메뉴
Task: {{ task }} Final best success_rate: {{ best_score }} (target was {{ target_success }}). Write the markdown report for this hill-climb run.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | report_narrative |
| description | Task: {{ task }} Final best success_rate: {{ best_score }} (target was {{ target_success }}). Write the markdown report for this hill-climb run. |
| tools | ["read_file","write_file","run_command","git_log"] |
SKILL_ID: report_narrative
You are writing the final report for an automated hill-climb over LeWM training on OGBench-Cube. First gather the facts:
research_log.md — the goal, the paper's reference numbers, and one line per
experiment (keep = improved and was committed, revert = did not improve).experiments.jsonl — structured ledger: each record has the candidate
score, the running best, status, and the proposal text that produced it.git_log — the kept experiments are commits on this branch
(saage: keep experiment, ...).config/train/lewm.yaml / config/train/model/lewm.yaml — the
winning configuration that survived.Then write report_narrative.md in markdown. Be ACCURATE — use only the real
scores and experiments from those files; do not invent results. Include:
Baseline score at the fixed {{ train_epochs }}-epoch budget, the final best score, and how both compare to the paper's 74% target and the earlier manual runs (60-64% at epoch 54 of a 100-epoch run).
The KEPT experiments in order: what changed, why it plausibly helped, score before -> after. Then notable reverted ideas and why a fixed short budget might reject them.
2-4 bullets — e.g. which surviving config to scale up to a full-length training run, and what to try next if the target was not reached.
Write ONLY to report_narrative.md. Finish with a one-line confirmation.
Run a tiny human-in-the-loop session: ask the user a couple of questions on the console, then write a short personalized note from their answers.
Task: {{ task }} Implement (or, if a proposal is given below, modify) the ML pipeline so it trains and evaluates end-to-end. Target metric to beat: {{ target_accuracy }}. Proposed change for THIS experiment (empty on the baseline): {{ current_proposal | default("(none — build a simple baseline)") }}
Task: {{ task }} Best VALIDATION accuracy (the hill-climb selection metric): {{ best_score }} (target {{ target_accuracy }}, higher is better). Held-out TEST accuracy of the retrained winner — the HEADLINE number, selected on validation and reported once on the test set: {{ final_test_score }}. Write the final HTML research report for this ML auto-research run.
Review the applied change against the proposal, check the contract, and run the smoke tests. Decide pass or fail. Proposal that was supposed to be applied: {{ current_proposal | default("(none — baseline build)") }}
Competition: {{ competition_id }} Metric: {{ metric_name }} ({{ "lower is better" if lower_is_better else "higher is better" }}). Final best validation score: {{ best_score }} (target {{ target_score }}). Write the final HTML report for this kaggle-solver run.
Condense the current kaggle experiment proposal into one short paragraph for the running research log.