Skill UI

Speeds up LLM inference by combining speculative decoding with Medusa multiple heads and lookahead techniques so chatbots, code assistants, and other real-time workloads can run with 1.5-3.6× lower latency while preserving target-model quality.

A comprehensive guide to leveraging Stable Diffusion models via the HuggingFace Diffusers library. This skill facilitates state-of-the-art image creation, including Text-to-Image generation from prompts, Image-to-Image translation, inpainting, outpainting, and advanced workflow building. Supports various models like SDXL and SD 3.0, ideal for professional creative and computational vision tasks.

Automates guidance for streaming inference setup, covering best practices in ML deployment, serving, monitoring, and production-ready configurations whenever you ask about streaming inference patterns or implementation.

Provides a structured workflow for defining agent-agnostic subagents with isolated context, covering purpose definition, metadata configuration, prompt writing, and common specialist patterns for verification, debugging, security auditing, and review.

Guides Rube MCP users through Supportivekoala automation via Composio, emphasizing tool discovery, connection checks, schema compliance, and multi-step execution for reliable workflow orchestration.

Automate Synthflow AI workflows through Composio by invoking Rube MCP operations, always discovering current tool schemas before managing connections or executing tool runs.

Automates TensorFlow SavedModel deployment support with step-by-step guidance, best-practice patterns, production-ready code/config generation, and validation for serving, monitoring, and optimization scenarios.

Automates Tinyurl operations through Composio’s Rube MCP by guiding you to discover tool schemas, manage Tinyurl connections, and execute actions via RUBE_MULTI_EXECUTE_TOOL with session and memory handling.

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.

Assists with torchscript exporter tasks for ML deployment, offering automated guidance, code/config generation, and validation against industry standards whenever torchscript exporters are mentioned.

Automates generating TorchServe deployment configurations with guided best practices for model serving, MLOps pipelines, inference, and production monitoring so ML deployment workflows stay consistent and reliable.

Automates Triton inference config guidance for ML deployment, covering serving, MLOps pipelines, monitoring, and production-ready configurations aligned with best practices.