Qwen3-1.7B QLoRA specialized for one falsifiable rule: diagnose only when two independent evidence channels agree; otherwise WATCH without naming a mode. Responses must identify the asset, cite independent evidence, and give exactly one non-compound action.
This is below the declared 0.95 bar and is not a threshold pass.
The adapter was trained against the listed bitsandbytes 4-bit base. Exact adapter evaluation requires CUDA. Applying the LoRA weights to upstream full-precision Qwen on CPU/MPS changes outputs and is non-comparable.
1python eval.py \
2 --model lecporr/rotating-equip-sft \
3 --revision 784a90897112cd94effb2da39f2fc151f55468ae \
4 --eval-set data/eval.jsonl \
5 --device cuda
A merged 16-bit checkpoint is required for portable CPU/MPS one-command inference; see notebooks/COLAB_MERGE.md.
Synthetic snapshots only. This model is not a physical diagnostic authority or safety controller. The 30-row public eval is small; staff-held-out evaluation remains required.
