# Stage 1 state

Task: `docs/stage1-training-task.md`. Settings and weights: `train/README.md`. Updated 2026-10-03 18:35.

## Done

- Step 0 (setup): `train/.venv` (Python 3.11.17, uv), `mlx-lm` 0.32.0 / `mlx` 0.32.3.
  Model `mlx-community/Qwen3.8-27B-4bit` at `~/models/Qwen3.8-27B-4bit` (affine 4 bit, group 64, 16.1 GB;
  base `Qwen/Qwen3.8-27B`). The Ollama NVFP4 weights do not load in mlx_lm (global scale).
- Server: `train/serve.sh` (`mlx_lm.server`, port 8080, thinking on with `reasoning_effort` medium,
  temperature 0.2, top_p 0.95, top_k 20, min_p 0, max tokens 32768). Tool-call test passed.
- Eval subset (Kral decision): `train/subset.json` (copy `runs/stage1/subset.json`), 25 tasks. The task
  document says "all accepted tasks" for step 2; Kral changed it to this subset. Use the same list after
  training.
- Runner: `train/baseline.py --label <label>` (one task at a time, results `runs/stage1/<label>.json`,
  run directories `runs/stage1/<label>/`).

## Running (detached)

- MLX server PID 59352, log `runs/stage1/server.log`.
- T01 test (budget 40, harness check), PID 59414, log `runs/stage1/baseline_t01.log`.
- Night chain `train/night_chain.sh` PID 60017, log `runs/stage1/night_chain.log`: after the DeepSeek reruns
  and the T01 test, it starts the baseline on the 25 tasks (budget 60), log `runs/stage1/baseline.log`,
  macOS notification at the end. Expected end: 4 October, morning to noon.

## Next

1. B2: read the last lines of `runs/stage1/baseline.log`; summary from `runs/stage1/baseline.json`
   (`t01_test_budget40` holds the T01 test result). Commit.
2. Step 1: `train/prepare.py` (real token counts with the base model tokenizer, length filter 16384,
   95/5 split by document, `train/data/train.jsonl` and `valid.jsonl`, report). A4H is not needed.
3. Step 2, second part: valid loss of the base model (`mlx_lm.lora --test` without adapter; check the
   options with `--help` first). Add it to `runs/stage1/baseline.json`.
4. Step 3 (training): Kral stops A4H; stop the MLX server; no other model loaded. Short test of 20
   iterations first.

## Notes

- The earlier T01 score 41.7 (run 103) used the Ollama NVFP4 weights. The baseline of 2026-10-03 (MLX 4 bit)
  is the new reference.
- Harness changes that matter for stage 1 runs: ADT activation fallback for PROG/FUNC (EPOD bug), G2 finds
  the FUNCTION statement after local classes, call budget floor 60, empty-turn retry in the agent
  (`max_tokens` only for cloud models; the MLX server limit is `--max-tokens 32768`).