Downloads · 30 days
0
dougalldeepmind/2026-08-26-qwen36-lora-table2-9284-post-action-retrospection-716-rank-64-dynbatch
2026-08-26-qwen36-lora-table2-9284-post-action-retrospection-716-rank-64-dynbatch is a machine learning model from dougalldeepmind. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
0
Access
Public
Updated Aug 31, 2026
Repo size
1.3 GB
Likes
0
Public
Click a slice to open those files.
.safetensors1.3 GB · 98%
From the Hugging Face model README
| field | value |
|---|---|
experiment | LoRA SFT adapter -- lora_qwen36_t2_9284_par716_dynbatch_2xh200. THE POST-ACTION-RETROSPECTION ARM (design B): Qwen3.6-27B trained on 9,284 Table-2 rows + 716 five-turn rows -- a difficult-advice prompt, a bare refusal (Sonnet, form instructed, never trained), the person's pushback, and the trained turn doing the reasoning the refusal skipped under DA's own rewrite contract; only the last assistant turn is in the loss. Same organism as the da716 arm (LASR-Callum/2026-08-14-qwen36-lora-table2-9284-difficult-advice-716-rank-64-dynbatch) in every respect but the 716 rows' shape. |
date_generated | 2026-08-26 |
constitution | constitutions/claude_distilled_12_principles_mid/constitution.md (9 principles), the same as difficult advice's; inherited from LASR-Callum/2026-08-26-table2-9284-post-action-retrospection-716-train |
source_repo | https://github.com/Matthew-Bozoukov/Lessons_from_constituitional_AFT.git @ ccb2d4b31f931f66482e88a399049d74142881b0 |
models | base: Qwen/Qwen3.6-27B |
generation_config | {"seed": 0, "thinking": true, "epochs": 1, "lr": 0.0001, "batch_size": 1, "grad_accum": 16, "max_seq_len": 8192, "dynamic_batching": "token budget from ModelProfile.train_memory (H200 8000)", "ddp_ranks": 2, "lora": {"r": 64, "alpha": 128, "dropout": 0.05}} |
schema | PEFT LoRA adapter (safetensors) + tokenizer + training_meta.json {thinking, train_config, base_model, dataset{repo,file,revision}, git_sha, timestamp} |
provenance | torchrun --nproc_per_node=2 scripts/train/train_lora.py --config configs/train/2026-08-27_lora_qwen36_table2_9284_post_action_retrospection_716_dynbatch.yaml push=false (on pod), then uv run python scratch/par_b/train_pod.py push |
dataset | hf.co/datasets/LASR-Callum/2026-08-26-table2-9284-post-action-retrospection-716-train@42c8a7407b2111443d86479dd0517bc0dedab406 (mixture_think.jsonl) |
comparison_arm | LASR-Callum/2026-08-14-qwen36-lora-table2-9284-difficult-advice-716-rank-64-dynbatch |