Skill UI

Mamba is a selective state-space model architecture delivering O(n) inference for million-token sequences, offering hardware-aware Mamba-1 and Mamba-2 variants that beat Transformers on speed without KV cache overhead.

This skill provides methodologies for designing sophisticated, layered memory architectures for AI agents. It moves beyond simple context windows and basic vector stores by incorporating short-term, long-term, and graph-based memory systems. Implementing this is crucial for agents that require persistence across sessions, maintaining complex entity consistency, and performing advanced, relationship-driven reasoning over accumulated knowledge bases, including time-aware fact tracking.

This guide provides advanced techniques for optimizing Mistral AI API usage, focusing on latency reduction and throughput maximization. Learn to implement robust caching strategies, utilize streaming responses (SSE), select the optimal model for specific use cases (e.g., Mistral Small for chat, Codestral for code), manage concurrent requests using rate limiting, and efficiently use the Batch API. Essential for building high-performance, production-grade AI applications.

This skill provides comprehensive guidance for building production-grade MLOps pipelines. It covers the entire machine learning lifecycle, from initial data ingestion and rigorous feature engineering to model training, advanced validation (A/B testing, performance drift), and robust deployment strategies (Canary, Blue-green). It enables the design and orchestration of scalable, reproducible, and observable ML systems using tools like Airflow and Kubeflow.

nanoGPT is a simplified, minimalist implementation designed specifically for educational purposes. It reproduces the core architecture of GPT-2 (124M) using clean, hackable code, allowing users to understand the entire transformer pipeline from scratch. The framework supports full workflows, including data preparation on various datasets (e.g., Shakespeare, OpenWebText), model training, and text generation, making it ideal for students and researchers learning NLP and deep learning principles.

A comprehensive workflow designed for clinical professionals to search, retrieve, and synthesize medical evidence. It allows users to query clinical literature with specialized filters (e.g., specialty, evidence level), check potential drug interactions against patient profiles, and look up the latest specialty guidelines from major health organizations (like ACC/AHA, NICE). The output provides structured citations, confidence scores, and evidence grading to support informed clinical decision-making.

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.

Use this tool to query, filter, and compare over 400 available AI models from OpenRouter. It allows users to check pricing, context lengths, model capabilities (e.g., tool calling), and select the best model based on specific needs, ensuring optimal performance and cost efficiency in AI development.

Outlines guarantees valid JSON/XML/code outputs by applying CFG/FSM-constrained token filtering, Pydantic schemas, and grammar-level sampling while supporting local Transformers, llama.cpp, and vLLM backends for fast inference.

This workflow executes a single-query search using the Perplexity Sonar model, ensuring all returned answers are web-grounded and accompanied by inline citations. It is essential for building high-trust applications, fact-checking tools, or any system requiring verifiable, real-time information retrieval. The process includes API interaction, detailed response parsing (answer, citations, search results), and formatting the output into a usable, source-linked format.

A comprehensive toolkit designed to manage the complex data flow associated with Perplexity search workflows. This skill implements critical security measures, including PII sanitization in user queries, validates external citations for accuracy, optimizes performance via result caching with freshness policies, and manages conversation context to maintain coherence and reliability. Essential for building robust, compliance-aware AI applications.

This guide provides comprehensive instructions for deploying applications powered by the Perplexity Sonar API. It covers setup for major cloud platforms like Vercel, Google Cloud Run, and Docker. Users can deploy production-ready endpoints, manage required secrets, implement edge functions, and utilize caching mechanisms (e.g., Redis) to ensure high availability and performance in production environments.