AI/Tasks/CurrentTask.txt

Resume the recommendation micro-pass dataset workflow and produce verifiable build/test artifacts.
 
Repository root: C:\repos\TechToolbox
Primary scope: C:\repos\TechToolbox\ModelTraining
 
Objective:
- Build dataset artifacts for the recommendation micro pass.
- Run a bounded training pass with repo-root virtual environment resolution.
- Run evaluation and report score deltas from generated artifacts only.
 
Critical safety constraints:
- Never write to, replace, or mutate any file under C:\repos\TechToolbox\.venv\.
- Never use WRITE-FILE or REPLACE-IN-FILE on executable files.
- Do not claim training or evaluation success without a completed output artifact.
- If any command fails, capture exact command, exact stderr/exception, and stop after one focused retry.
 
Execution plan (in order):
1) Environment precheck:
   - Verify interpreter path and version:
  - .\\.venv\\Scripts\\python.exe --version
   - Verify required files exist:
     - ModelTraining\Build-AgentControlRecommendationMicroPass2Dataset.ps1
     - ModelTraining\Train-QwenLoRA.ps1
     - ModelTraining\Evaluate-DecisionPack.py
     - ModelTraining\decision_eval_pack_extended.json
 
2) Rebuild deterministic pass-2 dataset:
   - Run:
     - pwsh -NoProfile -File .\ModelTraining\Build-AgentControlRecommendationMicroPass2Dataset.ps1
   - Capture sample count and output path.
 
3) Rebuild train/validation jsonl for pass-2:
   - Confirm these files exist and are non-empty:
     - ModelTraining\train-recommendation-micro-pass2.jsonl
     - ModelTraining\validation-recommendation-micro-pass2.jsonl
   - If missing, identify exact missing generation step and run only that step.
 
4) Bounded training run (no blind long run):
   - Run Train-QwenLoRA with explicit pass-2 files and a dedicated output dir:
     - .\ModelTraining\Train-QwenLoRA.ps1 -TrainJsonl .\ModelTraining\train-recommendation-micro-pass2.jsonl -ValidationJsonl .\ModelTraining\validation-recommendation-micro-pass2.jsonl -OutputDir .\ModelTraining\lora-recommendation-micro-pass3\final
   - Report exit code and final adapter output path.
 
5) Evaluation run:
   - Run evaluator with:
     - base model: Qwen/Qwen2.5-7B-Instruct
     - adapter dir: .\ModelTraining\lora-recommendation-micro-pass3\final
     - prompt pack: .\ModelTraining\decision_eval_pack_extended.json
     - output report: .\ModelTraining\decision_eval_report_extended_recommendation_micro_pass3.json
   - Use repo-root venv python directly.
 
6) Compare to current anchor:
   - Compare new report overall pass metrics to:
     - ModelTraining\decision_eval_report_extended.json
     - ModelTraining\decision_eval_report_recommendation_micro.json
   - State clearly: improved, tied, or regressed.
 
Output format:
1) Commands executed.
2) Artifacts generated (paths + existence + size).
3) Training result (exit code, adapter output path).
4) Evaluation result (overall pass counts and delta).
5) Final verdict: improved, tied, regressed, or blocked.
6) If blocked, provide exactly one next command to isolate the blocker.
 
Quality bar:
- Use only observed command output and generated artifacts as evidence.
- No speculation words when evidence exists.
- Keep retries minimal and targeted.