241 | 0.9952 | nan | 0.3818 | 0.9709 |
| 0.0604 | 30.0 | 60 | 0.2019 | 0.4644 | 0.7568 | 0.9704 | nan | 0.5239 | 0.9897 | 0.0 | 0.4236 | 0.9697 |
| 0.0497 | 35.0 | 70 | 0.1793 | 0.6838 | 0.7345 | 0.9700 | nan | 0.4775 | 0.9914 | nan | 0.3983 | 0.9694 |
| 0.0492 | 40.0 | 80 | 0.2000 | 0.4639 | 0.7567 | 0.9702 | nan | 0.5239 | 0.9896 | 0.0 | 0.4223 | 0.9695 |
| 0.0409 | 45.0 | 90 | 0.1893 | 0.7030 | 0.7778 | 0.9696 | nan | 0.5687 | 0.9869 | nan | 0.4372 | 0.9688 |
| 0.0328 | 50.0 | 100 | 0.1842 | 0.7040 | 0.7715 | 0.9704 | nan | 0.5545 | 0.9885 | nan | 0.4382 | 0.9697 |
| 0.0332 | 55.0 | 110 | 0.1781 | 0.7015 | 0.7563 | 0.9715 | nan | 0.5216 | 0.9910 | nan | 0.4322 | 0.9709 |
| 0.0314 | 60.0 | 120 | 0.1732 | 0.6890 | 0.7305 | 0.9717 | nan | 0.4675 | 0.9935 | nan | 0.4068 | 0.9711 |
| 0.0318 | 65.0 | 130 | 0.1786 | 0.6971 | 0.7477 | 0.9715 | nan | 0.5037 | 0.9918 | nan | 0.4233 | 0.9709 |
| 0.0291 | 70.0 | 140 | 0.1814 | 0.7119 | 0.7687 | 0.9725 | nan | 0.5466 | 0.9909 | nan | 0.4521 | 0.9718 |
| 0.0273 | 75.0 | 150 | 0.1755 | 0.7101 | 0.7677 | 0.9722 | nan | 0.5446 | 0.9907 | nan | 0.4487 | 0.9715 |
| 0.0274 | 80.0 | 160 | 0.1786 | 0.7006 | 0.7494 | 0.9720 | nan | 0.5066 | 0.9922 | nan | 0.4297 | 0.9714 |
| 0.0248 | 85.0 | 170 | 0.1741 | 0.7029 | 0.7526 | 0.9722 | nan | 0.5131 | 0.9921 | nan | 0.4341 | 0.9716 |
| 0.0248 | 90.0 | 180 | 0.1832 | 0.7050 | 0.7595 | 0.9719 | nan | 0.5278 | 0.9912 | nan | 0.4387 | 0.9713 |
| 0.0242 | 95.0 | 190 | 0.1808 | 0.7028 | 0.7539 | 0.9720 | nan | 0.5160 | 0.9918 | nan | 0.4341 | 0.9714 |
| 0.024 | 100.0 | 200 | 0.1796 | 0.7022 | 0.7501 | 0.9723 | nan | 0.5077 | 0.9925 | nan | 0.4327 | 0.9717 |
| 0.0231 | 105.0 | 210 | 0.1835 | 0.7137 | 0.7731 | 0.9724 | nan | 0.5557 | 0.9905 | nan | 0.4556 | 0.9717 |
| 0.0238 | 110.0 | 220 | 0.1823 | 0.7046 | 0.7565 | 0.9721 | nan | 0.5214 | 0.9917 | nan | 0.4376 | 0.9715 |
| 0.0228 | 115.0 | 230 | 0.1833 | 0.7009 | 0.7504 | 0.9720 | nan | 0.5088 | 0.9921 | nan | 0.4305 | 0.9714 |
| 0.0255 | 120.0 | 240 | 0.1841 | 0.7025 | 0.7532 | 0.9721 | nan | 0.5145 | 0.9919 | nan | 0.4336 | 0.9715 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
jaCappella/DPTNet_jaCappella_VES_48k
|
jaCappella
| 2024-02-12T14:55:00Z | 0 | 0 | null |
[
"music",
"speech",
"audio",
"audio-to-audio",
"a cappella",
"vocal ensemble",
"ja",
"dataset:jaCappella",
"arxiv:2211.16028",
"license:cc-by-nc-4.0",
"region:us"
] |
audio-to-audio
| 2023-01-21T06:42:35Z |
---
license: cc-by-nc-4.0
language:
- ja
tags:
- music
- speech
- audio
- audio-to-audio
- a cappella
- vocal ensemble
datasets:
- jaCappella
metrics:
- SI-SDR
---
# DPTNet trained with the jaCappella corpus for vocal ensemble separation
This model was trained by Tomohiko Nakamura using [the codebase](https://github.com/TomohikoNakamura/asteroid_jaCappella)).
It was trained on the vocal ensemble separation task of [the jaCappella dataset](https://tomohikonakamura.github.io/jaCappella_corpus/).
[The paper](https://doi.org/10.1109/ICASSP49357.2023.10095569) was published in ICASSP 2023 ([arXiv](https://arxiv.org/abs/2211.16028)).
# License
See [the jaCappella dataset page](https://tomohikonakamura.github.io/jaCappella_corpus/).
# Citation
See [the jaCappella dataset page](https://tomohikonakamura.github.io/jaCappella_corpus/).
# Configuration
```yaml
data:
num_workers: 12
sample_rate: 48000
samples_per_track: 13
seed: 42
seq_dur: 5.046
source_augmentations:
- gain
sources:
- vocal_percussion
- bass
- alto
- tenor
- soprano
- lead_vocal
filterbank:
kernel_size: 32
n_filters: 64
stride: 16
masknet:
bidirectional: true
chunk_size: 174
dropout: 0
ff_activation: relu
ff_hid: 256
hop_size: 128
in_chan: 64
mask_act: sigmoid
n_repeats: 8
n_src: 6
norm_type: gLN
out_chan: 64
optim:
lr: 0.005
optimizer: adam
weight_decay: 1.0e-05
training:
batch_size: 1
early_stop: true
epochs: 600
gradient_clipping: 5
half_lr: true
loss_func: pit_sisdr
```
# Results (SI-SDR [dB]) on vocal ensemble separation
| Method | Lead vocal | Soprano | Alto | Tenor | Bass |Vocal percussion|
|:---------------:|:--------------:|:--------------:|:--------------:|:--------------:|:--------------:|:--------------:|
| DPTNet | 8.9 | 8.5 | 11.9 | 14.9 | 19.7 | 21.9 |
|
Kavin0211/results
|
Kavin0211
| 2024-02-12T14:54:59Z | 0 | 0 |
peft
|
[
"peft",
"tensorboard",
"safetensors",
"trl",
"sft",
"generated_from_trainer",
"base_model:microsoft/phi-2",
"base_model:adapter:microsoft/phi-2",
"license:mit",
"region:us"
] | null | 2024-02-12T14:54:51Z |
---
license: mit
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: microsoft/phi-2
model-index:
- name: results
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# results
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- training_steps: 500
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- PEFT 0.8.2
- Transformers 4.37.2
- Pytorch 2.2.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
ddamir/boris_boy
|
ddamir
| 2024-02-12T14:51:42Z | 0 | 0 | null |
[
"license:other",
"region:us"
] | null | 2024-02-12T14:48:48Z |
---
license: other
license_name: boy
license_link: LICENSE
---
|
Shijia/furina_seed42_eng_amh_esp_roman
|
Shijia
| 2024-02-12T14:51:27Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:yihongLiu/furina",
"base_model:finetune:yihongLiu/furina",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T14:50:32Z |
---
base_model: yihongLiu/furina
tags:
- generated_from_trainer
model-index:
- name: furina_seed42_eng_amh_esp_roman
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# furina_seed42_eng_amh_esp_roman
This model is a fine-tuned version of [yihongLiu/furina](https://huggingface.co/yihongLiu/furina) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0144
- Spearman Corr: 0.8461
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 128
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Spearman Corr |
|:-------------:|:-----:|:----:|:---------------:|:-------------:|
| No log | 0.59 | 200 | 0.0299 | 0.6782 |
| No log | 1.18 | 400 | 0.0251 | 0.7278 |
| No log | 1.76 | 600 | 0.0202 | 0.7493 |
| 0.0425 | 2.35 | 800 | 0.0194 | 0.7584 |
| 0.0425 | 2.94 | 1000 | 0.0184 | 0.7737 |
| 0.0425 | 3.53 | 1200 | 0.0189 | 0.7734 |
| 0.0184 | 4.12 | 1400 | 0.0180 | 0.7906 |
| 0.0184 | 4.71 | 1600 | 0.0188 | 0.7909 |
| 0.0184 | 5.29 | 1800 | 0.0171 | 0.7971 |
| 0.0184 | 5.88 | 2000 | 0.0165 | 0.8055 |
| 0.0134 | 6.47 | 2200 | 0.0162 | 0.8059 |
| 0.0134 | 7.06 | 2400 | 0.0164 | 0.8085 |
| 0.0134 | 7.65 | 2600 | 0.0169 | 0.8131 |
| 0.0098 | 8.24 | 2800 | 0.0169 | 0.8171 |
| 0.0098 | 8.82 | 3000 | 0.0158 | 0.8169 |
| 0.0098 | 9.41 | 3200 | 0.0152 | 0.8201 |
| 0.0073 | 10.0 | 3400 | 0.0165 | 0.8197 |
| 0.0073 | 10.59 | 3600 | 0.0150 | 0.8234 |
| 0.0073 | 11.18 | 3800 | 0.0152 | 0.8284 |
| 0.0073 | 11.76 | 4000 | 0.0141 | 0.8338 |
| 0.0059 | 12.35 | 4200 | 0.0144 | 0.8315 |
| 0.0059 | 12.94 | 4400 | 0.0147 | 0.8348 |
| 0.0059 | 13.53 | 4600 | 0.0157 | 0.8327 |
| 0.0049 | 14.12 | 4800 | 0.0147 | 0.8379 |
| 0.0049 | 14.71 | 5000 | 0.0149 | 0.8365 |
| 0.0049 | 15.29 | 5200 | 0.0142 | 0.8360 |
| 0.0049 | 15.88 | 5400 | 0.0140 | 0.8409 |
| 0.0042 | 16.47 | 5600 | 0.0135 | 0.8414 |
| 0.0042 | 17.06 | 5800 | 0.0141 | 0.8410 |
| 0.0042 | 17.65 | 6000 | 0.0144 | 0.8402 |
| 0.0037 | 18.24 | 6200 | 0.0151 | 0.8435 |
| 0.0037 | 18.82 | 6400 | 0.0140 | 0.8431 |
| 0.0037 | 19.41 | 6600 | 0.0140 | 0.8454 |
| 0.0033 | 20.0 | 6800 | 0.0136 | 0.8453 |
| 0.0033 | 20.59 | 7000 | 0.0137 | 0.8446 |
| 0.0033 | 21.18 | 7200 | 0.0144 | 0.8461 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
Commandante/german-party-sentiment-bert-complete-synonyms-5e-5
|
Commandante
| 2024-02-12T14:45:39Z | 93 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"text-classification",
"generated_from_trainer",
"base_model:mdraw/german-news-sentiment-bert",
"base_model:finetune:mdraw/german-news-sentiment-bert",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-09T02:21:11Z |
---
base_model: mdraw/german-news-sentiment-bert
tags:
- generated_from_trainer
model-index:
- name: german-party-sentiment-bert-complete-synonyms-5e-5
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# german-party-sentiment-bert-complete-synonyms-5e-5
This model is a fine-tuned version of [mdraw/german-news-sentiment-bert](https://huggingface.co/mdraw/german-news-sentiment-bert) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8769
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 20
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 120
- num_epochs: 14
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 0.9596 | 1.0 | 70 | 0.9676 |
| 0.9122 | 2.0 | 140 | 0.8769 |
| 0.7382 | 3.0 | 210 | 0.9984 |
| 0.5708 | 4.0 | 280 | 1.1080 |
| 0.3579 | 5.0 | 350 | 1.4137 |
| 0.3066 | 6.0 | 420 | 1.8204 |
| 0.1716 | 7.0 | 490 | 1.8167 |
| 0.1974 | 8.0 | 560 | 2.1479 |
| 0.1164 | 9.0 | 630 | 2.3899 |
| 0.0878 | 10.0 | 700 | 2.5266 |
| 0.07 | 11.0 | 770 | 2.7014 |
| 0.0604 | 12.0 | 840 | 2.7048 |
| 0.0278 | 13.0 | 910 | 2.8119 |
| 0.0376 | 14.0 | 980 | 2.8799 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.2.0+cu121
- Tokenizers 0.15.1
|
alitolga/electra-base-generator-rank64
|
alitolga
| 2024-02-12T14:41:55Z | 0 | 0 | null |
[
"safetensors",
"generated_from_trainer",
"base_model:google/electra-base-generator",
"base_model:finetune:google/electra-base-generator",
"license:apache-2.0",
"region:us"
] | null | 2024-02-12T14:41:00Z |
---
license: apache-2.0
base_model: google/electra-base-generator
tags:
- generated_from_trainer
model-index:
- name: electra-base-generator-rank64
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# electra-base-generator-rank64
This model is a fine-tuned version of [google/electra-base-generator](https://huggingface.co/google/electra-base-generator) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2951
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 8.2066 | 1.0 | 179 | 3.8785 |
| 3.6834 | 2.0 | 358 | 3.3549 |
| 3.4351 | 3.0 | 537 | 3.2951 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
