| # FineWeb-Edu document Causal LM pack (DocumentDataset). | |
| # Build with tools/pack_document_data.py, then pass data.path=<dir> (DATA_PATH in scripts/train.sh). | |
| format: document | |
| path: ??? # Override at CLI: data.path=/path/to/packed_data | |
| val_path: null | |
| target_only: false | |