Skill UI

Weights & Biases (W&B) automates experiment logging, real-time dashboards, hyperparameter sweeps, artifact tracking, and model registry/versioning for ML teams using PyTorch, TensorFlow, or HuggingFace workflows.

Hardware-agnostic quantum ML framework enabling gradient-based training of quantum circuits, hybrid quantum-classical models, and seamless integration with PyTorch/JAX/TensorFlow alongside device portability across IBM/Google/Rigetti/IonQ.

PyTorch Lightning Toolkit streamlines PyTorch model training by organizing LightningModules, LightningDataModules, Trainer configs, callbacks, logging, and distributed strategies for scalable multi-GPU/TPU runs in research or production workflows.

Probabilistic deep-learning framework built on PyTorch for single-cell omics; handles batch correction, integration, multimodal and spatial data, differential expression, and transfer learning with unified APIs.

Stable Baselines3 is a robust, PyTorch-based library providing production-ready implementations of core reinforcement learning algorithms (PPO, SAC, DQN, etc.). It facilitates the training of single-agent RL models, supports the creation of custom Gymnasium environments, and handles advanced parallel training via vectorized environments. It is ideal for rapid prototyping and rigorous RL experiments.

Reference for building Graph Neural Networks with PyTorch Geometric, covering data objects, datasets, transforms, convolution layers, lazy initialization, and heterogeneous support so you can quickly tackle node, link, and graph tasks.

TorchDrug is a PyTorch-native toolbox for building graph neural networks tailored to molecules, proteins, and biomedical knowledge graphs, covering property prediction, generation, retrosynthesis, and knowledge graph reasoning with curated datasets and models.

A comprehensive technical module covering foundational deep learning principles and cutting-edge computer vision architectures, drawing from Yann LeCun's research. Topics include CNN fundamentals, backpropagation theory, and advanced self-supervised learning (SSL) models such as SimCLR, MAE, and the modern Joint Embedding Predictive Architecture (JEPA). Provides detailed theoretical explanations and complete PyTorch code for implementation.

Guides on using SwanLab for open-source ML experiment tracking, offering local or self-hosted dashboards, rich media logging, and integrations with PyTorch, Transformers, Lightning, or Fastai for comparing runs and monitoring metrics.

Provides end-to-end guidance for fine-tuning and serving Physical Intelligence’s OpenPI pi0 family (pi0, pi0-fast, pi0.5) with both JAX and PyTorch backends, covering dataset prep, norm stats, checkpoint conversion, and low-latency WebSocket policy inference for ALOHA, DROID, and LIBERO tasks.

This skill provides guidance on implementing common architectural patterns and best practices within the PyTorch framework. It is designed for developers building sophisticated deep learning models, helping to improve structure, efficiency, and code maintainability when working with neural networks.

Collection of battle-tested PyTorch training recipes covering architecture selection, optimizers, LR scheduling, precision tips, and debugging to train or fine-tune LLMs, vision, diffusion, and biomedical models with high GPU throughput.