ratchet-loop-qwen25-05b

LoRA adapter for mlx-community/Qwen2.5-0.5B-Instruct-4bit, trained locally on Apple Silicon with mlx-lm by the local-model-factory pipeline (prepare -> train -> eval-base -> eval-tuned -> gate -> publish).

Task: Synthetic ratchet-loop proposal task (self-generated 10k-example corpus, ledger: ratchet-loop-demo-synthetic) fine-tuned on Qwen 2.5 0.5B Instruct for local-only development-loop experiments. Demo/pipeline-proof adapter: it is only good at this synthetic task shape.

Gate decision: PROMOTE

• field_accuracy: 0.112 -> 1.0 (+0.888) meets min improvement 0.05
• json_validity: 0.92 -> 1.0 (no meaningful regression)
• exact_match: 0.0 -> 1.0 (no meaningful regression)

Held-out evaluation (same test split, deterministic)

metric	base model	base + this adapter
exact_match	0.000	1.000
examples	50.000	50.000
field_accuracy	0.112	1.000
json_examples	50.000	50.000
json_validity	0.920	1.000
mean_latency_s	0.283	0.332

Training

• iterations: 300, batch size: 4, learning rate: 0.0001, LoRA layers: 8, seed: 11
• final validation loss: 0.065
• data: 10000 raw examples → 8000 train / 1000 valid / 1000 test (seed 20260701, 0 duplicates dropped)
• data source: ratchet-loop-qwen25-05b task (tasks/ratchet-loop-qwen25-05b)
• data ledger: ratchet-loop-demo-synthetic

Use it

pip install mlx-lm

mlx_lm.generate --model mlx-community/Qwen2.5-0.5B-Instruct-4bit \
  --adapter-path <local-download-of-this-repo> \
  --prompt "..."

The full run artifacts (train config, both eval reports, gate decision, data manifest) are included in this repo so the numbers above can be audited against exactly what produced them.