Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →nvidia/torch-harmonics-skills
Index of project-local Claude Code skills for torch-harmonics. Use this to discover which skills are available for this codebase. Individual skills live in the .skills/ directory at the project root.
Passed 16 of 21 test cases with the skill, and 12 of 21 without it.
and 4 more in the full report
5 fixed12 passed either way4 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 4 test cases the skill fixed. Same model, same prompt, with and without the skill.
When testing backward passes for DISCO kernels in torch-harmonics, what check should be performed to verify that non-differentiable or integer arguments do not trigger autograd error failures?
| Checked | Without skill | With skill |
|---|---|---|
| Passes requires_grad=True only to floating-point input tensors in gradcheck. | Passed | Passed |
| Sets requires_grad=False on discrete grid dimensions or mode count arguments. | Missed | Passed |
The instructions your model reads when the skill loads.
Project-local skills live in .skills/<skill-name>/SKILL.md.
| Skill | Trigger | Description |
|---|---|---|
| test-writing | Writing or extending tests (serial or distributed), tolerance values, comparing tensors, gradient testing, autocast in tests | Guide for writing numerical tests following torch-harmonics conventions |
| disco-kernel-dev | Adding/optimizing/debugging DISCO kernels, wiring dispatch layers, bf16/fp16 autocast, distributed a2a, test coverage | Expert guide for DISCO convolution kernel development |
test cases passed, vs 12 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.