AGENT SKILL

msfm_mechanism_evaluation

Evaluate whether an MSFM recovery result exercised BatchOT coupling, Joint CFM loss, optimizer evidence, and soft-mode proxy constraints.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
msfm_mechanism_evaluation
description
Evaluate whether an MSFM recovery result exercised BatchOT coupling, Joint CFM loss, optimizer evidence, and soft-mode proxy constraints.

02 资源文件 2

03 SKILL.md 正文

MSFM Mechanism Evaluation

Use this skill after a Multisample Flow Matching recovery experiment has produced recovery_result.json, training_trace.json, and generated skill invocation logs.

Do not accept a proxy run only because its metric is positive; this skill checks whether the recovery exercised the paper mechanisms and respected source/runtime boundaries.

Inputs

  • Recovery result JSON.
  • Training trace JSON.
  • Generated skill invocation log.
  • Optional minimum transport-cost reduction threshold.

Outputs

  • Mechanism pass/fail report.
  • Metric gap summary.
  • Refinement hints assigned to module-level causes.

Workflow

  1. Confirm proxy status is declared when full recovery is not claimed.
  2. Verify BatchOT and uniform couplings were evaluated and double-stochastic evidence passed.
  3. Verify Joint CFM loss and an optimizer step were executed.
  4. Verify reduced/full booleans are honest and no original repository was read.
  5. Confirm generated skill invocations cover coupling, loss, recovery, and evaluation modules.
  6. Return actionable hints for missing mechanism evidence.

Validation

Run:

python scripts/evaluate_mechanism.py --self-test
python tests/test_evaluate_mechanism.py

Limitations

  • This skill evaluates reduced/proxy mechanism evidence; it does not compare full ImageNet FID.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351