sam1120/dropoff-utcustom-train-SF-RGB-b5_2
|
sam1120
| 2024-02-12T14:41:07Z | 151 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T14:24:47Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGB-b5_2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGB-b5_2
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4848
- Mean Iou: 0.4257
- Mean Accuracy: 0.7972
- Overall Accuracy: 0.9466
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.6343
- Accuracy Undropoff: 0.9601
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.3321
- Iou Undropoff: 0.9451
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0108 | 5.0 | 10 | 1.0721 | 0.1514 | 0.5401 | 0.4205 | nan | 0.6706 | 0.4096 | 0.0 | 0.0494 | 0.4047 |
| 0.9654 | 10.0 | 20 | 0.9802 | 0.2190 | 0.6570 | 0.5944 | nan | 0.7253 | 0.5887 | 0.0 | 0.0745 | 0.5826 |
| 0.9175 | 15.0 | 30 | 0.9047 | 0.2553 | 0.7350 | 0.6792 | nan | 0.7960 | 0.6741 | 0.0 | 0.0973 | 0.6686 |
| 0.9052 | 20.0 | 40 | 0.8427 | 0.2812 | 0.7661 | 0.7377 | nan | 0.7971 | 0.7351 | 0.0 | 0.1146 | 0.7290 |
| 0.8555 | 25.0 | 50 | 0.7970 | 0.3063 | 0.7827 | 0.7900 | nan | 0.7748 | 0.7906 | 0.0 | 0.1357 | 0.7832 |
| 0.8291 | 30.0 | 60 | 0.7543 | 0.3289 | 0.7891 | 0.8332 | nan | 0.7410 | 0.8372 | 0.0 | 0.1586 | 0.8282 |
| 0.7923 | 35.0 | 70 | 0.7327 | 0.3375 | 0.7961 | 0.8471 | nan | 0.7405 | 0.8517 | 0.0 | 0.1701 | 0.8425 |
| 0.7724 | 40.0 | 80 | 0.6994 | 0.3529 | 0.7968 | 0.8719 | nan | 0.7149 | 0.8787 | 0.0 | 0.1906 | 0.8682 |
| 0.7215 | 45.0 | 90 | 0.6675 | 0.3694 | 0.7935 | 0.8954 | nan | 0.6824 | 0.9047 | 0.0 | 0.2157 | 0.8926 |
| 0.6907 | 50.0 | 100 | 0.6521 | 0.3742 | 0.7998 | 0.9000 | nan | 0.6904 | 0.9091 | 0.0 | 0.2252 | 0.8973 |
| 0.6768 | 55.0 | 110 | 0.6260 | 0.3850 | 0.8022 | 0.9118 | nan | 0.6827 | 0.9217 | 0.0 | 0.2455 | 0.9094 |
| 0.659 | 60.0 | 120 | 0.6010 | 0.3965 | 0.7973 | 0.9244 | nan | 0.6586 | 0.9359 | 0.0 | 0.2671 | 0.9224 |
| 0.6265 | 65.0 | 130 | 0.5847 | 0.4005 | 0.7992 | 0.9276 | nan | 0.6592 | 0.9393 | 0.0 | 0.2757 | 0.9258 |
| 0.6134 | 70.0 | 140 | 0.5673 | 0.4060 | 0.8022 | 0.9316 | nan | 0.6611 | 0.9433 | 0.0 | 0.2881 | 0.9297 |
| 0.5864 | 75.0 | 150 | 0.5401 | 0.4132 | 0.7961 | 0.9383 | nan | 0.6410 | 0.9511 | 0.0 | 0.3029 | 0.9366 |
| 0.5686 | 80.0 | 160 | 0.5289 | 0.4153 | 0.7974 | 0.9395 | nan | 0.6424 | 0.9524 | 0.0 | 0.3080 | 0.9379 |
| 0.5597 | 85.0 | 170 | 0.5386 | 0.4114 | 0.8079 | 0.9350 | nan | 0.6692 | 0.9465 | 0.0 | 0.3011 | 0.9331 |
| 0.5718 | 90.0 | 180 | 0.5080 | 0.4210 | 0.7947 | 0.9438 | nan | 0.6321 | 0.9573 | 0.0 | 0.3208 | 0.9423 |
| 0.517 | 95.0 | 190 | 0.5026 | 0.4222 | 0.7956 | 0.9445 | nan | 0.6332 | 0.9580 | 0.0 | 0.3236 | 0.9430 |
| 0.5252 | 100.0 | 200 | 0.4990 | 0.4232 | 0.7969 | 0.9450 | nan | 0.6354 | 0.9584 | 0.0 | 0.3261 | 0.9435 |
| 0.5174 | 105.0 | 210 | 0.4951 | 0.4223 | 0.8012 | 0.9437 | nan | 0.6457 | 0.9567 | 0.0 | 0.3249 | 0.9422 |
| 0.5217 | 110.0 | 220 | 0.4882 | 0.4238 | 0.7993 | 0.9450 | nan | 0.6404 | 0.9582 | 0.0 | 0.3280 | 0.9435 |
| 0.5224 | 115.0 | 230 | 0.4846 | 0.4258 | 0.7968 | 0.9467 | nan | 0.6333 | 0.9603 | 0.0 | 0.3321 | 0.9452 |
| 0.5399 | 120.0 | 240 | 0.4848 | 0.4257 | 0.7972 | 0.9466 | nan | 0.6343 | 0.9601 | 0.0 | 0.3321 | 0.9451 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGB-b5_1
|
sam1120
| 2024-02-12T14:40:35Z | 147 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T14:24:17Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGB-b5_1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGB-b5_1
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6279
- Mean Iou: 0.4054
- Mean Accuracy: 0.7471
- Overall Accuracy: 0.8860
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.5956
- Accuracy Undropoff: 0.8986
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.3318
- Iou Undropoff: 0.8843
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0071 | 5.0 | 10 | 1.0206 | 0.1745 | 0.2748 | 0.5034 | nan | 0.0255 | 0.5241 | 0.0 | 0.0147 | 0.5087 |
| 0.9688 | 10.0 | 20 | 0.9873 | 0.2140 | 0.3486 | 0.5771 | nan | 0.0992 | 0.5979 | 0.0 | 0.0582 | 0.5838 |
| 0.9406 | 15.0 | 30 | 0.9313 | 0.2613 | 0.4446 | 0.6655 | nan | 0.2038 | 0.6855 | 0.0 | 0.1135 | 0.6705 |
| 0.9278 | 20.0 | 40 | 0.8851 | 0.2930 | 0.5149 | 0.7111 | nan | 0.3009 | 0.7289 | 0.0 | 0.1648 | 0.7142 |
| 0.8956 | 25.0 | 50 | 0.8563 | 0.3118 | 0.5642 | 0.7358 | nan | 0.3770 | 0.7514 | 0.0 | 0.1985 | 0.7370 |
| 0.8674 | 30.0 | 60 | 0.8260 | 0.3303 | 0.6086 | 0.7664 | nan | 0.4366 | 0.7807 | 0.0 | 0.2246 | 0.7664 |
| 0.8438 | 35.0 | 70 | 0.8149 | 0.3347 | 0.6355 | 0.7671 | nan | 0.4921 | 0.7790 | 0.0 | 0.2381 | 0.7660 |
| 0.8309 | 40.0 | 80 | 0.7881 | 0.3459 | 0.6472 | 0.7847 | nan | 0.4972 | 0.7972 | 0.0 | 0.2539 | 0.7839 |
| 0.8069 | 45.0 | 90 | 0.7640 | 0.3567 | 0.6617 | 0.8041 | nan | 0.5063 | 0.8170 | 0.0 | 0.2668 | 0.8033 |
| 0.7779 | 50.0 | 100 | 0.7486 | 0.3637 | 0.6792 | 0.8145 | nan | 0.5316 | 0.8268 | 0.0 | 0.2778 | 0.8132 |
| 0.7695 | 55.0 | 110 | 0.7354 | 0.3684 | 0.6936 | 0.8214 | nan | 0.5542 | 0.8329 | 0.0 | 0.2858 | 0.8195 |
| 0.7568 | 60.0 | 120 | 0.7164 | 0.3757 | 0.7032 | 0.8365 | nan | 0.5577 | 0.8486 | 0.0 | 0.2924 | 0.8347 |
| 0.7285 | 65.0 | 130 | 0.6976 | 0.3836 | 0.7119 | 0.8484 | nan | 0.5630 | 0.8608 | 0.0 | 0.3042 | 0.8467 |
| 0.7217 | 70.0 | 140 | 0.6922 | 0.3857 | 0.7217 | 0.8499 | nan | 0.5817 | 0.8616 | 0.0 | 0.3091 | 0.8480 |
| 0.7095 | 75.0 | 150 | 0.6708 | 0.3926 | 0.7287 | 0.8624 | nan | 0.5828 | 0.8745 | 0.0 | 0.3172 | 0.8605 |
| 0.6944 | 80.0 | 160 | 0.6637 | 0.3951 | 0.7320 | 0.8660 | nan | 0.5858 | 0.8781 | 0.0 | 0.3212 | 0.8641 |
| 0.6878 | 85.0 | 170 | 0.6632 | 0.3942 | 0.7397 | 0.8673 | nan | 0.6005 | 0.8788 | 0.0 | 0.3175 | 0.8652 |
| 0.6868 | 90.0 | 180 | 0.6468 | 0.3998 | 0.7391 | 0.8756 | nan | 0.5902 | 0.8880 | 0.0 | 0.3257 | 0.8739 |
| 0.6581 | 95.0 | 190 | 0.6444 | 0.4003 | 0.7421 | 0.8776 | nan | 0.5942 | 0.8899 | 0.0 | 0.3249 | 0.8759 |
| 0.6587 | 100.0 | 200 | 0.6383 | 0.4026 | 0.7427 | 0.8814 | nan | 0.5914 | 0.8940 | 0.0 | 0.3281 | 0.8797 |
| 0.6525 | 105.0 | 210 | 0.6334 | 0.4032 | 0.7434 | 0.8825 | nan | 0.5918 | 0.8951 | 0.0 | 0.3289 | 0.8808 |
| 0.658 | 110.0 | 220 | 0.6345 | 0.4026 | 0.7451 | 0.8811 | nan | 0.5968 | 0.8934 | 0.0 | 0.3285 | 0.8793 |
| 0.6575 | 115.0 | 230 | 0.6300 | 0.4050 | 0.7463 | 0.8851 | nan | 0.5948 | 0.8977 | 0.0 | 0.3314 | 0.8835 |
| 0.6625 | 120.0 | 240 | 0.6279 | 0.4054 | 0.7471 | 0.8860 | nan | 0.5956 | 0.8986 | 0.0 | 0.3318 | 0.8843 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
hugo-massonnat/q-FrozenLake-v1-4x4-noSlippery
|
hugo-massonnat
| 2024-02-12T14:37:02Z | 0 | 0 | null |
[
"FrozenLake-v1-4x4-no_slippery",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T14:37:00Z |
---
tags:
- FrozenLake-v1-4x4-no_slippery
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-FrozenLake-v1-4x4-noSlippery
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: FrozenLake-v1-4x4-no_slippery
type: FrozenLake-v1-4x4-no_slippery
metrics:
- type: mean_reward
value: 1.00 +/- 0.00
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **FrozenLake-v1**
This is a trained model of a **Q-Learning** agent playing **FrozenLake-v1** .
## Usage
```python
model = load_from_hub(repo_id="hugo-massonnat/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
nashirab/vit-base-beans
|
nashirab
| 2024-02-12T14:28:31Z | 179 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"vit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"base_model:google/vit-base-patch16-224",
"base_model:finetune:google/vit-base-patch16-224",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2024-02-12T14:28:14Z |
---
license: apache-2.0
base_model: google/vit-base-patch16-224
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: vit-base-beans
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.49375
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# vit-base-beans
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2658
- Accuracy: 0.4938
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 4.7295 | 0.25 | 10 | 2.7467 | 0.1875 |
| 2.3133 | 0.5 | 20 | 2.1258 | 0.2437 |
| 2.031 | 0.75 | 30 | 1.9442 | 0.3187 |
| 1.8773 | 1.0 | 40 | 1.6159 | 0.375 |
| 1.4132 | 1.25 | 50 | 1.5585 | 0.4188 |
| 1.4581 | 1.5 | 60 | 1.5269 | 0.35 |
| 1.4697 | 1.75 | 70 | 1.5535 | 0.3625 |
| 1.3575 | 2.0 | 80 | 1.3056 | 0.4375 |
| 1.0615 | 2.25 | 90 | 1.4774 | 0.4 |
| 1.1895 | 2.5 | 100 | 1.2384 | 0.45 |
| 1.0659 | 2.75 | 110 | 1.3315 | 0.4938 |
| 1.1517 | 3.0 | 120 | 1.1040 | 0.575 |
| 0.7957 | 3.25 | 130 | 1.3480 | 0.4375 |
| 0.8037 | 3.5 | 140 | 1.2879 | 0.525 |
| 1.0157 | 3.75 | 150 | 1.1900 | 0.5 |
| 0.7665 | 4.0 | 160 | 1.2039 | 0.4938 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
sophiayk20/bert-finetuned-squad
|
sophiayk20
| 2024-02-12T14:24:38Z | 112 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"bert",
"question-answering",
"generated_from_trainer",
"base_model:google-bert/bert-base-cased",
"base_model:finetune:google-bert/bert-base-cased",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] |
question-answering
| 2024-02-12T11:32:45Z |
---
license: apache-2.0
base_model: bert-base-cased
tags:
- generated_from_trainer
model-index:
- name: bert-finetuned-squad
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-finetuned-squad
This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
xncy/q-FrozenLake-v1-4x4-noSlippery
|
xncy
| 2024-02-12T14:24:11Z | 0 | 0 | null |
[
"FrozenLake-v1-4x4-no_slippery",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T14:24:08Z |
---
tags:
- FrozenLake-v1-4x4-no_slippery
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-FrozenLake-v1-4x4-noSlippery
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: FrozenLake-v1-4x4-no_slippery
type: FrozenLake-v1-4x4-no_slippery
metrics:
- type: mean_reward
value: 1.00 +/- 0.00
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **FrozenLake-v1**
This is a trained model of a **Q-Learning** agent playing **FrozenLake-v1** .
