Tags: #optimization

Heretic is an AI model utility that automatically removes censorship and safety alignment from transformer-based language models without requiring expensive post-training.

LMCache is an LLM serving engine extension designed to significantly reduce Time-To-First-Token (TTFT) and boost throughput by intelligently reusing KV caches across various storage tiers and serving instances.

A comprehensive, curated list of research papers and associated code implementations focused on optimizing Large Language Model (LLM) and Vision-Language Model (VLM) inference.

A versatile framework designed to train and optimize AI agents from any framework with minimal code changes, leveraging advanced algorithms like Reinforcement Learning.

An open-source framework that automates the evaluation and optimization of Retrieval-Augmented Generation (RAG) pipelines using AutoML-style automation for specific datasets.