README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*.parquet

LangMap-TheStack-java-100M

Code finetuning dataset for java streamed from bigcode/the-stack.

• Tokens collected: 100,000,000 (target: 100,000,000)
• Tokenizer: allenai/OLMo-3-1025-7B
• Schema: {"text": [...]} (sanitised source code)