Hugging Face's logo

Datasets:
aadarsh99
/
kyvo-datasets-and-codebooks

Modalities:
Image
License:
Dataset card Data Studio Files Community
2
nielsr HF Staff commited on
Commit
8fae39c
·
verified ·
1 Parent(s): dfa44e9

Add task category, link to paper, project page and Github repo

Browse files

This PR adds the `text-to-3d` task category to the dataset card, links the dataset to the paper, adds a link to the project page and adds a link to the Github repository.

Files changed (1) hide show
  1. README.md +12 -9
README.md CHANGED
@@ -1,9 +1,17 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
4
  # Kyvo Dataset and Codebooks Details
5
 
6
- This document provides details about the dataset and codebooks provided in the `kyvo-datasets-and-codebooks` repository. We will provide the details about each of the folders in the repository and the contents of each folder.
 
 
 
 
7
 
8
  ## Data Generation Pipeline
9
 
@@ -13,9 +21,6 @@ The pipeline that we follow to generate the pre-tokenized data is as follows:
13
  * **Images**: Image --> VQGAN Codebook Indices --> Tokenized Image
14
  * **Text**: Text --> Tokenized Text
15
 
16
-
17
-
18
-
19
  ## Pre-tokenized Data
20
 
21
  The `pretokenized-data` folder contains all the pre-tokenized data for the datasets used in the Kyvo project. The pre-tokenized data is stored in the following structure:
@@ -34,7 +39,6 @@ pretokenized-data/
34
  | |-- images/ # contains all pre-tokenized images for Objectron for all tasks
35
  ```
36
 
37
-
38
  For a given task, an input can be any combination of 3d-scenes, images, and text. The output can be any combination of images, text, and 3d-scenes. In the following table we outline the tasks for each dataset and the corresponding input and output data that are needed for each task.
39
 
40
  | **Task** | **Input Image** | **Input 3D Scene** | **Input Text** | **Output Image** | **Output 3D Scene** | **Output Text** |
@@ -54,7 +58,6 @@ For a given task, an input can be any combination of 3d-scenes, images, and text
54
 
55
  For the exact files that correspond to the input and output data for each task, please refer to the corresponding configuration files in the `configs/llama3_2/train` folder.
56
 
57
-
58
  ## VQGAN Models and Codebooks
59
 
60
  The `vqgan-models-and-codebooks` folder contains all the VQGAN model checkpoints and codebooks for the datasets used in the Kyvo project. The VQGAN model checkpoints and codebooks are stored in the following structure:
 
1
+ ---
2
+ license: apache-2.0
3
+ task_categories:
4
+ - text-to-3d
5
+ library_name: torchtune
6
+ ---
7
+
8
  # Kyvo Dataset and Codebooks Details
9
 
10
+ This document provides details about the dataset and codebooks provided in the `kyvo-datasets-and-codebooks` repository. This dataset is associated with the paper [Aligning Text, Images, and 3D Structure Token-by-Token](https://huggingface.co/papers/2506.08002).
11
+ The Github repository can be found here: https://github.com/AadSah/kyvo.
12
+ The project page can be found here: https://glab-caltech.github.io/kyvo/
13
+
14
+ We will provide the details about each of the folders in the repository and the contents of each folder.
15
 
16
  ## Data Generation Pipeline
17
 
 
21
  * **Images**: Image --> VQGAN Codebook Indices --> Tokenized Image
22
  * **Text**: Text --> Tokenized Text
23
 
 
 
 
24
  ## Pre-tokenized Data
25
 
26
  The `pretokenized-data` folder contains all the pre-tokenized data for the datasets used in the Kyvo project. The pre-tokenized data is stored in the following structure:
 
39
  | |-- images/ # contains all pre-tokenized images for Objectron for all tasks
40
  ```
41
 
 
42
  For a given task, an input can be any combination of 3d-scenes, images, and text. The output can be any combination of images, text, and 3d-scenes. In the following table we outline the tasks for each dataset and the corresponding input and output data that are needed for each task.
43
 
44
  | **Task** | **Input Image** | **Input 3D Scene** | **Input Text** | **Output Image** | **Output 3D Scene** | **Output Text** |
 
58
 
59
  For the exact files that correspond to the input and output data for each task, please refer to the corresponding configuration files in the `configs/llama3_2/train` folder.
60
 
 
61
  ## VQGAN Models and Codebooks
62
 
63
  The `vqgan-models-and-codebooks` folder contains all the VQGAN model checkpoints and codebooks for the datasets used in the Kyvo project. The VQGAN model checkpoints and codebooks are stored in the following structure: