skills.rest
TTAWDTT
TTAWDTTCOMMUNITY
Explore all skills by this author →
GitHub

ml-training-recipes

Optimizes PyTorch training workflows with architecture selection and debugging.

5|Updated Jun 7, 2026
Search Tags:#debugging#machine learning#PyTorch#training optimization#neural networks#mixed precision#experiment management#learning rate scheduling#optimizer configuration
One-click install
npx skills add https://github.com/TTAWDTT/elegant-researcher-skill --skill ml-training-recipes-ttawdtt
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ml-training-recipes
Source: https://github.com/TTAWDTT/elegant-researcher-skill/tree/main/skills/ml-training-recipes
Command: npx skills add https://github.com/TTAWDTT/elegant-researcher-skill --skill ml-training-recipes-ttawdtt

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, torchvision, torchvision.transforms, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill provides a comprehensive set of guidelines and recipes for optimizing PyTorch training processes, covering a wide range of topics from architecture selection to debugging.

Core Features & Use Cases

  • Training Loop Best Practices: Offers best practices for efficient training loops, including gradient accumulation, memory management, and performance monitoring.
  • Optimizer Configuration: Includes detailed instructions for configuring modern optimizers like AdamW and Muon for various parameter types.
  • Learning Rate Scheduling: Provides guidelines for choosing and implementing learning rate schedules for optimal training dynamics.
  • Mixed Precision & Compilation: Discusses the use of mixed precision training and the benefits of model compilation for performance.
  • Memory & Performance: Offers strategies for managing memory and optimizing performance, including model FLOPs estimation and OOM solutions.
  • Hyperparameter Search: Details a systematic approach to hyperparameter tuning, including a priority order and best practices.
  • Debugging Checklist: Provides a comprehensive checklist for troubleshooting common issues in PyTorch training.
  • Experiment Management: Offers a framework for tracking and managing experiments using TSV files and best practices for systematic experimentation.
  • Evaluation Metrics: Defines key evaluation metrics for different types of machine learning tasks.
  • Biomedical & Pharmaceutical ML: Provides insights into using machine learning in the biomedical and pharmaceutical domains, including molecular property prediction, drug discovery, and medical imaging.
  • Genomic & Sequence Models: Offers guidance on using machine learning for genomic and sequence analysis.
  • Single-Cell Omics: Discusses the use of machine learning in single-cell omics analysis.
  • Clinical NLP: Provides insights into using machine learning for clinical NLP tasks like named entity recognition and relation extraction.
  • EHR & Survival Analysis: Offers guidance on using machine learning for analyzing electronic health records and survival analysis.
  • Biomedical Training Tricks: Provides strategies for training machine learning models on small datasets and handling class imbalance.

Quick Start

To get started with the ml-training-recipes Skill, use the following command:

npx skills add TTAWDTT/elegant-researcher-skill -g -y --copy

Frequently Asked Questions about ml-training-recipes

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I configure optimizers like AdamW and Muon for PyTorch neural network training?▼

To configure optimizers for PyTorch training, you can apply detailed parameter-specific instructions for modern optimizers like AdamW and Muon to stabilize gradient updates and improve convergence dynamics across various neural network architectures.

What is the best way to implement learning rate scheduling and mixed precision in PyTorch?▼

The best way to implement learning rate scheduling and mixed precision in PyTorch involves applying systematic guidelines for choosing schedules and enabling mixed precision training to achieve optimal training dynamics and accelerate model performance.

How do I fix out of memory errors during PyTorch model training?▼

To fix out of memory errors during PyTorch model training, apply targeted memory management strategies such as gradient accumulation, model FLOPs estimation, and specific OOM solutions to optimize memory usage within the training loop.

What is the recommended priority order for systematic hyperparameter tuning in machine learning?▼

Systematic hyperparameter tuning in machine learning follows a specific priority order and best practices, providing a structured framework to efficiently search and identify optimal configurations for neural network training tasks.

Why does my PyTorch training loop fail to converge and how can I debug it?▼

If your PyTorch training loop fails to converge, use a comprehensive debugging checklist to troubleshoot common issues, verify gradient accumulation, and monitor performance metrics to identify and resolve training anomalies.

Can I use these PyTorch training recipes for biomedical and pharmaceutical machine learning tasks?▼

Yes, you can use these PyTorch training recipes for biomedical and pharmaceutical machine learning tasks, utilizing specialized insights for molecular property prediction, drug discovery, medical imaging, and handling class imbalance on small datasets.

Related Skills

Explore More Software Engineering Skills→
robotlearning123robotlearning123

torch-geometric

Develop and train graph neural networks with PyTorch.

Community
Advanced
zechenzhangAGIzechenzhangAGI
11.5k

moe-training

Train Mixture-of-Experts models with DeepSpeed MoE and HuggingFace Transformers.

Community
Advanced
ldzhhxxldzhhxx
1

pytorch-fsdp

Manage fully sharded data parallelism in PyTorch with FSDP.

Community
Advanced
jstzwjjstzwj
4

bitsandbytes

Quantize weights and optimizer states for PyTorch model training.

Community
Advanced