metadata
size_categories:
- 1K<n<10K
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.