Skill UI

This skill automates the setup and configuration of Ollama, allowing users to deploy Large Language Models (LLMs) locally. It performs system assessment (OS, GPU, RAM), selects optimal models based on hardware availability, and handles installation across macOS, Linux, and Docker. Use this when you require free, self-hosted AI alternatives, need offline inference capabilities, or wish to eliminate reliance on costly hosted API services.

High-performance RLHF framework that combines Ray, vLLM, and ZeRO-3 to accelerate PPO/GRPO/RLOO/DPO training for 7B-70B+ models on distributed GPU clusters, providing a single-stack workflow for reward models and policy optimization.

Flash Attention optimizer delivers 2-4x speedup and 10-20x memory savings for transformer attention, ideal when training or running long-context models that hit GPU memory or latency limits; supports PyTorch native scaled_dot_product_attention, the flash-attn library, H100 FP8 precision, and sliding-window attention.

Covers HuggingFace PEFT workflows for LoRA/QLoRA-based LLM fine-tuning, enabling sub-1% parameter updates on limited GPUs plus adapter serving tips.

Ray Data offers scalable distributed data processing for ML workloads, streaming across CPU/GPU clusters and integrating with PyTorch, TensorFlow, and Ray Train. Use it for batch inference, ETL, and multi-modal preprocessing from a laptop up to hundreds of nodes.

Ray Train orchestrates distributed PyTorch, TensorFlow, and HuggingFace workloads from laptops to clusters, automating GPU placement, fault tolerance, checkpointing, and hyperparameter sweeps so teams can scale massive model training with minimal code changes.

Serve production-grade LLM APIs with vLLM, leveraging PagedAttention, continuous batching, quantization, and tensor parallelism to maximize throughput while keeping latency, GPU usage, and scalability under control for OpenAI-compatible deployments.

SGLang is a high-performance serving framework for LLMs and VLMs that uses RadixAttention prefix caching to deliver structured outputs (JSON/regex/grammar) and agentic workflows with function calls, providing up to 5× faster inference than vLLM across multi-GPU production deployments.

A comprehensive guide for writing efficient GLSL shaders (Vertex/Fragment) used in WebGL and game engines. Learn fundamental concepts like shader structure, uniforms, varying variables, and advanced techniques such as raymarching and post-processing effects, optimizing graphics rendering performance on the GPU.

SkyPilot orchestrates ML training, batch jobs, and serving across 20+ clouds with automatic cost-aware region selection, spot-instance recovery, managed clusters, and GPU-aware distributed training, letting you optimize spend and avoid vendor lock-in.

Optimizes large language model inference on NVIDIA GPUs with TensorRT, delivering 10-100× faster throughput, quantized precision (FP8/INT4), multi-GPU scaling, and serving-ready tooling for production deployments.

torchforge is Meta's advanced PyTorch-native library designed for agentic Reinforcement Learning (RL). It fundamentally separates core RL algorithms from complex distributed infrastructure concerns. This allows researchers to rapidly prototype and experiment with novel RL methods (like GRPO) using clean, PyTorch-native abstractions, while automatically handling large-scale tasks like distributed training across thousands of GPUs using Monarch and TorchTitan.