---

name: training-evaluation description: "Builds and troubleshoots FCOS training, evaluation, distributed launch, checkpoint, and benchmark-style command workflows." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

FCOS Training and Evaluation

Use this sub-skill when the user asks for FCOS training, COCO/VOC/Cityscapes evaluation, distributed launch, checkpoint handling, AP reproduction, MODEL.WEIGHT, OUTPUT_DIR, TEST.IMS_PER_BATCH, or OOM/runtime failures during train/test workflows.

Start here

1. Read references/cli-reference.md for flags and override semantics.
2. Read references/workflows.md for single-GPU, multi-GPU, evaluation, and OOM-safe command recipes.
3. Use scripts/build_train_command.py and scripts/build_eval_command.py to produce commands without starting long jobs.
4. Use scripts/remove_solver_states.py to create a checkpoint copy without optimizer/scheduler/iteration state.
5. Read references/checkpoints.md when changing MODEL.WEIGHT, resuming, or publishing weights.
6. Read references/troubleshooting.md for dataset, CUDA/NCCL, OOM, SyncBN, and benchmark limitations.

Boundaries

• Route YAML selection, dataset layout validation, and custom dataset planning to ../data-configs/SKILL.md.
• Route one-image detector API or display demos to ../inference-demo/SKILL.md.
• Route ONNX export to ../onnx-export/SKILL.md.
• Route source-code changes or test triage to ../internals-maintenance/SKILL.md.

Backend/cost rule

Training and full evaluation are not safe smoke tests. They require datasets, weights, compiled extensions, and usually GPUs. Use command builders and config validation first; run full jobs only after the user approves data, weight downloads, GPU use, output locations, and expected runtime.