Verify machine learning code and hyperparameters against official documentation.
npx skills add https://github.com/Leeroo-AI/superml --skill ml-verify
Please help me install this Agent Skill. Skill: ml-verify Source: https://github.com/Leeroo-AI/superml/tree/main/skills/ml-verify Command: npx skills add https://github.com/Leeroo-AI/superml --skill ml-verify
This Skill prevents costly mistakes by verifying ML code, configurations, and mathematical formulas against official documentation and best practices before expensive training jobs or deployments.
Use the ml-verify skill to check my QLoRA config for training Llama-3 8B.
High-intent search queries and answers about installing and using this skill.
You can verify machine learning code by checking configurations, API usage, and mathematical formulas against official documentation and established heuristics before initiating expensive training jobs. This pre-run verification catches configuration errors and incorrect API usage.
Debugging common ML errors like OOM and NaN involves validating training plans and checking mathematical formulas against known-good examples. This process identifies configuration issues and mathematical inaccuracies that cause out-of-memory or not-a-number errors.
Yes, you can validate Hugging Face and DeepSpeed configurations by verifying training plans against official framework documentation. This ensures your model loading, data batching, and learning rate setups are correctly configured before multi-day fine-tuning jobs.
To check a QLoRA config, verify the learning rate, data loading pipeline, and LoRA parameters against Hugging Face PEFT and Transformers documentation. This documentation grounding ensures your setup follows established best practices.
Pre-training verification requires grounding via knowledge base tools or web fetching of API documentation. This ensures that code checks and configuration validations cite official framework documentation and known-good examples accurately.
A dry run checklist for ML pre-training checks should guide you through model loading, data batching, and initial loss or gradient sanity checks. This ensures mathematical formulas and hyperparameter configurations are validated before execution.
shinnytechGenerate reStructuredText docstrings for PyTorch functions and methods.
BagelHoleSet up QLoRA, LoRA, and full fine-tuning infrastructure with Hugging Face TRL and Axolotl.
fusengineScan the Claude Code changelog and generate a structured update report.
scooter-lacroixCreate and configure specialized AI sub-agents with custom prompts, tools, and models.
