|
Download README.md from weizhou03/lbs-dmax-data: direct link, hf CLI and curl.
- Browser
- Download file 585 Bytes
-
https://huggingface.co/datasets/weizhou03/lbs-dmax-data/resolve/main/README.md
- Command line
-
hf download hf://datasets/weizhou03/lbs-dmax-data/README.md
-
curl -L -o README.md https://huggingface.co/datasets/weizhou03/lbs-dmax-data/resolve/main/README.md
585 Bytes
| # LBS on DMax: data | |
| Private working data for the EB-Decoding LBS-on-DMax rebuttal experiments (code: EB-Decoding branch `dmax-lbs`, folder `dmax/`). | |
| - `trajectories/dmax16b_traj/`: DMax-16B self-generated trajectories (math.parquet 8,558 rows, code.parquet 11,990 rows; question/answer). | |
| - `features/`: router training features (`feats_op_*` from DMax-16B trajectories; `feats_inblock*` from the LLaDA-2.0-mini trajectories released by DMax; `feats_sdar`, `feats` from earlier integrations). | |
| - `runs/`: per-question eval outputs (`python summarize.py runs/<name>`); `logs/`: run logs. | |