## Usage
```python
model = load_from_hub(repo_id="xncy/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
Guilherme34/Jennifer-uwu-version
|
Guilherme34
| 2024-02-12T14:23:35Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"arxiv:1910.09700",
"endpoints_compatible",
"region:us"
] | null | 2024-02-12T14:23:32Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
jaykchen/tiny
|
jaykchen
| 2024-02-12T14:20:30Z | 94 | 0 |
transformers
|
[
"transformers",
"safetensors",
"llama",
"text-generation",
"text-generation-inference",
"unsloth",
"trl",
"en",
"base_model:unsloth/tinyllama-bnb-4bit",
"base_model:finetune:unsloth/tinyllama-bnb-4bit",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-12T14:19:02Z |
---
language:
- en
license: apache-2.0
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
base_model: unsloth/tinyllama-bnb-4bit
---
# Uploaded model
- **Developed by:** jaykchen
- **License:** apache-2.0
- **Finetuned from model :** unsloth/tinyllama-bnb-4bit
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
Shijia/furina_seed42_eng_kin_amh_roman
|
Shijia
| 2024-02-12T14:19:22Z | 91 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:yihongLiu/furina",
"base_model:finetune:yihongLiu/furina",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T14:18:30Z |
---
base_model: yihongLiu/furina
tags:
- generated_from_trainer
model-index:
- name: furina_seed42_eng_kin_amh_roman
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# furina_seed42_eng_kin_amh_roman
This model is a fine-tuned version of [yihongLiu/furina](https://huggingface.co/yihongLiu/furina) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0284
- Spearman Corr: 0.7771
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 128
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Spearman Corr |
|:-------------:|:-----:|:----:|:---------------:|:-------------:|
| No log | 0.65 | 200 | 0.0373 | 0.5747 |
| No log | 1.3 | 400 | 0.0297 | 0.6851 |
| No log | 1.95 | 600 | 0.0311 | 0.7236 |
| 0.0545 | 2.61 | 800 | 0.0305 | 0.7322 |
| 0.0545 | 3.26 | 1000 | 0.0281 | 0.7496 |
| 0.0545 | 3.91 | 1200 | 0.0278 | 0.7582 |
| 0.0208 | 4.56 | 1400 | 0.0278 | 0.7528 |
| 0.0208 | 5.21 | 1600 | 0.0238 | 0.7556 |
| 0.0208 | 5.86 | 1800 | 0.0235 | 0.7631 |
| 0.0143 | 6.51 | 2000 | 0.0245 | 0.7634 |
| 0.0143 | 7.17 | 2200 | 0.0243 | 0.7619 |
| 0.0143 | 7.82 | 2400 | 0.0242 | 0.7651 |
| 0.0102 | 8.47 | 2600 | 0.0257 | 0.7645 |
| 0.0102 | 9.12 | 2800 | 0.0271 | 0.7713 |
| 0.0102 | 9.77 | 3000 | 0.0255 | 0.7661 |
| 0.0079 | 10.42 | 3200 | 0.0218 | 0.7720 |
| 0.0079 | 11.07 | 3400 | 0.0250 | 0.7658 |
| 0.0079 | 11.73 | 3600 | 0.0266 | 0.7628 |
| 0.0064 | 12.38 | 3800 | 0.0267 | 0.7657 |
| 0.0064 | 13.03 | 4000 | 0.0261 | 0.7680 |
| 0.0064 | 13.68 | 4200 | 0.0232 | 0.7720 |
| 0.0055 | 14.33 | 4400 | 0.0256 | 0.7737 |
| 0.0055 | 14.98 | 4600 | 0.0237 | 0.7736 |
| 0.0055 | 15.64 | 4800 | 0.0284 | 0.7771 |
### Framework versions
- Transformers 4.37.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
Augusto777/swin-tiny-patch4-window7-224-dmae-va-U
|
Augusto777
| 2024-02-12T14:09:53Z | 201 | 1 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"swin",
"image-classification",
"generated_from_trainer",
"dataset:Augusto777/dmae-U",
"base_model:microsoft/swin-tiny-patch4-window7-224",
"base_model:finetune:microsoft/swin-tiny-patch4-window7-224",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2023-12-13T20:40:53Z |
---
license: apache-2.0
base_model: microsoft/swin-tiny-patch4-window7-224
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: swin-tiny-patch4-window7-224-dmae-va-U
results: []
datasets:
- Augusto777/dmae-U
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# swin-tiny-patch4-window7-224-dmae-va-U
This model is a fine-tuned version of [microsoft/swin-tiny-patch4-window7-224](https://huggingface.co/microsoft/swin-tiny-patch4-window7-224) on an AMD dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0900
- Accuracy: 0.9725
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 40
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 0.9 | 7 | 1.4643 | 0.2110 |
| 1.4218 | 1.94 | 15 | 1.4070 | 0.3303 |
| 1.3226 | 2.97 | 23 | 1.3454 | 0.3486 |
| 1.1908 | 4.0 | 31 | 1.1438 | 0.4220 |
| 1.1908 | 4.9 | 38 | 0.8730 | 0.5780 |
| 0.9441 | 5.94 | 46 | 0.8100 | 0.6422 |
| 0.7185 | 6.97 | 54 | 0.6099 | 0.7339 |
| 0.6526 | 8.0 | 62 | 0.5096 | 0.7890 |
| 0.6526 | 8.9 | 69 | 0.4925 | 0.8165 |
| 0.5185 | 9.94 | 77 | 0.3989 | 0.8349 |
| 0.4946 | 10.97 | 85 | 0.3276 | 0.8807 |
| 0.4469 | 12.0 | 93 | 0.3023 | 0.8899 |
| 0.376 | 12.9 | 100 | 0.3112 | 0.8991 |
| 0.376 | 13.94 | 108 | 0.2117 | 0.9266 |
| 0.3156 | 14.97 | 116 | 0.2024 | 0.9174 |
| 0.366 | 16.0 | 124 | 0.2065 | 0.9450 |
| 0.2806 | 16.9 | 131 | 0.1942 | 0.9174 |
| 0.2806 | 17.94 | 139 | 0.2393 | 0.9174 |
| 0.2695 | 18.97 | 147 | 0.1498 | 0.9541 |
| 0.2357 | 20.0 | 155 | 0.1465 | 0.9358 |
| 0.2345 | 20.9 | 162 | 0.1522 | 0.9633 |
| 0.2157 | 21.94 | 170 | 0.1403 | 0.9450 |
| 0.2157 | 22.97 | 178 | 0.0999 | 0.9541 |
| 0.1894 | 24.0 | 186 | 0.1427 | 0.9633 |
| 0.2195 | 24.9 | 193 | 0.0949 | 0.9633 |
| 0.1874 | 25.94 | 201 | 0.1152 | 0.9633 |
| 0.1874 | 26.97 | 209 | 0.1226 | 0.9541 |
| 0.1815 | 28.0 | 217 | 0.0964 | 0.9633 |
| 0.1619 | 28.9 | 224 | 0.0912 | 0.9633 |
| 0.201 | 29.94 | 232 | 0.0903 | 0.9633 |
| 0.1659 | 30.97 | 240 | 0.0745 | 0.9633 |
| 0.1659 | 32.0 | 248 | 0.0781 | 0.9633 |
| 0.1459 | 32.9 | 255 | 0.0930 | 0.9633 |
| 0.1459 | 33.94 | 263 | 0.0900 | 0.9725 |
| 0.1487 | 34.97 | 271 | 0.0796 | 0.9725 |
| 0.1487 | 36.0 | 279 | 0.0784 | 0.9725 |
| 0.1504 | 36.13 | 280 | 0.0784 | 0.9725 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
NBA55/llama2-7B-without-diversity-epoch-10-new
|
NBA55
| 2024-02-12T14:09:12Z | 0 | 0 |
peft
|
[
"peft",
"region:us"
] | null | 2024-02-12T14:09:02Z |
---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: False
- bnb_4bit_compute_dtype: float16
### Framework versions
- PEFT 0.4.0
|
alitolga/electra-base-generator-rank16
|
alitolga
| 2024-02-12T14:00:44Z | 0 | 0 | null |
[
"safetensors",
"generated_from_trainer",
"base_model:google/electra-base-generator",
"base_model:finetune:google/electra-base-generator",
"license:apache-2.0",
"region:us"
] | null | 2024-02-12T13:59:43Z |
---
license: apache-2.0
base_model: google/electra-base-generator
tags:
- generated_from_trainer
model-index:
- name: electra-base-generator-rank16
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# electra-base-generator-rank16
This model is a fine-tuned version of [google/electra-base-generator](https://huggingface.co/google/electra-base-generator) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2684
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 8.2873 | 1.0 | 179 | 3.8892 |
| 3.6837 | 2.0 | 358 | 3.3411 |
| 3.4127 | 3.0 | 537 | 3.2684 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
sam1120/dropoff-utcustom-train-SF-RGBD-b5_7
|
sam1120
| 2024-02-12T13:58:42Z | 148 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T13:25:26Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b5_7
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b5_7
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1296
- Mean Iou: 0.6242
- Mean Accuracy: 0.6623
- Overall Accuracy: 0.9652
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.3319
- Accuracy Undropoff: 0.9926
- Iou Unlabeled: nan
- Iou Dropoff: 0.2838
- Iou Undropoff: 0.9647
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 0.9278 | 5.0 | 10 | 0.8454 | 0.3197 | 0.5545 | 0.8788 | nan | 0.2009 | 0.9082 | 0.0 | 0.0807 | 0.8785 |
| 0.5551 | 10.0 | 20 | 0.4668 | 0.3221 | 0.5042 | 0.9540 | nan | 0.0135 | 0.9948 | 0.0 | 0.0122 | 0.9540 |
| 0.3667 | 15.0 | 30 | 0.3354 | 0.3218 | 0.5035 | 0.9570 | nan | 0.0088 | 0.9982 | 0.0 | 0.0085 | 0.9570 |
| 0.2402 | 20.0 | 40 | 0.2678 | 0.5985 | 0.6492 | 0.9587 | nan | 0.3116 | 0.9868 | nan | 0.2388 | 0.9582 |
| 0.1562 | 25.0 | 50 | 0.2101 | 0.6240 | 0.6719 | 0.9631 | nan | 0.3544 | 0.9895 | nan | 0.2854 | 0.9625 |
| 0.1159 | 30.0 | 60 | 0.1704 | 0.6262 | 0.6641 | 0.9654 | nan | 0.3353 | 0.9928 | nan | 0.2875 | 0.9650 |
| 0.0869 | 35.0 | 70 | 0.1443 | 0.6380 | 0.6817 | 0.9657 | nan | 0.3720 | 0.9915 | nan | 0.3108 | 0.9652 |
