Datasetpublic

dougalldeepmind/2026-08-24-grok-responder-703-paired-bundle

2026-08-24-grokresp703-paired-bundle field value experiment Generator ablation ARM B (treatment): the SAME 703 questions answered by x-ai/grok-4.6 + 9,284 Table2. source_repo Matthew-Bozoukov/Lessons_from_constituitional_AFT @ b57a5afc94613752e8fd288c8315a190b9439b26 train_config configs/train/2026-08-24_lora_qwen36_table2_9284_grok_responder_703_paired.yaml (inside code.tar.gz) mixture t2_9284_grokresp703_10k.jsonl provenance uv run python… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-08-24-grok-responder-703-paired-bundle.

sourceHugging Faceapache-2.0updated 27d agoView on Hugging Face

0likes47downloads

Dataset Card

2026-08-24-grokresp703-paired-bundle

field	value
`experiment`	Generator ablation ARM B (treatment): the SAME 703 questions answered by x-ai/grok-4.6 + 9,284 Table2.
`source_repo`	Matthew-Bozoukov/Lessonsfromconstituitional_AFT @ b57a5afc94613752e8fd288c8315a190b9439b26
`train_config`	`configs/train/2026-08-24_lora_qwen36_table2_9284_grok_responder_703_paired.yaml` (inside code.tar.gz)
`mixture`	`t2_9284_grokresp703_10k.jsonl`
`provenance`	uv run python scripts/gpu/runpodtrain.py up --bundle LASR-Callum/2026-08-24-grok-responder-703-paired-bundle --trainconfig configs/train/2026-08-24loraqwen36table29284grokresponder703paired.yaml --mixture t29284grokresp70310k.jsonl --gpu 'NVIDIA H200' --gpucount 2
`notes`	Training bundle, not a corpus: code.tar.gz + the rendered mixture. Public because the training pod holds no credentials and reads it anonymously. Paired sibling: LASR-Callum/2026-08-24-sonnet-703-paired-bundle. grok answers are 1.70x shorter; length-AUC between arms is 0.864, so report length as a covariate.