Hugging Face's logo

Datasets:
tokyotech-llm
/
swallow-code-v0.1

Tasks:
Text Generation
Languages:
English
Japanese
Tags:
code
License:
Dataset card Files Community
kazukifujii commited on
Commit
b2b7ddc
·
verified ·
1 Parent(s): 8b3ec48

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -0
README.md CHANGED
@@ -66,6 +66,12 @@ The dataset is released in four versions:
66
  - **Swallow Code v0.1 stage 3**: 20B tokens, 24M documents containing Python scripts that are filtered with pylint score.
67
  - **Swallow Code v0.1 stage 4**: 16B tokens, 21M documents containing Python scripts that are filtered with code comments and literal language detection(English and Japanese).
68
 
 
 
 
 
 
 
69
  ## Dataset Schema
70
 
71
  ```python
 
66
  - **Swallow Code v0.1 stage 3**: 20B tokens, 24M documents containing Python scripts that are filtered with pylint score.
67
  - **Swallow Code v0.1 stage 4**: 16B tokens, 21M documents containing Python scripts that are filtered with code comments and literal language detection(English and Japanese).
68
 
69
+ ## Results and Performance
70
+
71
+ Llama-3.1-8B Performance after Continual Pretraining on 50B tokens Japanese, English, and Code(= swallow-code-v0.1) datasets.
72
+
73
+ ![](swallow-code-eval.png)
74
+
75
  ## Dataset Schema
76
 
77
  ```python