mrzjy commited on
Commit
a83f2d9
·
verified ·
1 Parent(s): 06522fc

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +14 -2
README.md CHANGED
@@ -5,7 +5,7 @@ license: cc-by-sa-4.0
5
  # Genshin Impact Fandom Wiki Multimodal Dataset
6
 
7
  ## Description
8
- This dataset is a comprehensive collection of 22,162 fandom wiki pages for the popular game *Genshin Impact*.
9
 
10
  The dataset includes *markdown*-formatted content from the wiki, featuring interleaved text, as well as image, video, and audio file links. Additionally, the associated multimodal files (images, videos, and audio) have been downloaded and organized to facilitate the multimodal dataset usage.
11
 
@@ -14,6 +14,12 @@ The dataset includes *markdown*-formatted content from the wiki, featuring inter
14
  - **Videos**: Embedded gameplay clips, tutorials, etc, mostly in `mp4`, `gif` format.
15
  - **Audio**: Sound effects, voice lines associated with the game, mostly in `.ogg` format.
16
 
 
 
 
 
 
 
17
  ## Intended Use
18
  This dataset is intended for research and development in multimodal model pretraining, fine-tuning, and game studies. Potential use cases include:
19
  - Training and evaluating multimodal models that can process and generate text, images, and audio.
@@ -23,7 +29,13 @@ This dataset is intended for research and development in multimodal model pretra
23
  ## Dataset Structure
24
  The dataset is organized as follows:
25
  - `genshin.jsonl`: Contains markdown content as well as some meta info for each wiki page.
26
- - `images/`: Stores downloaded multimodal files (images, videos, and audio) referenced in the markdown content.
 
 
 
 
 
 
27
 
28
  ### Example
29
 
 
5
  # Genshin Impact Fandom Wiki Multimodal Dataset
6
 
7
  ## Description
8
+ This dataset is a comprehensive collection of **22,162** fandom wiki pages for the popular game *Genshin Impact*.
9
 
10
  The dataset includes *markdown*-formatted content from the wiki, featuring interleaved text, as well as image, video, and audio file links. Additionally, the associated multimodal files (images, videos, and audio) have been downloaded and organized to facilitate the multimodal dataset usage.
11
 
 
14
  - **Videos**: Embedded gameplay clips, tutorials, etc, mostly in `mp4`, `gif` format.
15
  - **Audio**: Sound effects, voice lines associated with the game, mostly in `.ogg` format.
16
 
17
+ ### Statistics
18
+
19
+ The total number of tokens in the markdown content is **47,674,290**, as calculated by `tiktoken`.
20
+ For individual markdown files, the maximum token count is 655,266, and the minimum is only 38.
21
+
22
+
23
  ## Intended Use
24
  This dataset is intended for research and development in multimodal model pretraining, fine-tuning, and game studies. Potential use cases include:
25
  - Training and evaluating multimodal models that can process and generate text, images, and audio.
 
29
  ## Dataset Structure
30
  The dataset is organized as follows:
31
  - `genshin.jsonl`: Contains markdown content as well as some meta info for each wiki page.
32
+ - `images/`: Stores 184,773 downloaded multimodal files (images, videos, and audio) referenced in the markdown content.
33
+
34
+ - Multimodal Distribution
35
+
36
+ ```
37
+ {'ogg': 152274, 'png': 28762, 'mp4': 2515, 'gif': 422, 'other': 390, 'jpg': 332, 'svg': 44, 'jpeg': 23, 'webp': 11}
38
+ ```
39
 
40
  ### Example
41