|
|
--- |
|
|
size_categories: |
|
|
- 1K<n<10K |
|
|
--- |
|
|
Repo [PythonCurseforgeScraper](https://github.com/MrDoritos/PythonCurseforgeScraper) |
|
|
|
|
|
Current dataset is 2k different addons and the 500k+ source files from each. For reference there is a total of about 250k total addons, 50k mods. |
|
|
|
|
|
Search criteria ignores binary files. |
|
|
|
|
|
### TODO |
|
|
There is no ability to specify specific versions. That might be useful. |
|
|
|
|
|
Create descriptions for each row using another LLM by exporting metadata from a scrape instance as a RAG reference for specific Minecraft context. |
|
|
|
|
|
Increase dataset by 10-20x, near the limit of available source count. |
|
|
|
|
|
Train using the dataset for minecraft mod code completion. |
|
|
|
|
|
Create another dataset using decompiled mods..!? All the mods for all the versions. |
