agentleFS
MacroMan5MacroMan5/autotrain-yolo/.claude/skills/analyze-results/SKILL.md

analyze-results

Skill8 starsChanged 8 months agoMITView on GitHub

Analyze YOLO training runs — compares to baseline/best, checks per-class regression, analyzes training dynamics and tune convergence, writes actionable recommendations.

11 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use experiment-plan.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • raja21068Skill

    experiment-plan

    Turn a refined research proposal or method idea into a detailed, claim-driven experiment roadmap. Use after `research-refine`, or when the user asks for a detailed experiment plan, ablation matrix, evaluation protocol, run order, compute budget, or paper-ready validation that supports the core problem, novelty, simplicity, and any LLM / VLM / Diffusion / RL-based contribution.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write experiments/analysis.md with findings and recommendations.

    line 53
  2. Distinctive:

    Keep analysis concise — focus on what changed and what to try next, not restating all historical metrics.

    line 60

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Analyze Results

Single-pass analysis run after each experiment or on-demand via /analyze.

Data

Read:

  • Latest experiment's report + results.csv or tune_results.csv
  • Current best metrics from experiments/summary.md journal entries
  • experiments/dataset_profile.yaml
  • training-plan.md (thresholds, goals, constraints)

Results.csv Lookup Order

  1. experiments/exp_NNN_name/train/results.csv
  2. experiments/exp_NNN_name/results.csv
  3. Fall back to metrics.yaml — note that training dynamics analysis is unavailable

For tune runs, also check:

  • experiments/exp_NNN_name/tune/tune_results.csv

Analysis

Comparison: Latest vs baseline AND current best (mAP50-95, per-class AP)

Per-class regression check: Flag any class exceeding training-plan.md threshold vs current best

Training dynamics (if results.csv available):

  • Convergence speed: epochs to reach 90% of final mAP
  • Overfitting onset: epoch where val loss diverges from train loss
  • Loss plateau: epochs with <0.1% change in val loss
  • LR effectiveness: correlation between LR schedule phase and metric improvement

Tune analysis (if tune_results.csv available):

  • Tune convergence: are later iterations improving or stalled?
  • Best params found vs defaults — which params moved the most?
  • Top-N trial spread: how tight is the performance range?

Diminishing returns detection (when 3+ experiments show <0.5% change):

  • Same-category stall (all augmentation, or all LR variations): that lever is exhausted, try a different category
  • Cross-category stall (multiple lever types tried): model may be at ceiling for this dataset/architecture
  • Recommendation differs per stall type — surface this explicitly

Recommendations: Ordered by expected impact. Each cites evidence from the analysis.

Output

Write experiments/analysis.md with findings and recommendations. Print top 3 findings + top 3 recommendations.

Findings also inform the journal entry's learning field and the next DIAGNOSE step in the reasoning loop.

Important

  • The reasoning loop's DIAGNOSE step reads analysis.md from the previous experiment or last explicit /analyze. It does NOT read analysis of the experiment that just ran before it has been analyzed.
  • Keep analysis concise — focus on what changed and what to try next, not restating all historical metrics.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.