Download

Skill UI

Browse and discover 5096+ curated skills

All Development Artificial Intelligence Design & Creative Product & Business Data Science Marketing Soft Skills Productivity Engineering Languages

Search Pytorch , found 29 results

Default Newest Most Downloaded

Transfer Learning Adaptation Skill

adapting-transfer-learning-models

jeremylongshore/claude-code-plugins-plus-skills

Automates adapting pretrained models via transfer learning by analyzing requirements, generating PyTorch or TensorFlow fine-tuning code, validating data, tracking KPIs, and saving artifacts so you can quickly reuse existing models for new datasets and tasks.

TorchTitan LLM Pretraining

distributed-llm-pretraining-torchtitan

Orchestra-Research/AI-Research-SKILLs

TorchTitan delivers PyTorch-native distributed LLM pretraining with composable 4D parallelism (FSDP2, TP, PP, CP), Float8, torch.compile, and checkpointing so you can scale Llama 3.1, DeepSeek V3, or custom models across 8 to 512+ GPUs.

Early Stopping Callback

early-stopping-callback

jeremylongshore/claude-code-plugins-plus-skills

Automated guidance for configuring and applying early stopping callbacks across ML training workflows, covering best practices, code generation, and validation for PyTorch, TensorFlow, and scikit-learn experiments.

Gradient Clipping Helper

gradient-clipping-helper

jeremylongshore/claude-code-plugins-plus-skills

Gradient Clipping Helper automates ML training work by guiding gradient clipping workflows, generating production-ready code and configs, and validating outputs for PyTorch, TensorFlow, and related pipelines.

Half Quadratic Quantization

hqq-quantization

Orchestra-Research/AI-Research-SKILLs

Calibrate-free half-quadratic quantization for LLMs that delivers 1-8 bit model compression with configurable backends (PyTorch, TorchAO, BitBlas, Marlin). Ideal for fast quantization workflows, zero-data deployment, and PEFT/LoRA fine-tuning with HuggingFace or vLLM stacks.

Accelerate Distributed Training

huggingface-accelerate

Orchestra-Research/AI-Research-SKILLs

Accelerate wraps PyTorch scripts with a four-line integration so you can launch DDP, DeepSpeed, FSDP, Megatron, or mixed-precision training on any hardware via a single command, handling device placement, sharding, and automatic config generation.

Learning Rate Scheduler

learning-rate-scheduler

jeremylongshore/claude-code-plugins-plus-skills

Provides automated guidance for configuring learning rate schedulers during ML training, covering best practices, setup patterns, and framework-ready code snippets for PyTorch, TensorFlow, and Scikit-learn while validating against common standards.

Production ML Engineering Playbook

sickn33/antigravity-awesome-skills

Guides ML engineers in building production-ready systems with PyTorch 2.x, TensorFlow, and modern tooling, covering serving pipelines, feature engineering, monitoring, and MLOps infrastructure for scalable, reliable deployments.

MLflow Lifecycle Tracker

Orchestra-Research/AI-Research-SKILLs

MLflow provides a framework-agnostic ML lifecycle toolkit for tracking experiments, logging models/artifacts, managing registries with versions/stages, and deploying across platforms so teams can reproduce and collaborate on PyTorch, TensorFlow, and Scikit-Learn workflows.

Minimalist GPT Training

Orchestra-Research/AI-Research-SKILLs

NanoGPT is a hackable ~300-line PyTorch reproduction of GPT-2 for learning transformer internals; train Shakespeare or OpenWebText on CPUs or multi-GPU setups, experiment with fine-tuning/custom datasets, and prototype quickly.

Nnsight Remote Interpretability

nnsight-remote-interpretability

Orchestra-Research/AI-Research-SKILLs

Guides researchers through nnsight workflows for probing PyTorch models locally or via NDIF remote nodes, covering activation analysis, patching, and tracing tools for massive GPT/Llama families.

Flash Attention Optimization

optimizing-attention-flash

Orchestra-Research/AI-Research-SKILLs

Speeds up transformer attention with Flash Attention techniques, offering 2-4× throughput gains and 10-20× memory reduction. Ideal for long-context models on PyTorch (native or flash-attn lib), H100 FP8, and sliding window scenarios to resolve GPU memory bottlenecks and improve inference.

Language