| 0.079 | 40.0 | 80 | 0.1350 | 0.6072 | 0.6360 | 0.9654 | nan | 0.2766 | 0.9953 | nan | 0.2494 | 0.9650 |
| 0.0647 | 45.0 | 90 | 0.1370 | 0.5800 | 0.6031 | 0.9643 | nan | 0.2090 | 0.9971 | nan | 0.1959 | 0.9640 |
| 0.0587 | 50.0 | 100 | 0.1336 | 0.6276 | 0.6796 | 0.9628 | nan | 0.3707 | 0.9885 | nan | 0.2929 | 0.9622 |
| 0.0575 | 55.0 | 110 | 0.1313 | 0.6189 | 0.6531 | 0.9654 | nan | 0.3126 | 0.9937 | nan | 0.2729 | 0.9649 |
| 0.0527 | 60.0 | 120 | 0.1298 | 0.6252 | 0.6655 | 0.9648 | nan | 0.3391 | 0.9920 | nan | 0.2860 | 0.9643 |
| 0.0491 | 65.0 | 130 | 0.1313 | 0.6110 | 0.6492 | 0.9635 | nan | 0.3063 | 0.9920 | nan | 0.2589 | 0.9631 |
| 0.0441 | 70.0 | 140 | 0.1295 | 0.6103 | 0.6429 | 0.9648 | nan | 0.2919 | 0.9939 | nan | 0.2562 | 0.9643 |
| 0.0426 | 75.0 | 150 | 0.1233 | 0.6271 | 0.6633 | 0.9659 | nan | 0.3333 | 0.9933 | nan | 0.2887 | 0.9654 |
| 0.0477 | 80.0 | 160 | 0.1286 | 0.6255 | 0.6629 | 0.9655 | nan | 0.3328 | 0.9929 | nan | 0.2861 | 0.9650 |
| 0.039 | 85.0 | 170 | 0.1265 | 0.6380 | 0.6824 | 0.9656 | nan | 0.3735 | 0.9913 | nan | 0.3109 | 0.9650 |
| 0.0378 | 90.0 | 180 | 0.1309 | 0.6185 | 0.6543 | 0.9650 | nan | 0.3154 | 0.9932 | nan | 0.2725 | 0.9645 |
| 0.0362 | 95.0 | 190 | 0.1266 | 0.6311 | 0.6715 | 0.9655 | nan | 0.3508 | 0.9922 | nan | 0.2973 | 0.9650 |
| 0.0394 | 100.0 | 200 | 0.1307 | 0.6274 | 0.6635 | 0.9659 | nan | 0.3337 | 0.9934 | nan | 0.2894 | 0.9655 |
| 0.0362 | 105.0 | 210 | 0.1271 | 0.6366 | 0.6789 | 0.9658 | nan | 0.3661 | 0.9918 | nan | 0.3080 | 0.9653 |
| 0.0361 | 110.0 | 220 | 0.1274 | 0.6317 | 0.6736 | 0.9653 | nan | 0.3554 | 0.9918 | nan | 0.2987 | 0.9648 |
| 0.0353 | 115.0 | 230 | 0.1290 | 0.6216 | 0.6579 | 0.9652 | nan | 0.3228 | 0.9931 | nan | 0.2784 | 0.9647 |
| 0.0344 | 120.0 | 240 | 0.1296 | 0.6242 | 0.6623 | 0.9652 | nan | 0.3319 | 0.9926 | nan | 0.2838 | 0.9647 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b5_6
|
sam1120
| 2024-02-12T13:58:06Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T13:25:25Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b5_6
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b5_6
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1429
- Mean Iou: 0.6443
- Mean Accuracy: 0.6853
- Overall Accuracy: 0.9669
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.3782
- Accuracy Undropoff: 0.9925
- Iou Unlabeled: nan
- Iou Dropoff: 0.3223
- Iou Undropoff: 0.9664
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.159 | 5.0 | 10 | 1.0040 | 0.2283 | 0.5676 | 0.6267 | nan | 0.5031 | 0.6321 | 0.0 | 0.0644 | 0.6203 |
| 0.8345 | 10.0 | 20 | 0.7480 | 0.3236 | 0.5320 | 0.9158 | nan | 0.1134 | 0.9506 | 0.0 | 0.0555 | 0.9154 |
| 0.5406 | 15.0 | 30 | 0.5477 | 0.3223 | 0.5049 | 0.9513 | nan | 0.0179 | 0.9918 | 0.0 | 0.0157 | 0.9513 |
| 0.3695 | 20.0 | 40 | 0.4590 | 0.3215 | 0.5036 | 0.9519 | nan | 0.0146 | 0.9926 | 0.0 | 0.0125 | 0.9519 |
| 0.3053 | 25.0 | 50 | 0.3790 | 0.3196 | 0.5001 | 0.9565 | nan | 0.0023 | 0.9979 | 0.0 | 0.0022 | 0.9565 |
| 0.2436 | 30.0 | 60 | 0.3303 | 0.4812 | 0.5020 | 0.9568 | nan | 0.0059 | 0.9981 | nan | 0.0056 | 0.9568 |
| 0.2148 | 35.0 | 70 | 0.2739 | 0.4794 | 0.5002 | 0.9580 | nan | 0.0008 | 0.9996 | nan | 0.0008 | 0.9580 |
| 0.1983 | 40.0 | 80 | 0.2348 | 0.5079 | 0.5284 | 0.9595 | nan | 0.0582 | 0.9986 | nan | 0.0564 | 0.9594 |
| 0.1784 | 45.0 | 90 | 0.2178 | 0.6064 | 0.6440 | 0.9631 | nan | 0.2960 | 0.9920 | nan | 0.2501 | 0.9626 |
| 0.1631 | 50.0 | 100 | 0.1943 | 0.6223 | 0.6811 | 0.9607 | nan | 0.3760 | 0.9861 | nan | 0.2846 | 0.9601 |
| 0.1468 | 55.0 | 110 | 0.1759 | 0.6206 | 0.6731 | 0.9617 | nan | 0.3583 | 0.9879 | nan | 0.2801 | 0.9611 |
| 0.1353 | 60.0 | 120 | 0.1657 | 0.6014 | 0.6335 | 0.9639 | nan | 0.2731 | 0.9939 | nan | 0.2393 | 0.9635 |
| 0.1474 | 65.0 | 130 | 0.1590 | 0.5943 | 0.6228 | 0.9641 | nan | 0.2505 | 0.9951 | nan | 0.2249 | 0.9637 |
| 0.1172 | 70.0 | 140 | 0.1562 | 0.6272 | 0.6662 | 0.9653 | nan | 0.3400 | 0.9924 | nan | 0.2896 | 0.9648 |
| 0.1169 | 75.0 | 150 | 0.1538 | 0.6302 | 0.6696 | 0.9656 | nan | 0.3467 | 0.9925 | nan | 0.2954 | 0.9651 |
| 0.1263 | 80.0 | 160 | 0.1540 | 0.6372 | 0.6784 | 0.9661 | nan | 0.3645 | 0.9922 | nan | 0.3089 | 0.9656 |
| 0.1028 | 85.0 | 170 | 0.1512 | 0.6462 | 0.6948 | 0.9659 | nan | 0.3992 | 0.9904 | nan | 0.3271 | 0.9653 |
| 0.1163 | 90.0 | 180 | 0.1493 | 0.6469 | 0.6932 | 0.9663 | nan | 0.3953 | 0.9911 | nan | 0.3280 | 0.9658 |
| 0.0998 | 95.0 | 190 | 0.1481 | 0.6457 | 0.6894 | 0.9666 | nan | 0.3869 | 0.9918 | nan | 0.3253 | 0.9661 |
| 0.0997 | 100.0 | 200 | 0.1465 | 0.6454 | 0.6893 | 0.9665 | nan | 0.3869 | 0.9917 | nan | 0.3247 | 0.9660 |
| 0.0998 | 105.0 | 210 | 0.1473 | 0.6488 | 0.6937 | 0.9668 | nan | 0.3958 | 0.9916 | nan | 0.3313 | 0.9662 |
| 0.1003 | 110.0 | 220 | 0.1437 | 0.6401 | 0.6774 | 0.9671 | nan | 0.3614 | 0.9934 | nan | 0.3136 | 0.9666 |
| 0.0932 | 115.0 | 230 | 0.1434 | 0.6469 | 0.6898 | 0.9669 | nan | 0.3876 | 0.9920 | nan | 0.3275 | 0.9664 |
| 0.0942 | 120.0 | 240 | 0.1429 | 0.6443 | 0.6853 | 0.9669 | nan | 0.3782 | 0.9925 | nan | 0.3223 | 0.9664 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b5_4
|
sam1120
| 2024-02-12T13:56:36Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T13:24:40Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b5_4
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b5_4
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2351
- Mean Iou: 0.4792
- Mean Accuracy: 0.5
- Overall Accuracy: 0.9584
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.0
- Accuracy Undropoff: 1.0
- Iou Unlabeled: nan
- Iou Dropoff: 0.0
- Iou Undropoff: 0.9584
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 7e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0114 | 5.0 | 10 | 1.0037 | 0.2459 | 0.4345 | 0.7074 | nan | 0.1368 | 0.7322 | 0.0 | 0.0286 | 0.7089 |
| 0.9088 | 10.0 | 20 | 0.8245 | 0.3119 | 0.5046 | 0.8887 | nan | 0.0857 | 0.9235 | 0.0 | 0.0460 | 0.8897 |
| 0.8029 | 15.0 | 30 | 0.6620 | 0.3157 | 0.4998 | 0.9214 | nan | 0.0399 | 0.9596 | 0.0 | 0.0253 | 0.9219 |
| 0.6935 | 20.0 | 40 | 0.5662 | 0.3154 | 0.4959 | 0.9309 | nan | 0.0214 | 0.9704 | 0.0 | 0.0151 | 0.9311 |
| 0.635 | 25.0 | 50 | 0.5018 | 0.3175 | 0.4978 | 0.9401 | nan | 0.0153 | 0.9803 | 0.0 | 0.0121 | 0.9404 |
| 0.5579 | 30.0 | 60 | 0.4701 | 0.3178 | 0.4978 | 0.9422 | nan | 0.0131 | 0.9825 | 0.0 | 0.0111 | 0.9423 |
| 0.5086 | 35.0 | 70 | 0.4403 | 0.3181 | 0.4977 | 0.9459 | nan | 0.0088 | 0.9866 | 0.0 | 0.0080 | 0.9461 |
| 0.472 | 40.0 | 80 | 0.4328 | 0.3177 | 0.4971 | 0.9471 | nan | 0.0063 | 0.9879 | 0.0 | 0.0059 | 0.9473 |
| 0.4484 | 45.0 | 90 | 0.4136 | 0.3184 | 0.4981 | 0.9506 | nan | 0.0046 | 0.9916 | 0.0 | 0.0044 | 0.9508 |
| 0.4026 | 50.0 | 100 | 0.4013 | 0.3186 | 0.4985 | 0.9516 | nan | 0.0043 | 0.9926 | 0.0 | 0.0042 | 0.9517 |
| 0.3873 | 55.0 | 110 | 0.3621 | 0.3189 | 0.4991 | 0.9557 | nan | 0.0010 | 0.9971 | 0.0 | 0.0009 | 0.9557 |
| 0.3549 | 60.0 | 120 | 0.3479 | 0.3189 | 0.4992 | 0.9564 | nan | 0.0004 | 0.9979 | 0.0 | 0.0004 | 0.9564 |
| 0.3358 | 65.0 | 130 | 0.3282 | 0.3191 | 0.4994 | 0.9571 | nan | 0.0001 | 0.9986 | 0.0 | 0.0001 | 0.9571 |
| 0.3146 | 70.0 | 140 | 0.3141 | 0.3193 | 0.4996 | 0.9577 | nan | 0.0000 | 0.9993 | 0.0 | 0.0000 | 0.9577 |
| 0.3116 | 75.0 | 150 | 0.2941 | 0.3194 | 0.4999 | 0.9582 | nan | 0.0 | 0.9998 | 0.0 | 0.0 | 0.9582 |
| 0.3151 | 80.0 | 160 | 0.2809 | 0.3195 | 0.5000 | 0.9584 | nan | 0.0 | 0.9999 | 0.0 | 0.0 | 0.9584 |
| 0.2778 | 85.0 | 170 | 0.2750 | 0.3195 | 0.5000 | 0.9584 | nan | 0.0 | 1.0000 | 0.0 | 0.0 | 0.9584 |
| 0.2753 | 90.0 | 180 | 0.2615 | 0.3195 | 0.5000 | 0.9584 | nan | 0.0 | 1.0000 | 0.0 | 0.0 | 0.9584 |
| 0.2809 | 95.0 | 190 | 0.2547 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
| 0.2606 | 100.0 | 200 | 0.2464 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
| 0.2563 | 105.0 | 210 | 0.2459 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
| 0.2454 | 110.0 | 220 | 0.2393 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
| 0.2707 | 115.0 | 230 | 0.2368 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
| 0.2433 | 120.0 | 240 | 0.2351 | 0.4792 | 0.5 | 0.9584 | nan | 0.0 | 1.0 | nan | 0.0 | 0.9584 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b5_1
|
sam1120
