Model training & evaluation

The top-rated launch-quality Model training & evaluation skill for AI agents is nvflare-convert-pytorch (grade A), followed by nvflare-shared and nemotron-retrieval-recipes. Skillspace rates 636 Model training & evaluation skills; 194 are launch-quality.

Training classical and deep models, PyTorch, scikit-learn, evaluation and experiment tracking.

Launch-quality skills first, then by grade, then by our Skill Score, which adds judge reviews, test results, GitHub stars and adoption.

  1. nvflare-convert-pytorch A
  2. nvflare-shared A
  3. nemotron-retrieval-recipes B
  4. nemo-mbridge-recipe-recommender B
  5. training-llms-megatron B
  6. huggingface-accelerate B
  7. pytorch-lightning B
  8. explore-code B
  9. nemo-mbridge-perf-sequence-packing B
  10. create-tiny-model B
  11. ray-train B
  12. simpo-training B
  13. nvflare-autofl B
  14. alterlab-pufferlib B
  15. optimizing-attention-flash B
  16. sparse-autoencoder-training B
  17. weights-and-biases B
  18. distributed-llm-pretraining-torchtitan B
  19. pytorch-fsdp2 B
  20. tdhook B
  21. minimal-run-and-audit B
  22. explore-run B
  23. run-train B
  24. rl-execution B
  25. i4h-workflow-dataset-convert B

Frequently asked questions

What is the best Model training & evaluation skill for AI agents?

The top-rated launch-quality Model training & evaluation skill for AI agents is nvflare-convert-pytorch (grade A), followed by nvflare-shared and nemotron-retrieval-recipes. Skillspace rates 636 Model training & evaluation skills; 194 are launch-quality.

How does Skillspace rate Model training & evaluation skills?

Skillspace grades every Model training & evaluation skill from A to F with a published scoring engine that reads the skill's SKILL.md: architecture and maturity make the score, and a security scan can cap it. In crowded categories, a judge model also compares the strongest contenders side by side, and some skills are tested head-to-head on real tasks against other skills in the same category (including a deliberately weak one) and a no-skill baseline. Launch-quality skills are graded A to C with a clean safety verdict, and not excluded by a quality filter: the filters catch skills our model's safety screen flags for prompt injection, guardrail bypass, harmful content or credential exposure; test fixtures and attack benchmarks; skills written only for the repository they live in; templates and stubs; and near-duplicate variants and copies of other skills. Ratings were last updated on 24 September 2026; the full method is on the methodology page.

Are Model training & evaluation skills safe to install?

Of the 636 Model training & evaluation skills Skillspace rates, 600 passed its safety scan with no flag from its model, and 36 are flagged: 32 as risky, 3 as unsafe and 1 as a test fixture. Each skill's page shows its safety verdict. A pass is not a guarantee: the scan reads only the skill's SKILL.md, not the scripts or other files it ships, and looks for patterns such as piping a remote script into a shell, reading credentials or hiding actions from the user. Skills are third-party code that Skillspace indexes but does not host or endorse, so read a skill's source on GitHub before you install it.

Skillspace indexes and rates skills; it does not host or endorse them. Run third-party skills at your own risk.