Update README.md
Browse files
README.md
CHANGED
|
@@ -14,13 +14,14 @@ Using the [unreal-engine-5-code](https://huggingface.co/datasets/AdamCodd/unreal
|
|
| 14 |
|
| 15 |
### Dataset Description
|
| 16 |
Branches:
|
| 17 |
-
|
| 18 |
- Dataset is split/chunked by engine module (no max chunk size)
|
| 19 |
-
|
| 20 |
- Data is first split by module `{module_name}.jsonl` if ≤ 8000 tokens.
|
| 21 |
- If a module is ≥ 8000 tokens then it's further split by header name `{module_name}_{header_name}.jsonl`
|
| 22 |
- If a header is ≥ 8000 tokens then it's incrementally split `{module_name}_{header_name}_{chunk_number}.jsonl`
|
| 23 |
-
|
|
|
|
| 24 |
### Dataset Sources
|
| 25 |
|
| 26 |
- **Forked from:** [unreal-engine-5-code by AdamCodd](https://huggingface.co/datasets/AdamCodd/unreal-engine-5-code)
|
|
|
|
| 14 |
|
| 15 |
### Dataset Description
|
| 16 |
Branches:
|
| 17 |
+
**`main`**:
|
| 18 |
- Dataset is split/chunked by engine module (no max chunk size)
|
| 19 |
+
**`chunked-8k`**:
|
| 20 |
- Data is first split by module `{module_name}.jsonl` if ≤ 8000 tokens.
|
| 21 |
- If a module is ≥ 8000 tokens then it's further split by header name `{module_name}_{header_name}.jsonl`
|
| 22 |
- If a header is ≥ 8000 tokens then it's incrementally split `{module_name}_{header_name}_{chunk_number}.jsonl`
|
| 23 |
+
- All markdown URI syntax (`[Link](Link/Address)`) has been removed, leaving just the relevant `Link` text to preserve the data
|
| 24 |
+
|
| 25 |
### Dataset Sources
|
| 26 |
|
| 27 |
- **Forked from:** [unreal-engine-5-code by AdamCodd](https://huggingface.co/datasets/AdamCodd/unreal-engine-5-code)
|