| 2024-02-12T13:42:04Z | 147 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T13:23:03Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b5_1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b5_1
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3428
- Mean Iou: 0.4792
- Mean Accuracy: 0.5000
- Overall Accuracy: 0.9583
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.0001
- Accuracy Undropoff: 0.9999
- Iou Unlabeled: nan
- Iou Dropoff: 0.0001
- Iou Undropoff: 0.9583
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 0.8047 | 5.0 | 10 | 0.9867 | 0.2744 | 0.6315 | 0.7475 | nan | 0.5049 | 0.7581 | 0.0 | 0.0812 | 0.7422 |
| 0.7528 | 10.0 | 20 | 0.8526 | 0.3461 | 0.5957 | 0.9213 | nan | 0.2406 | 0.9508 | 0.0 | 0.1178 | 0.9205 |
| 0.7087 | 15.0 | 30 | 0.7023 | 0.3450 | 0.5533 | 0.9467 | nan | 0.1243 | 0.9824 | 0.0 | 0.0887 | 0.9464 |
| 0.6601 | 20.0 | 40 | 0.6251 | 0.3381 | 0.5390 | 0.9462 | nan | 0.0948 | 0.9832 | 0.0 | 0.0684 | 0.9460 |
| 0.6274 | 25.0 | 50 | 0.5828 | 0.3286 | 0.5178 | 0.9486 | nan | 0.0479 | 0.9876 | 0.0 | 0.0374 | 0.9485 |
| 0.5929 | 30.0 | 60 | 0.5478 | 0.3257 | 0.5122 | 0.9488 | nan | 0.0359 | 0.9884 | 0.0 | 0.0284 | 0.9487 |
| 0.5672 | 35.0 | 70 | 0.5237 | 0.3240 | 0.5088 | 0.9494 | nan | 0.0283 | 0.9893 | 0.0 | 0.0227 | 0.9493 |
| 0.5454 | 40.0 | 80 | 0.4966 | 0.4856 | 0.5072 | 0.9529 | nan | 0.0212 | 0.9933 | nan | 0.0183 | 0.9528 |
| 0.5261 | 45.0 | 90 | 0.4700 | 0.3234 | 0.5062 | 0.9553 | nan | 0.0163 | 0.9960 | 0.0 | 0.0149 | 0.9552 |
| 0.5012 | 50.0 | 100 | 0.4576 | 0.4832 | 0.5041 | 0.9563 | nan | 0.0107 | 0.9974 | nan | 0.0101 | 0.9563 |
| 0.4875 | 55.0 | 110 | 0.4430 | 0.4811 | 0.5018 | 0.9566 | nan | 0.0058 | 0.9978 | nan | 0.0056 | 0.9565 |
| 0.4622 | 60.0 | 120 | 0.4328 | 0.4800 | 0.5007 | 0.9570 | nan | 0.0031 | 0.9983 | nan | 0.0030 | 0.9570 |
| 0.4394 | 65.0 | 130 | 0.4179 | 0.4796 | 0.5004 | 0.9572 | nan | 0.0021 | 0.9986 | nan | 0.0021 | 0.9572 |
| 0.4352 | 70.0 | 140 | 0.4048 | 0.4795 | 0.5002 | 0.9573 | nan | 0.0016 | 0.9988 | nan | 0.0016 | 0.9573 |
| 0.426 | 75.0 | 150 | 0.3881 | 0.4796 | 0.5003 | 0.9577 | nan | 0.0015 | 0.9992 | nan | 0.0014 | 0.9577 |
| 0.4175 | 80.0 | 160 | 0.3794 | 0.4797 | 0.5004 | 0.9579 | nan | 0.0014 | 0.9994 | nan | 0.0014 | 0.9579 |
| 0.4087 | 85.0 | 170 | 0.3742 | 0.3196 | 0.5002 | 0.9577 | nan | 0.0012 | 0.9992 | 0.0 | 0.0012 | 0.9577 |
| 0.3887 | 90.0 | 180 | 0.3645 | 0.4792 | 0.4999 | 0.9581 | nan | 0.0003 | 0.9996 | nan | 0.0003 | 0.9581 |
| 0.3799 | 95.0 | 190 | 0.3540 | 0.4791 | 0.4999 | 0.9581 | nan | 0.0001 | 0.9997 | nan | 0.0001 | 0.9581 |
| 0.376 | 100.0 | 200 | 0.3511 | 0.4792 | 0.4999 | 0.9582 | nan | 0.0001 | 0.9998 | nan | 0.0001 | 0.9582 |
| 0.3677 | 105.0 | 210 | 0.3452 | 0.4792 | 0.4999 | 0.9582 | nan | 0.0001 | 0.9998 | nan | 0.0001 | 0.9582 |
| 0.358 | 110.0 | 220 | 0.3437 | 0.4792 | 0.4999 | 0.9582 | nan | 0.0001 | 0.9998 | nan | 0.0001 | 0.9582 |
| 0.3997 | 115.0 | 230 | 0.3434 | 0.4792 | 0.5000 | 0.9583 | nan | 0.0001 | 0.9999 | nan | 0.0001 | 0.9583 |
| 0.3769 | 120.0 | 240 | 0.3428 | 0.4792 | 0.5000 | 0.9583 | nan | 0.0001 | 0.9999 | nan | 0.0001 | 0.9583 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
alitolga/electra-base-generator-rank8
|
alitolga
| 2024-02-12T13:41:55Z | 0 | 0 | null |
[
"safetensors",
"generated_from_trainer",
"base_model:google/electra-base-generator",
"base_model:finetune:google/electra-base-generator",
"license:apache-2.0",
"region:us"
] | null | 2024-02-12T13:41:17Z |
---
license: apache-2.0
base_model: google/electra-base-generator
tags:
- generated_from_trainer
model-index:
- name: electra-base-generator-rank8
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# electra-base-generator-rank8
This model is a fine-tuned version of [google/electra-base-generator](https://huggingface.co/google/electra-base-generator) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2562
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 8.2296 | 1.0 | 179 | 3.8171 |
| 3.6406 | 2.0 | 358 | 3.3218 |
| 3.395 | 3.0 | 537 | 3.2562 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
sam1120/dropoff-utcustom-train-SF-RGBD-b5_3
|
sam1120
| 2024-02-12T13:40:47Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T13:23:45Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b5_3
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b5_3
This model is a fine-tuned version of [nvidia/mit-b5](https://huggingface.co/nvidia/mit-b5) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2768
- Mean Iou: 0.3194
- Mean Accuracy: 0.4999
- Overall Accuracy: 0.9578
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.0006
- Accuracy Undropoff: 0.9993
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.0006
- Iou Undropoff: 0.9578
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 15
- eval_batch_size: 15
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0992 | 5.0 | 10 | 1.0599 | 0.1938 | 0.4241 | 0.5281 | nan | 0.3106 | 0.5376 | 0.0 | 0.0540 | 0.5273 |
| 1.0188 | 10.0 | 20 | 0.9493 | 0.2781 | 0.4808 | 0.7846 | nan | 0.1494 | 0.8122 | 0.0 | 0.0476 | 0.7868 |
| 0.9218 | 15.0 | 30 | 0.8130 | 0.3074 | 0.4913 | 0.8851 | nan | 0.0618 | 0.9209 | 0.0 | 0.0364 | 0.8858 |
| 0.8411 | 20.0 | 40 | 0.7253 | 0.3089 | 0.4866 | 0.9038 | nan | 0.0315 | 0.9416 | 0.0 | 0.0221 | 0.9047 |
| 0.7583 | 25.0 | 50 | 0.6719 | 0.3097 | 0.4890 | 0.9069 | nan | 0.0331 | 0.9448 | 0.0 | 0.0216 | 0.9076 |
| 0.688 | 30.0 | 60 | 0.6303 | 0.3109 | 0.4883 | 0.9170 | nan | 0.0207 | 0.9559 | 0.0 | 0.0149 | 0.9179 |
| 0.6279 | 35.0 | 70 | 0.5919 | 0.3139 | 0.4918 | 0.9276 | nan | 0.0164 | 0.9671 | 0.0 | 0.0133 | 0.9283 |
| 0.5533 | 40.0 | 80 | 0.5375 | 0.3168 | 0.4961 | 0.9377 | nan | 0.0144 | 0.9777 | 0.0 | 0.0125 | 0.9380 |
| 0.5116 | 45.0 | 90 | 0.5111 | 0.3176 | 0.4970 | 0.9442 | nan | 0.0093 | 0.9847 | 0.0 | 0.0083 | 0.9445 |
| 0.4801 | 50.0 | 100 | 0.4696 | 0.3183 | 0.4981 | 0.9492 | nan | 0.0062 | 0.9901 | 0.0 | 0.0057 | 0.9492 |
| 0.4744 | 55.0 | 110 | 0.4317 | 0.3187 | 0.4987 | 0.9543 | nan | 0.0018 | 0.9956 | 0.0 | 0.0017 | 0.9543 |
| 0.4494 | 60.0 | 120 | 0.3991 | 0.3189 | 0.4991 | 0.9555 | nan | 0.0013 | 0.9969 | 0.0 | 0.0012 | 0.9555 |
| 0.386 | 65.0 | 130 | 0.3737 | 0.3189 | 0.4990 | 0.9565 | nan | 0.0000 | 0.9980 | 0.0 | 0.0000 | 0.9565 |
| 0.3674 | 70.0 | 140 | 0.3538 | 0.3191 | 0.4994 | 0.9567 | nan | 0.0007 | 0.9981 | 0.0 | 0.0007 | 0.9567 |
| 0.3601 | 75.0 | 150 | 0.3413 | 0.3192 | 0.4995 | 0.9573 | nan | 0.0002 | 0.9988 | 0.0 | 0.0002 | 0.9573 |
| 0.3626 | 80.0 | 160 | 0.3225 | 0.3193 | 0.4996 | 0.9569 | nan | 0.0009 | 0.9984 | 0.0 | 0.0009 | 0.9569 |
| 0.3331 | 85.0 | 170 | 0.3163 | 0.3195 | 0.5000 | 0.9576 | nan | 0.0009 | 0.9991 | 0.0 | 0.0009 | 0.9576 |
| 0.319 | 90.0 | 180 | 0.3004 | 0.3200 | 0.5008 | 0.9577 | nan | 0.0024 | 0.9991 | 0.0 | 0.0024 | 0.9577 |
| 0.3163 | 95.0 | 190 | 0.2931 | 0.3198 | 0.5004 | 0.9575 | nan | 0.0020 | 0.9989 | 0.0 | 0.0020 | 0.9575 |
| 0.3185 | 100.0 | 200 | 0.2920 | 0.3194 | 0.4999 | 0.9577 | nan | 0.0006 | 0.9992 | 0.0 | 0.0006 | 0.9577 |
| 0.3122 | 105.0 | 210 | 0.2831 | 0.3194 | 0.4999 | 0.9578 | nan | 0.0005 | 0.9994 | 0.0 | 0.0005 | 0.9578 |
| 0.3218 | 110.0 | 220 | 0.2788 | 0.3195 | 0.5000 | 0.9576 | nan | 0.0009 | 0.9991 | 0.0 | 0.0009 | 0.9576 |
| 0.3037 | 115.0 | 230 | 0.2752 | 0.3194 | 0.4999 | 0.9577 | nan | 0.0006 | 0.9992 | 0.0 | 0.0006 | 0.9577 |
| 0.3319 | 120.0 | 240 | 0.2768 | 0.3194 | 0.4999 | 0.9578 | nan | 0.0006 | 0.9993 | 0.0 | 0.0006 | 0.9578 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
yemen2016/MeMo-BERT-WSD_old
|
yemen2016
| 2024-02-12T13:40:04Z | 48 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"bert",
"text-classification",
"generated_from_trainer",
"da",
"base_model:MiMe-MeMo/MeMo-BERT-01",
"base_model:finetune:MiMe-MeMo/MeMo-BERT-01",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-09T12:37:02Z |
---
base_model: MiMe-MeMo/MeMo-BERT-01
tags:
- generated_from_trainer
model-index:
- name: new_memo_model
results: []
language: da # <-- my language
widget:
- text: "Men havde Gud vendt sig fra ham , saa kunde han ogsaa vende sig fra Gud . Havde Gud ingen Øren , saa havde han heller ingen Læber , havde Gud ingen Naade , saa havde han heller ingen Tilbedelse , og han trodsede og viste Gud ud af sit Hjærte ."
