Menu
Run GPU smoke checks for lra and pdebench models in this repo (compile, instantiate, strict state_dict reload, and CUDA forward pass). Use this whenever validating that model wiring/loading/execution still works after refactors or syncs.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | gpu-model-smoke |
| description | Run GPU smoke checks for lra and pdebench models in this repo (compile, instantiate, strict state_dict reload, and CUDA forward pass). Use this whenever validating that model wiring/loading/execution still works after refactors or syncs. |
Use this skill when the user asks to validate that models in lra and pdebench compile, load, and run on GPU.
.venv exists with project dependencies.torch.cuda.is_available()).source .venv/bin/activate && python tests/gpu/smoke_models_gpu.py
compile_errors=0unexpected_failures=0pdebench.UPT may appear under expected_failures (known incomplete implementation in this repo).tests/gpu/smoke_models_gpu.py.unexpected_failures=0.