菜单
Diagnose a training failure or stall by inspecting recent logs, loss curves, OOM traces, NaN events, and config. Activate when the user asks "why did my training crash", "loss went to NaN", "OOM during step X", "training is not improving", or "help me debug this run". Delegates to the failure-diagnoser agent.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Run a short, reproducible benchmark of one optimizer step (forward + backward + optimizer step over N microbatches) using the project's registered runtime. Activate when the user asks to "benchmark a training step", "measure throughput", "time one optimizer step", or "smoke test the runtime". Wraps run_accumulated_step from curry_train.benchmark.
Lightning Fabric integration recipe — minimal 5-line setup that gives DDP / FSDP / mixed precision / mixed-precision while keeping a raw PyTorch training loop. Activate when the user asks "Lightning Fabric", "torchrun", "DDP setup", "FSDP setup", "mixed precision", or wires up the launch script.
Bootstrap a new curryTrain project by copying the framework Python template into the user's working directory. Use when the user runs /curry-train:init, asks to "start a new training project with curryTrain", "initialize curryTrain", "scaffold a curryTrain project", or wants the framework code copied locally for editing.
Scaffold a new model + dataset + config triple inside an existing curryTrain project. Activate when the user asks to "add a new model to curryTrain", "scaffold a new experiment", "generate model.py / dataset.py / config.yaml for X", or describes an idea they want to start training. Delegates the actual generation to the scaffolder agent.
Compare two training runs and produce a concise markdown diff covering config, key metrics, loss curves, and grad-norm trajectory. Activate when the user asks to "compare run A and run B", "diff two experiments", "did this change actually help", or "is this run better than the previous one". This is both the implementation of the action and the methodology guide for variance-aware comparison.
Define an abort condition before launching a run, so that a clearly broken or clearly under-performing run stops itself instead of consuming the full compute budget. Activate when the user asks "when should I kill a run", "abort condition", "early stop a bad run", "kill criterion", or before any expensive run.
| name | diagnose |
| description | Diagnose a training failure or stall by inspecting recent logs, loss curves, OOM traces, NaN events, and config. Activate when the user asks "why did my training crash", "loss went to NaN", "OOM during step X", "training is not improving", or "help me debug this run". Delegates to the failure-diagnoser agent. |
Read the artifacts of a recent run, classify the failure mode, and produce a ranked list of likely causes with concrete fixes. Most of the analytical work is delegated to the failure-diagnoser agent.
$1 (optional): path to a run directory or a single log file. If omitted, look for the most recently modified directory under runs/ or outputs/ in the current project.Resolve the target. If the user gave a directory, use it. Else find the most recent runs/<timestamp>/ or outputs/<timestamp>/. If nothing exists, ask the user where their logs are.
Collect the evidence bundle. Read at most:
*.log or stderr.txt in the directory.config.yaml (the resolved Hydra config).metrics.jsonl (last 200 entries).git_sha.txt, seed.txt, cuda_info.txt if present.core.* traceback or Python traceback files.Spawn the failure-diagnoser agent. Hand it the evidence bundle and ask for a structured diagnosis (see agents/failure-diagnoser.md). Output schema:
classification: <NaN | OOM | divergence | stall | crash | unclear>
primary cause: <one line>
contributing factors: <bulleted>
fix candidates: <ranked, each with concrete code or config change>
verification next step: <a single command or skill to confirm the fix>
Suggest the verification skill. Match the classification to one of:
stage5-loss-spike-rollback (recipe to recover) or stage2-init-loss-check (early sanity).primitive-recompute (activation checkpointing) or batch-size reduction.stage2-overfit-single-batch (verify pipeline at all) or stage2-grad-flow-viz (find dead layers).classification: unclear and ask for more information./tmp/ or to a job scheduler.