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# MeMo Model (Word Sense Disambiguation)
This model is a fine-tuned version of [MiMe-MeMo/MeMo-BERT-01](https://huggingface.co/MiMe-MeMo/MeMo-BERT-01) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7214
- F1-score: 0.6667
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
### Training results
| Training Loss | Epoch | Step | Validation Loss | F1-score |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| No log | 1.0 | 11 | 0.7214 | 0.6667 |
| No log | 2.0 | 22 | 1.2543 | 0.5429 |
| No log | 3.0 | 33 | 1.0829 | 0.6837 |
| No log | 4.0 | 44 | 1.3815 | 0.7552 |
| No log | 5.0 | 55 | 1.4733 | 0.7005 |
| No log | 6.0 | 66 | 2.3876 | 0.5513 |
| No log | 7.0 | 77 | 1.3215 | 0.8004 |
| No log | 8.0 | 88 | 1.4006 | 0.7608 |
| No log | 9.0 | 99 | 1.4862 | 0.7608 |
| No log | 10.0 | 110 | 1.4974 | 0.7608 |
| No log | 11.0 | 121 | 1.4966 | 0.7608 |
| No log | 12.0 | 132 | 1.5040 | 0.7608 |
| No log | 13.0 | 143 | 1.5010 | 0.7608 |
| No log | 14.0 | 154 | 1.4741 | 0.7608 |
| No log | 15.0 | 165 | 1.4507 | 0.7608 |
| No log | 16.0 | 176 | 1.4420 | 0.7608 |
| No log | 17.0 | 187 | 1.4398 | 0.7608 |
| No log | 18.0 | 198 | 1.4426 | 0.7608 |
| No log | 19.0 | 209 | 1.4438 | 0.7608 |
| No log | 20.0 | 220 | 1.4439 | 0.7608 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
Deadwalker0/phitune
|
Deadwalker0
| 2024-02-12T13:38:06Z | 0 | 0 |
peft
|
[
"peft",
"tensorboard",
"safetensors",
"phi",
"generated_from_trainer",
"custom_code",
"base_model:microsoft/phi-1_5",
"base_model:adapter:microsoft/phi-1_5",
"license:mit",
"4-bit",
"bitsandbytes",
"region:us"
] | null | 2024-02-12T13:30:14Z |
---
license: mit
library_name: peft
tags:
- generated_from_trainer
base_model: microsoft/phi-1_5
model-index:
- name: phi-sft-out
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
<details><summary>See axolotl config</summary>
axolotl version: `0.4.0`
```yaml
base_model: microsoft/phi-1_5
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
load_in_8bit: false
load_in_4bit: true
strict: false
datasets:
- path: garage-bAInd/Open-Platypus
type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./phi-sft-out
sequence_len: 2048
sample_packing: true
pad_to_sequence_len: true
adapter: qlora
lora_model_dir:
lora_r: 64
lora_alpha: 32
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 1
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_torch
adam_beta2: 0.95
adam_epsilon: 0.00001
max_grad_norm: 1.0
lr_scheduler: cosine
learning_rate: 0.000003
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: true
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: True
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 100
evals_per_epoch: 4
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
resize_token_embeddings_to_32x: true
special_tokens:
pad_token: "<|endoftext|>"
```
</details><br>
# phi-sft-out
This model is a fine-tuned version of [microsoft/phi-1_5](https://huggingface.co/microsoft/phi-1_5) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2548
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-06
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 4
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.0668 | 0.0 | 1 | 1.2826 |
| 0.9408 | 0.25 | 580 | 1.2613 |
| 1.2121 | 0.5 | 1160 | 1.2559 |
| 0.9644 | 0.75 | 1740 | 1.2562 |
| 0.9582 | 1.0 | 2320 | 1.2556 |
| 1.0009 | 1.23 | 2900 | 1.2559 |
| 0.7816 | 1.48 | 3480 | 1.2556 |
| 0.9843 | 1.73 | 4060 | 1.2552 |
| 0.8877 | 1.98 | 4640 | 1.2559 |
| 0.8498 | 2.21 | 5220 | 1.2554 |
| 0.9163 | 2.46 | 5800 | 1.2550 |
| 1.0539 | 2.71 | 6380 | 1.2545 |
| 0.9533 | 2.96 | 6960 | 1.2547 |
| 0.6969 | 3.19 | 7540 | 1.2547 |
| 0.6204 | 3.44 | 8120 | 1.2547 |
| 0.891 | 3.69 | 8700 | 1.2548 |
### Framework versions
- PEFT 0.8.2
- Transformers 4.38.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.17.0
- Tokenizers 0.15.0
|
ramsi-k/Reinforce-PixelCopter_64
|
ramsi-k
| 2024-02-12T13:36:38Z | 0 | 0 | null |
[
"Pixelcopter-PLE-v0",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T13:36:09Z |
---
tags:
- Pixelcopter-PLE-v0
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: Reinforce-PixelCopter_64
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Pixelcopter-PLE-v0
type: Pixelcopter-PLE-v0
metrics:
- type: mean_reward
value: 33.00 +/- 41.93
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **Pixelcopter-PLE-v0**
This is a trained model of a **Reinforce** agent playing **Pixelcopter-PLE-v0** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
hugo-massonnat/ppo-Huggy
|
hugo-massonnat
| 2024-02-12T13:35:12Z | 0 | 0 |
ml-agents
|
[
"ml-agents",
"tensorboard",
"onnx",
"Huggy",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-Huggy",
"region:us"
] |
reinforcement-learning
| 2024-02-12T13:35:03Z |
---
library_name: ml-agents
tags:
- Huggy
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-Huggy
---
# **ppo** Agent playing **Huggy**
This is a trained model of a **ppo** agent playing **Huggy**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: hugo-massonnat/ppo-Huggy
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
alitolga/electra-base-generator-rank2
|
alitolga
| 2024-02-12T13:31:54Z | 0 | 0 | null |
[
"safetensors",
"generated_from_trainer",
"base_model:google/electra-base-generator",
"base_model:finetune:google/electra-base-generator",
"license:apache-2.0",
"region:us"
] | null | 2024-02-12T13:25:46Z |
---
license: apache-2.0
base_model: google/electra-base-generator
tags:
- generated_from_trainer
model-index:
- name: electra-base-generator-rank2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# electra-base-generator-rank2
This model is a fine-tuned version of [google/electra-base-generator](https://huggingface.co/google/electra-base-generator) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2155
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 8.206 | 1.0 | 179 | 3.8146 |
| 3.5779 | 2.0 | 358 | 3.2736 |
| 3.3568 | 3.0 | 537 | 3.2155 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
eren23/sd15-FantasyMix-filmGrain-segmoe
|
eren23
| 2024-02-12T13:31:44Z | 30 | 0 |
diffusers
|
[
"diffusers",
"safetensors",
"text-to-image",
"stable-diffusion",
"segmoe",
"merge",
"moe",
"sd1.5",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2024-02-12T13:17:43Z |
---
library_name: diffusers
pipeline_tag: text-to-image
tags:
- text-to-image
- stable-diffusion
- segmoe
- merge
- moe
- sd1.5
---
This model is a segmoe merge of 2 models from civitAI:
https://civitai.com/models/234898/vixons-fantasy-mix
https://civitai.com/models/43977?modelVersionId=113623
Merged using the great project at: https://github.com/segmind/segmoe
To do something similar you can either follow the guide in readme or you can follow this blogpost: https://huggingface.co/blog/segmoe
The setting I used:
base_model: https://civitai.com/api/download/models/306781
num_experts: 4
moe_layers: all
num_experts_per_tok: 2
type: sd
experts:
- source_model: https://civitai.com/api/download/models/306781
positive_prompt: "cinematic, portrait, photograph, instagram, fashion, movie, macro shot, 8K, RAW, fantastic, ultra high quality"
negative_prompt: " (deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime), text, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck"
- source_model: https://civitai.com/api/download/models/113623
positive_prompt: "photo realistic scenes, fantastic view, impressive view, movie scene, 8K, RAW, hyperrealistic, ultra realistic"
negative_prompt: "simple background, duplicate, retro style, low quality, lowest quality, 1980s, 1990s, 2000s, 2005 2006 2007 2008 2009 2010 2011 2012 2013, bad anatomy, bad proportions, extra digits, lowres, username, artist name, error, duplicate, watermark, signature, text, extra digit, fewer digits, worst quality, jpeg artifacts, blurry"
# Useage
!pip install -U segmoe diffusers transformers
from segmoe import SegMoEPipeline
pipeline = SegMoEPipeline("eren23/sd15-FantasyMix-filmGrain-segmoe", device="cuda")
prompt = "fantastic land canvas, knight cat standing next to a purple medieval village wall"
negative_prompt = "nsfw, bad quality, worse quality"
img = pipeline(
prompt=prompt,
negative_prompt=negative_prompt,
height=512,
width=512,
num_inference_steps=30,
guidance_scale=7.5,
).images[0]
img.save("image.png")
|
Annikaijak/bert_classification
|
Annikaijak
| 2024-02-12T13:31:36Z | 93 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"text-classification",
"arxiv:1910.09700",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T13:31:04Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
hweemiin/ppo-LunarLander-v2
|
hweemiin
| 2024-02-12T13:31:35Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T13:31:13Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 214.81 +/- 68.51
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
davmel/ka_word2vec
|
davmel
| 2024-02-12T13:31:29Z | 0 | 0 | null |
[
"ka",
"license:mit",
"region:us"
] | null | 2024-02-12T13:23:58Z |
---
license: mit
language:
- ka
---
This is a word2vec model trained on CC100 Georgian dataset.
|
Camillahannesbo/Camillas_bert_model
|
Camillahannesbo
| 2024-02-12T13:27:32Z | 94 | 0 |
transformers
|
[
"transformers",
"safetensors",
"bert",
"text-classification",
"arxiv:1910.09700",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2024-02-12T13:26:44Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
Shruthi-S/bertbase-ten
|
Shruthi-S
| 2024-02-12T13:19:04Z | 45 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"pretraining",
"generated_from_keras_callback",
"endpoints_compatible",
"region:us"
] | null | 2024-02-12T13:18:41Z |
---
tags:
- generated_from_keras_callback
model-index:
- name: bertbase-ten
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# bertbase-ten
This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 8.1691
- Validation Loss: 8.0864
- Epoch: 0
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': 0.001, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Validation Loss | Epoch |
|:----------:|:---------------:|:-----:|
| 8.1691 | 8.0864 | 0 |
### Framework versions
- Transformers 4.38.0.dev0
- TensorFlow 2.15.0
- Datasets 2.17.0
- Tokenizers 0.15.1
|
Ostfriese/q-taxi
|
Ostfriese
| 2024-02-12T13:08:23Z | 0 | 0 | null |
[
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T12:54:02Z |
---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-taxi
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.56 +/- 2.71
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="Ostfriese/q-taxi", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
y-oguchi/codeparrot-ds
|
y-oguchi
| 2024-02-12T13:05:37Z | 93 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"gpt2",
"text-generation",
"generated_from_trainer",
"base_model:openai-community/gpt2",
"base_model:finetune:openai-community/gpt2",
"license:mit",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-12T10:39:30Z |
---
license: mit
base_model: gpt2
tags:
- generated_from_trainer
model-index:
- name: codeparrot-ds
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# codeparrot-ds
This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on an unknown dataset.
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 96
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 768
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 1
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
|
alitolga/deberta-v3-base-rank8
|
alitolga
| 2024-02-12T13:04:41Z | 0 | 0 | null |
[
"safetensors",
"generated_from_trainer",
"base_model:microsoft/deberta-v3-base",
"base_model:finetune:microsoft/deberta-v3-base",
"license:mit",
"region:us"
] | null | 2024-02-12T13:01:18Z |
---
license: mit
base_model: microsoft/deberta-v3-base
