Update README.md
Browse files
README.md
CHANGED
@@ -1,34 +1,50 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
dataset_info:
|
4 |
-
features:
|
5 |
-
- name: repo_name
|
6 |
-
dtype: string
|
7 |
-
- name: sub_path
|
8 |
-
dtype: string
|
9 |
-
- name: file_name
|
10 |
-
dtype: string
|
11 |
-
- name: file_ext
|
12 |
-
dtype: string
|
13 |
-
- name: file_size_in_byte
|
14 |
-
dtype: int64
|
15 |
-
- name: line_count
|
16 |
-
dtype: int64
|
17 |
-
- name: lang
|
18 |
-
dtype: string
|
19 |
-
- name: program_lang
|
20 |
-
dtype: string
|
21 |
-
- name: doc_type
|
22 |
-
dtype: string
|
23 |
-
splits:
|
24 |
-
- name: The_Stack_V2
|
25 |
-
num_bytes: 46577045485
|
26 |
-
num_examples: 336845710
|
27 |
-
download_size: 20019085005
|
28 |
-
dataset_size: 46577045485
|
29 |
-
configs:
|
30 |
-
- config_name: default
|
31 |
-
data_files:
|
32 |
-
- split: The_Stack_V2
|
33 |
-
path: data/The_Stack_V2-*
|
34 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
dataset_info:
|
4 |
+
features:
|
5 |
+
- name: repo_name
|
6 |
+
dtype: string
|
7 |
+
- name: sub_path
|
8 |
+
dtype: string
|
9 |
+
- name: file_name
|
10 |
+
dtype: string
|
11 |
+
- name: file_ext
|
12 |
+
dtype: string
|
13 |
+
- name: file_size_in_byte
|
14 |
+
dtype: int64
|
15 |
+
- name: line_count
|
16 |
+
dtype: int64
|
17 |
+
- name: lang
|
18 |
+
dtype: string
|
19 |
+
- name: program_lang
|
20 |
+
dtype: string
|
21 |
+
- name: doc_type
|
22 |
+
dtype: string
|
23 |
+
splits:
|
24 |
+
- name: The_Stack_V2
|
25 |
+
num_bytes: 46577045485
|
26 |
+
num_examples: 336845710
|
27 |
+
download_size: 20019085005
|
28 |
+
dataset_size: 46577045485
|
29 |
+
configs:
|
30 |
+
- config_name: default
|
31 |
+
data_files:
|
32 |
+
- split: The_Stack_V2
|
33 |
+
path: data/The_Stack_V2-*
|
34 |
+
---
|
35 |
+
|
36 |
+
|
37 |
+
This dataset consists of meta information (including the repository and file path of files) of the raw code data from **RefineCode**.
|
38 |
+
|
39 |
+
**Note:** Currently, we have uploaded the meta data covered by The Stack V2. We are working hard to make the remaining part available.
|
40 |
+
|
41 |
+
---
|
42 |
+
**RefineCode** is a **high-quality**, **reproducible** code pretraining corpus comprising **960 billion** tokens across **607** programming languages and **75 billion** code-related token recalled from web corpus, incorporating over **130** language-specific rules with customized weight assignments.
|
43 |
+
Our dataset shows better training efficacy and efficiency compared with the training subset of The Stack V2.
|
44 |
+
|
45 |
+
<img src="https://raw.githubusercontent.com/OpenCoder-llm/opencoder-llm.github.io/refs/heads/main/static/images/opencoder_banner.png" alt="OpenCoder banner" style="zoom:30%;" />
|
46 |
+
|
47 |
+
We also use PCA to visualize the embeddings extracted from CodeBERT for The Stack V2 and **RefineCode**, showing a clear advance of our pretraining dataset.
|
48 |
+
|
49 |
+
<img src="https://raw.githubusercontent.com/OpenCoder-llm/opencoder-llm.github.io/refs/heads/main/static/images/compare_refinecode_stack_v2.jpg" alt="Distribution Comparsion" style="zoom:50%;" />
|
50 |
+
|