tags:
- generated_from_trainer
model-index:
- name: deberta-v3-base-rank8
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# deberta-v3-base-rank8
This model is a fine-tuned version of [microsoft/deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 4.8366
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 13.9313 | 1.0 | 179 | 8.8102 |
| 7.5134 | 2.0 | 358 | 5.4105 |
| 5.566 | 3.0 | 537 | 4.8366 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
|
sam1120/dropoff-utcustom-train-SF-RGBD-b0_7
|
sam1120
| 2024-02-12T13:01:30Z | 147 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T12:53:02Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b0_7
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b0_7
This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2075
- Mean Iou: 0.6372
- Mean Accuracy: 0.6861
- Overall Accuracy: 0.9647
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.3822
- Accuracy Undropoff: 0.9900
- Iou Unlabeled: nan
- Iou Dropoff: 0.3104
- Iou Undropoff: 0.9641
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 8e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 0.9508 | 5.0 | 10 | 1.0263 | 0.3104 | 0.5474 | 0.8717 | nan | 0.1937 | 0.9011 | 0.0 | 0.0605 | 0.8706 |
| 0.7814 | 10.0 | 20 | 0.7568 | 0.4971 | 0.5339 | 0.9361 | nan | 0.0952 | 0.9726 | nan | 0.0584 | 0.9359 |
| 0.642 | 15.0 | 30 | 0.5907 | 0.5134 | 0.5443 | 0.9494 | nan | 0.1026 | 0.9861 | nan | 0.0777 | 0.9492 |
| 0.5118 | 20.0 | 40 | 0.4804 | 0.3658 | 0.5923 | 0.9513 | nan | 0.2006 | 0.9839 | 0.0 | 0.1464 | 0.9509 |
| 0.4581 | 25.0 | 50 | 0.4405 | 0.3715 | 0.5915 | 0.9569 | nan | 0.1930 | 0.9900 | 0.0 | 0.1578 | 0.9565 |
| 0.4213 | 30.0 | 60 | 0.4146 | 0.3828 | 0.6136 | 0.9580 | nan | 0.2379 | 0.9892 | 0.0 | 0.1910 | 0.9575 |
| 0.3571 | 35.0 | 70 | 0.3750 | 0.3846 | 0.6180 | 0.9578 | nan | 0.2474 | 0.9887 | 0.0 | 0.1963 | 0.9574 |
| 0.3205 | 40.0 | 80 | 0.3478 | 0.5777 | 0.6202 | 0.9576 | nan | 0.2522 | 0.9882 | nan | 0.1982 | 0.9571 |
| 0.3114 | 45.0 | 90 | 0.3461 | 0.3895 | 0.6423 | 0.9541 | nan | 0.3022 | 0.9824 | 0.0 | 0.2150 | 0.9535 |
| 0.2747 | 50.0 | 100 | 0.3253 | 0.5875 | 0.6357 | 0.9575 | nan | 0.2847 | 0.9867 | nan | 0.2180 | 0.9570 |
| 0.2593 | 55.0 | 110 | 0.3083 | 0.5967 | 0.6599 | 0.9552 | nan | 0.3377 | 0.9820 | nan | 0.2387 | 0.9546 |
| 0.2293 | 60.0 | 120 | 0.2762 | 0.5966 | 0.6389 | 0.9606 | nan | 0.2880 | 0.9898 | nan | 0.2331 | 0.9601 |
| 0.2306 | 65.0 | 130 | 0.2655 | 0.6016 | 0.6587 | 0.9577 | nan | 0.3326 | 0.9848 | nan | 0.2462 | 0.9571 |
| 0.2118 | 70.0 | 140 | 0.2446 | 0.6039 | 0.6509 | 0.9605 | nan | 0.3133 | 0.9886 | nan | 0.2479 | 0.9600 |
| 0.2038 | 75.0 | 150 | 0.2395 | 0.6164 | 0.6708 | 0.9607 | nan | 0.3547 | 0.9870 | nan | 0.2727 | 0.9601 |
| 0.1895 | 80.0 | 160 | 0.2196 | 0.6254 | 0.6721 | 0.9636 | nan | 0.3542 | 0.9900 | nan | 0.2878 | 0.9630 |
| 0.1681 | 85.0 | 170 | 0.2176 | 0.6302 | 0.6829 | 0.9630 | nan | 0.3773 | 0.9884 | nan | 0.2979 | 0.9624 |
| 0.1612 | 90.0 | 180 | 0.2175 | 0.6334 | 0.6870 | 0.9633 | nan | 0.3857 | 0.9884 | nan | 0.3042 | 0.9627 |
| 0.1545 | 95.0 | 190 | 0.2140 | 0.6337 | 0.6816 | 0.9644 | nan | 0.3732 | 0.9900 | nan | 0.3035 | 0.9638 |
| 0.1551 | 100.0 | 200 | 0.2134 | 0.6357 | 0.6891 | 0.9637 | nan | 0.3896 | 0.9886 | nan | 0.3083 | 0.9631 |
| 0.1508 | 105.0 | 210 | 0.2090 | 0.6359 | 0.6865 | 0.9642 | nan | 0.3837 | 0.9894 | nan | 0.3083 | 0.9636 |
| 0.1536 | 110.0 | 220 | 0.2057 | 0.6346 | 0.6801 | 0.9650 | nan | 0.3694 | 0.9908 | nan | 0.3048 | 0.9644 |
| 0.1392 | 115.0 | 230 | 0.2083 | 0.6387 | 0.6890 | 0.9646 | nan | 0.3883 | 0.9896 | nan | 0.3133 | 0.9640 |
| 0.1446 | 120.0 | 240 | 0.2075 | 0.6372 | 0.6861 | 0.9647 | nan | 0.3822 | 0.9900 | nan | 0.3104 | 0.9641 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b0_3
|
sam1120
| 2024-02-12T13:01:18Z | 146 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T12:52:47Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b0_3
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b0_3
This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3666
- Mean Iou: 0.6400
- Mean Accuracy: 0.7120
- Overall Accuracy: 0.9610
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.4404
- Accuracy Undropoff: 0.9836
- Iou Unlabeled: nan
- Iou Dropoff: 0.3196
- Iou Undropoff: 0.9603
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 4e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0352 | 5.0 | 10 | 1.0676 | 0.2560 | 0.5776 | 0.7142 | nan | 0.4286 | 0.7266 | 0.0 | 0.0589 | 0.7090 |
| 0.9564 | 10.0 | 20 | 0.9743 | 0.3355 | 0.5576 | 0.9248 | nan | 0.1571 | 0.9581 | 0.0 | 0.0822 | 0.9243 |
| 0.8577 | 15.0 | 30 | 0.8504 | 0.3318 | 0.5283 | 0.9409 | nan | 0.0782 | 0.9784 | 0.0 | 0.0545 | 0.9407 |
| 0.7512 | 20.0 | 40 | 0.6972 | 0.3270 | 0.5122 | 0.9527 | nan | 0.0318 | 0.9926 | 0.0 | 0.0283 | 0.9526 |
| 0.6955 | 25.0 | 50 | 0.5761 | 0.3259 | 0.5099 | 0.9545 | nan | 0.0250 | 0.9948 | 0.0 | 0.0234 | 0.9544 |
| 0.6691 | 30.0 | 60 | 0.5209 | 0.3360 | 0.5271 | 0.9525 | nan | 0.0632 | 0.9911 | 0.0 | 0.0557 | 0.9524 |
| 0.626 | 35.0 | 70 | 0.5297 | 0.3408 | 0.5362 | 0.9505 | nan | 0.0844 | 0.9881 | 0.0 | 0.0719 | 0.9503 |
| 0.5544 | 40.0 | 80 | 0.5263 | 0.3616 | 0.5757 | 0.9521 | nan | 0.1652 | 0.9862 | 0.0 | 0.1330 | 0.9518 |
| 0.5316 | 45.0 | 90 | 0.4825 | 0.3836 | 0.6353 | 0.9506 | nan | 0.2915 | 0.9792 | 0.0 | 0.2009 | 0.9500 |
| 0.4929 | 50.0 | 100 | 0.4763 | 0.3958 | 0.6588 | 0.9530 | nan | 0.3378 | 0.9797 | 0.0 | 0.2352 | 0.9524 |
| 0.468 | 55.0 | 110 | 0.4583 | 0.4077 | 0.6974 | 0.9528 | nan | 0.4188 | 0.9759 | 0.0 | 0.2713 | 0.9519 |
| 0.429 | 60.0 | 120 | 0.4268 | 0.3985 | 0.6526 | 0.9575 | nan | 0.3199 | 0.9852 | 0.0 | 0.2386 | 0.9569 |
| 0.4211 | 65.0 | 130 | 0.3988 | 0.3951 | 0.6406 | 0.9584 | nan | 0.2939 | 0.9872 | 0.0 | 0.2275 | 0.9578 |
| 0.3926 | 70.0 | 140 | 0.4085 | 0.4102 | 0.6780 | 0.9587 | nan | 0.3718 | 0.9842 | 0.0 | 0.2726 | 0.9581 |
| 0.4006 | 75.0 | 150 | 0.3944 | 0.6077 | 0.6574 | 0.9604 | nan | 0.3269 | 0.9879 | nan | 0.2555 | 0.9599 |
| 0.3978 | 80.0 | 160 | 0.3881 | 0.6216 | 0.6875 | 0.9591 | nan | 0.3912 | 0.9838 | nan | 0.2848 | 0.9585 |
| 0.3553 | 85.0 | 170 | 0.3877 | 0.6333 | 0.7077 | 0.9595 | nan | 0.4329 | 0.9824 | nan | 0.3079 | 0.9588 |
| 0.3637 | 90.0 | 180 | 0.4004 | 0.6428 | 0.7273 | 0.9594 | nan | 0.4741 | 0.9805 | nan | 0.3270 | 0.9586 |
| 0.3416 | 95.0 | 190 | 0.3835 | 0.6403 | 0.7166 | 0.9604 | nan | 0.4507 | 0.9825 | nan | 0.3210 | 0.9596 |
| 0.342 | 100.0 | 200 | 0.3634 | 0.6371 | 0.7061 | 0.9611 | nan | 0.4279 | 0.9842 | nan | 0.3137 | 0.9604 |
| 0.3393 | 105.0 | 210 | 0.3740 | 0.6429 | 0.7217 | 0.9604 | nan | 0.4614 | 0.9820 | nan | 0.3262 | 0.9596 |
| 0.3535 | 110.0 | 220 | 0.3771 | 0.6423 | 0.7199 | 0.9605 | nan | 0.4575 | 0.9823 | nan | 0.3249 | 0.9597 |
| 0.3159 | 115.0 | 230 | 0.3710 | 0.6423 | 0.7167 | 0.9610 | nan | 0.4502 | 0.9832 | nan | 0.3243 | 0.9603 |
| 0.3278 | 120.0 | 240 | 0.3666 | 0.6400 | 0.7120 | 0.9610 | nan | 0.4404 | 0.9836 | nan | 0.3196 | 0.9603 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b0_1
|
sam1120
| 2024-02-12T13:01:12Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T12:52:21Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b0_1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b0_1
This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4979
- Mean Iou: 0.4170
- Mean Accuracy: 0.6846
- Overall Accuracy: 0.9603
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.3839
- Accuracy Undropoff: 0.9853
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.2914
- Iou Undropoff: 0.9597
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0495 | 5.0 | 10 | 1.0890 | 0.1852 | 0.3572 | 0.4990 | nan | 0.2026 | 0.5119 | 0.0 | 0.0474 | 0.5081 |
| 0.9941 | 10.0 | 20 | 1.0479 | 0.3452 | 0.8357 | 0.8479 | nan | 0.8225 | 0.8490 | 0.0 | 0.1931 | 0.8425 |
| 0.9448 | 15.0 | 30 | 0.9839 | 0.3790 | 0.8217 | 0.9010 | nan | 0.7351 | 0.9082 | 0.0 | 0.2390 | 0.8980 |
| 0.8912 | 20.0 | 40 | 0.9041 | 0.3845 | 0.7150 | 0.9247 | nan | 0.4863 | 0.9437 | 0.0 | 0.2303 | 0.9233 |
| 0.8458 | 25.0 | 50 | 0.7997 | 0.3835 | 0.6687 | 0.9326 | nan | 0.3808 | 0.9565 | 0.0 | 0.2188 | 0.9316 |
| 0.8299 | 30.0 | 60 | 0.7387 | 0.3751 | 0.6333 | 0.9326 | nan | 0.3068 | 0.9597 | 0.0 | 0.1934 | 0.9318 |
| 0.7518 | 35.0 | 70 | 0.6810 | 0.3791 | 0.6322 | 0.9404 | nan | 0.2961 | 0.9683 | 0.0 | 0.1975 | 0.9397 |
| 0.6943 | 40.0 | 80 | 0.6322 | 0.3703 | 0.6069 | 0.9422 | nan | 0.2411 | 0.9726 | 0.0 | 0.1691 | 0.9417 |
| 0.6617 | 45.0 | 90 | 0.6071 | 0.3780 | 0.6240 | 0.9454 | nan | 0.2734 | 0.9746 | 0.0 | 0.1892 | 0.9449 |
| 0.634 | 50.0 | 100 | 0.5932 | 0.3765 | 0.6106 | 0.9497 | nan | 0.2407 | 0.9805 | 0.0 | 0.1802 | 0.9494 |
| 0.6157 | 55.0 | 110 | 0.5829 | 0.3982 | 0.6538 | 0.9524 | nan | 0.3281 | 0.9795 | 0.0 | 0.2425 | 0.9520 |
| 0.5814 | 60.0 | 120 | 0.5708 | 0.4038 | 0.6699 | 0.9533 | nan | 0.3608 | 0.9790 | 0.0 | 0.2586 | 0.9528 |
| 0.5988 | 65.0 | 130 | 0.5575 | 0.3974 | 0.6456 | 0.9569 | nan | 0.3061 | 0.9851 | 0.0 | 0.2357 | 0.9564 |
| 0.5583 | 70.0 | 140 | 0.5530 | 0.4224 | 0.7075 | 0.9576 | nan | 0.4346 | 0.9803 | 0.0 | 0.3103 | 0.9570 |
| 0.5596 | 75.0 | 150 | 0.5264 | 0.4034 | 0.6522 | 0.9598 | nan | 0.3167 | 0.9877 | 0.0 | 0.2510 | 0.9593 |
| 0.5524 | 80.0 | 160 | 0.5392 | 0.4208 | 0.7109 | 0.9567 | nan | 0.4429 | 0.9790 | 0.0 | 0.3065 | 0.9560 |
| 0.5294 | 85.0 | 170 | 0.5257 | 0.4161 | 0.6913 | 0.9582 | nan | 0.4002 | 0.9824 | 0.0 | 0.2909 | 0.9576 |
| 0.5477 | 90.0 | 180 | 0.5178 | 0.4207 | 0.6962 | 0.9591 | nan | 0.4095 | 0.9829 | 0.0 | 0.3035 | 0.9584 |
| 0.528 | 95.0 | 190 | 0.5185 | 0.4183 | 0.6939 | 0.9590 | nan | 0.4047 | 0.9831 | 0.0 | 0.2965 | 0.9584 |
| 0.5144 | 100.0 | 200 | 0.5004 | 0.4153 | 0.6788 | 0.9604 | nan | 0.3716 | 0.9860 | 0.0 | 0.2859 | 0.9599 |
| 0.5313 | 105.0 | 210 | 0.5032 | 0.4199 | 0.7005 | 0.9585 | nan | 0.4191 | 0.9819 | 0.0 | 0.3020 | 0.9578 |
| 0.5172 | 110.0 | 220 | 0.4993 | 0.4188 | 0.6931 | 0.9591 | nan | 0.4030 | 0.9832 | 0.0 | 0.2978 | 0.9585 |
| 0.5124 | 115.0 | 230 | 0.4999 | 0.4167 | 0.6828 | 0.9606 | nan | 0.3799 | 0.9858 | 0.0 | 0.2901 | 0.9600 |
| 0.5025 | 120.0 | 240 | 0.4979 | 0.4170 | 0.6846 | 0.9603 | nan | 0.3839 | 0.9853 | 0.0 | 0.2914 | 0.9597 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b0_2
|
sam1120
| 2024-02-12T13:01:09Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T12:52:41Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b0_2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b0_2
This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4274
- Mean Iou: 0.6102
- Mean Accuracy: 0.6603
- Overall Accuracy: 0.9607
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.3326
- Accuracy Undropoff: 0.9879
- Iou Unlabeled: nan
- Iou Dropoff: 0.2602
- Iou Undropoff: 0.9601
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.0555 | 5.0 | 10 | 1.0734 | 0.2254 | 0.4211 | 0.6018 | nan | 0.2240 | 0.6182 | 0.0 | 0.0622 | 0.6140 |
| 0.9825 | 10.0 | 20 | 1.0261 | 0.2992 | 0.6380 | 0.7780 | nan | 0.4852 | 0.7907 | 0.0 | 0.1170 | 0.7807 |
| 0.8991 | 15.0 | 30 | 0.8985 | 0.3231 | 0.5517 | 0.8892 | nan | 0.1836 | 0.9198 | 0.0 | 0.0776 | 0.8917 |
| 0.8191 | 20.0 | 40 | 0.7413 | 0.3270 | 0.5262 | 0.9299 | nan | 0.0858 | 0.9665 | 0.0 | 0.0513 | 0.9296 |
| 0.7562 | 25.0 | 50 | 0.6268 | 0.3259 | 0.5130 | 0.9436 | nan | 0.0433 | 0.9826 | 0.0 | 0.0343 | 0.9435 |
| 0.7395 | 30.0 | 60 | 0.5872 | 0.3235 | 0.5073 | 0.9498 | nan | 0.0246 | 0.9900 | 0.0 | 0.0206 | 0.9498 |
| 0.7272 | 35.0 | 70 | 0.5820 | 0.3379 | 0.5415 | 0.9411 | nan | 0.1055 | 0.9774 | 0.0 | 0.0729 | 0.9409 |
| 0.6525 | 40.0 | 80 | 0.5571 | 0.3445 | 0.5451 | 0.9498 | nan | 0.1036 | 0.9865 | 0.0 | 0.0839 | 0.9496 |
| 0.6161 | 45.0 | 90 | 0.5465 | 0.3480 | 0.5480 | 0.9528 | nan | 0.1064 | 0.9895 | 0.0 | 0.0914 | 0.9526 |
| 0.6131 | 50.0 | 100 | 0.5379 | 0.3712 | 0.5917 | 0.9555 | nan | 0.1949 | 0.9885 | 0.0 | 0.1584 | 0.9551 |
| 0.579 | 55.0 | 110 | 0.5229 | 0.3892 | 0.6411 | 0.9536 | nan | 0.3002 | 0.9819 | 0.0 | 0.2146 | 0.9530 |
| 0.5133 | 60.0 | 120 | 0.5113 | 0.3962 | 0.6596 | 0.9541 | nan | 0.3384 | 0.9808 | 0.0 | 0.2352 | 0.9535 |
| 0.535 | 65.0 | 130 | 0.4925 | 0.3981 | 0.6566 | 0.9561 | nan | 0.3299 | 0.9833 | 0.0 | 0.2386 | 0.9555 |
| 0.4866 | 70.0 | 140 | 0.4717 | 0.5993 | 0.6516 | 0.9584 | nan | 0.3169 | 0.9863 | nan | 0.2407 | 0.9579 |
| 0.5119 | 75.0 | 150 | 0.4712 | 0.5976 | 0.6513 | 0.9578 | nan | 0.3171 | 0.9856 | nan | 0.2380 | 0.9572 |
| 0.5034 | 80.0 | 160 | 0.4737 | 0.6120 | 0.6840 | 0.9562 | nan | 0.3872 | 0.9808 | nan | 0.2686 | 0.9554 |
| 0.4503 | 85.0 | 170 | 0.4496 | 0.6103 | 0.6618 | 0.9604 | nan | 0.3361 | 0.9875 | nan | 0.2607 | 0.9598 |
| 0.4653 | 90.0 | 180 | 0.4617 | 0.6201 | 0.6907 | 0.9580 | nan | 0.3992 | 0.9822 | nan | 0.2830 | 0.9572 |
| 0.4375 | 95.0 | 190 | 0.4412 | 0.6090 | 0.6592 | 0.9605 | nan | 0.3305 | 0.9878 | nan | 0.2580 | 0.9599 |
| 0.4306 | 100.0 | 200 | 0.4355 | 0.6120 | 0.6653 | 0.9602 | nan | 0.3436 | 0.9870 | nan | 0.2643 | 0.9597 |
| 0.4456 | 105.0 | 210 | 0.4414 | 0.6178 | 0.6756 | 0.9601 | nan | 0.3653 | 0.9860 | nan | 0.2760 | 0.9595 |
| 0.4435 | 110.0 | 220 | 0.4387 | 0.6150 | 0.6681 | 0.9608 | nan | 0.3489 | 0.9873 | nan | 0.2699 | 0.9602 |
| 0.4263 | 115.0 | 230 | 0.4348 | 0.6156 | 0.6692 | 0.9607 | nan | 0.3512 | 0.9872 | nan | 0.2711 | 0.9602 |
| 0.4123 | 120.0 | 240 | 0.4274 | 0.6102 | 0.6603 | 0.9607 | nan | 0.3326 | 0.9879 | nan | 0.2602 | 0.9601 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
sam1120/dropoff-utcustom-train-SF-RGBD-b0_4
|
sam1120
| 2024-02-12T13:01:07Z | 145 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"segformer",
"vision",
"image-segmentation",
"generated_from_trainer",
"license:other",
"endpoints_compatible",
"region:us"
] |
image-segmentation
| 2024-02-12T12:52:49Z |
---
license: other
tags:
- vision
- image-segmentation
- generated_from_trainer
model-index:
- name: dropoff-utcustom-train-SF-RGBD-b0_4
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# dropoff-utcustom-train-SF-RGBD-b0_4
This model is a fine-tuned version of [nvidia/mit-b0](https://huggingface.co/nvidia/mit-b0) on the sam1120/dropoff-utcustom-TRAIN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3688
- Mean Iou: 0.3485
- Mean Accuracy: 0.5433
- Overall Accuracy: 0.9606
- Accuracy Unlabeled: nan
- Accuracy Dropoff: 0.0881
- Accuracy Undropoff: 0.9984
- Iou Unlabeled: 0.0
- Iou Dropoff: 0.0851
- Iou Undropoff: 0.9604
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 120
### Training results
| Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Unlabeled | Accuracy Dropoff | Accuracy Undropoff | Iou Unlabeled | Iou Dropoff | Iou Undropoff |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:-------------:|:----------------:|:------------------:|:----------------:|:------------------:|:-------------:|:-----------:|:-------------:|
| 1.2008 | 5.0 | 10 | 1.0960 | 0.1205 | 0.4461 | 0.2825 | nan | 0.6246 | 0.2677 | 0.0 | 0.0943 | 0.2671 |
| 1.0485 | 10.0 | 20 | 1.0952 | 0.1603 | 0.6272 | 0.4049 | nan | 0.8696 | 0.3848 | 0.0 | 0.0965 | 0.3843 |
| 0.9156 | 15.0 | 30 | 1.0312 | 0.3080 | 0.5963 | 0.8333 | nan | 0.3377 | 0.8548 | 0.0 | 0.0924 | 0.8317 |
| 0.7435 | 20.0 | 40 | 0.9448 | 0.3221 | 0.5508 | 0.8937 | nan | 0.1769 | 0.9248 | 0.0 | 0.0733 | 0.8930 |
| 0.7336 | 25.0 | 50 | 0.7446 | 0.3191 | 0.4998 | 0.9461 | nan | 0.0129 | 0.9866 | 0.0 | 0.0113 | 0.9461 |
| 0.6585 | 30.0 | 60 | 0.6397 | 0.3183 | 0.4981 | 0.9534 | nan | 0.0014 | 0.9948 | 0.0 | 0.0013 | 0.9534 |
| 0.583 | 35.0 | 70 | 0.5785 | 0.3181 | 0.4978 | 0.9537 | nan | 0.0006 | 0.9951 | 0.0 | 0.0005 | 0.9537 |
| 0.5324 | 40.0 | 80 | 0.5458 | 0.3182 | 0.4980 | 0.9545 | nan | 0.0002 | 0.9958 | 0.0 | 0.0002 | 0.9545 |
| 0.5155 | 45.0 | 90 | 0.5347 | 0.3186 | 0.4987 | 0.9558 | nan | 0.0001 | 0.9973 | 0.0 | 0.0001 | 0.9558 |
| 0.4874 | 50.0 | 100 | 0.4954 | 0.3179 | 0.4976 | 0.9537 | nan | 0.0 | 0.9951 | 0.0 | 0.0 | 0.9537 |
| 0.4716 | 55.0 | 110 | 0.4646 | 0.3185 | 0.4985 | 0.9555 | nan | 0.0 | 0.9969 | 0.0 | 0.0 | 0.9555 |
| 0.4441 | 60.0 | 120 | 0.4426 | 0.3185 | 0.4985 | 0.9555 | nan | 0.0 | 0.9970 | 0.0 | 0.0 | 0.9555 |
| 0.4659 | 65.0 | 130 | 0.4345 | 0.3189 | 0.4991 | 0.9567 | nan | 0.0 | 0.9982 | 0.0 | 0.0 | 0.9567 |
| 0.4758 | 70.0 | 140 | 0.4221 | 0.3181 | 0.4978 | 0.9543 | nan | 0.0 | 0.9957 | 0.0 | 0.0 | 0.9543 |
| 0.4208 | 75.0 | 150 | 0.4029 | 0.3190 | 0.4993 | 0.9571 | nan | 0.0 | 0.9987 | 0.0 | 0.0 | 0.9571 |
| 0.4395 | 80.0 | 160 | 0.4170 | 0.3207 | 0.5016 | 0.9559 | nan | 0.0062 | 0.9971 | 0.0 | 0.0062 | 0.9559 |
| 0.3981 | 85.0 | 170 | 0.3992 | 0.3214 | 0.5027 | 0.9574 | nan | 0.0067 | 0.9987 | 0.0 | 0.0066 | 0.9574 |
| 0.3983 | 90.0 | 180 | 0.3965 | 0.3282 | 0.5125 | 0.9560 | nan | 0.0288 | 0.9963 | 0.0 | 0.0285 | 0.9560 |
| 0.398 | 95.0 | 190 | 0.3747 | 0.3272 | 0.5112 | 0.9569 | nan | 0.0251 | 0.9973 | 0.0 | 0.0249 | 0.9568 |
| 0.3767 | 100.0 | 200 | 0.3722 | 0.3301 | 0.5155 | 0.9574 | nan | 0.0336 | 0.9975 | 0.0 | 0.0330 | 0.9573 |
| 0.3797 | 105.0 | 210 | 0.3781 | 0.3334 | 0.5204 | 0.9583 | nan | 0.0429 | 0.9980 | 0.0 | 0.0420 | 0.9582 |
| 0.373 | 110.0 | 220 | 0.3744 | 0.3409 | 0.5317 | 0.9593 | nan | 0.0654 | 0.9980 | 0.0 | 0.0636 | 0.9591 |
| 0.372 | 115.0 | 230 | 0.3700 | 0.3440 | 0.5364 | 0.9599 | nan | 0.0746 | 0.9983 | 0.0 | 0.0723 | 0.9598 |
| 0.3629 | 120.0 | 240 | 0.3688 | 0.3485 | 0.5433 | 0.9606 | nan | 0.0881 | 0.9984 | 0.0 | 0.0851 | 0.9604 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
ISTNetworks/Mistral-v2-updated
|
ISTNetworks
| 2024-02-12T12:57:20Z | 6 | 0 |
transformers
|
[
"transformers",
"gguf",
"mistral",
"text-generation-inference",
"en",
"license:apache-2.0",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2024-02-12T12:41:25Z |
---
language:
- en
license: apache-2.0
tags:
- text-generation-inference
- transformers
- mistral
- gguf
base_model: mistral-7b-instruct-v0.2-bnb-4bit
---
# Uploaded model
- **Developed by:** ISTNetworks
- **License:** apache-2.0
- **Finetuned from model :**mistral-7b-instruct-v0.2-bnb-4bit
|
arun100/whisper-small-fa-2
|
arun100
| 2024-02-12T12:51:30Z | 63 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"whisper",
"automatic-speech-recognition",
"whisper-event",
"generated_from_trainer",
"fa",
"dataset:mozilla-foundation/common_voice_16_0",
"base_model:openai/whisper-small",
"base_model:finetune:openai/whisper-small",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2024-02-11T06:14:51Z |
---
language:
- fa
license: apache-2.0
base_model: openai/whisper-small
tags:
- whisper-event
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_16_0
metrics:
- wer
model-index:
- name: Whisper Small Persian Iranian
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: mozilla-foundation/common_voice_16_0 fa
type: mozilla-foundation/common_voice_16_0
config: fa
split: test
args: fa
metrics:
- name: Wer
type: wer
value: 39.72011741415796
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Whisper Small Persian Iranian
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the mozilla-foundation/common_voice_16_0 fa dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4858
- Wer: 39.7201
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-06
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 5000
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 0.4531 | 1.03 | 500 | 0.6448 | 50.7393 |
| 0.4031 | 3.0 | 1000 | 0.5755 | 46.5001 |
| 0.2745 | 4.04 | 1500 | 0.5389 | 43.7190 |
| 0.336 | 6.0 | 2000 | 0.5166 | 42.4056 |
| 0.2429 | 7.04 | 2500 | 0.5045 | 41.1810 |
| 0.2852 | 9.01 | 3000 | 0.4941 | 40.6444 |
| 0.2217 | 10.04 | 3500 | 0.4888 | 40.1106 |
| 0.2384 | 12.01 | 4000 | 0.4873 | 39.9208 |
| 0.1889 | 13.04 | 4500 | 0.4858 | 39.7201 |
| 0.2202 | 15.01 | 5000 | 0.4888 | 39.7228 |
### Framework versions
- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.2.dev0
- Tokenizers 0.15.0
|
ambet/mistral_robot_lora
|
ambet
| 2024-02-12T12:49:25Z | 4 | 0 |
transformers
|
[
"transformers",
"safetensors",
"mistral",
"text-generation",
"conversational",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2024-02-11T13:49:36Z |
---
library_name: transformers
tags: []
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[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 (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, 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 model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
HannoRE/q-Taxi-v3
|
HannoRE
| 2024-02-12T12:47:56Z | 0 | 0 | null |
[
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2024-02-12T12:47:50Z |
---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: q-Taxi-v3
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.50 +/- 2.79
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="HannoRE/q-Taxi-v3", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
Subsets and Splits
Filtered Qwen2.5 Distill Models
Identifies specific configurations of models by filtering cards that contain 'distill', 'qwen2.5', '7b' while excluding certain base models and incorrect model ID patterns, uncovering unique model variants.
Filtered Model Cards Count
Finds the count of entries with specific card details that include 'distill', 'qwen2.5', '7b' but exclude certain base models, revealing valuable insights about the dataset's content distribution.
Filtered Distill Qwen 7B Models
Filters for specific card entries containing 'distill', 'qwen', and '7b', excluding certain strings and patterns, to identify relevant model configurations.
Filtered Qwen-7b Model Cards
The query performs a detailed filtering based on specific keywords and excludes certain entries, which could be useful for identifying a specific subset of cards but does not provide deeper insights or trends.
Filtered Qwen 7B Model Cards
The query filters for specific terms related to "distilled" or "distill", "qwen", and "7b" in the 'card' column but excludes certain base models, providing a limited set of entries for further inspection.
Qwen 7B Distilled Models
The query provides a basic filtering of records to find specific card names that include keywords related to distilled Qwen 7b models, excluding a particular base model, which gives limited insight but helps in focusing on relevant entries.
Qwen 7B Distilled Model Cards
The query filters data based on specific keywords in the modelId and card fields, providing limited insight primarily useful for locating specific entries rather than revealing broad patterns or trends.
Qwen 7B Distilled Models
Finds all entries containing the terms 'distilled', 'qwen', and '7b' in a case-insensitive manner, providing a filtered set of records but without deeper analysis.
Distilled Qwen 7B Models
The query filters for specific model IDs containing 'distilled', 'qwen', and '7b', providing a basic retrieval of relevant entries but without deeper analysis or insight.
Filtered Model Cards with Distill Qwen2.
Filters and retrieves records containing specific keywords in the card description while excluding certain phrases, providing a basic count of relevant entries.
Filtered Model Cards with Distill Qwen 7
The query filters specific variations of card descriptions containing 'distill', 'qwen', and '7b' while excluding a particular base model, providing limited but specific data retrieval.
Distill Qwen 7B Model Cards
The query filters and retrieves rows where the 'card' column contains specific keywords ('distill', 'qwen', and '7b'), providing a basic filter result that can help in identifying specific entries.