headroomlabs-ai/headroom
Headroom is a context compression layer for AI agents, reducing token usage by 60-95% across tool outputs, logs, RAG chunks, files, and conversation history, while maintaining accuracy.
Awesome AI Agents › End-to-End Optimizers
microsoft/Trace
Trace is an innovative AutoDiff-like Python library designed for end-to-end generative optimization of AI agents using general feedback mechanisms such as numerical rewards, losses, natural language text, and compiler errors. It extends the traditional back-propagation algorithm by capturing and propagating an AI system's execution trace, enabling optimization beyond neural networks. Implemented with a PyTorch-like interface, Trace allows users to write Python code directly and optimize specific parts of their programs using Trace primitives like `node` and `bundle`. The `node` primitive defines nodes in the computation graph, while `bundle` wraps functions to make them trainable, facilitating the construction of complex, optimizable AI systems. Trace supports training AI agents by defining trainable nodes and functions, then applying optimizers such as OptoPrime to improve performance based on feedback functions that act like reward or loss signals. This approach enables optimization of diverse AI behaviors, including language understanding and decision-making, through a flexible and extensible framework. The library integrates with LiteLLM as the default backend for large language models and offers compatibility with AutoGen for legacy support. The project is actively maintained by Microsoft and has been featured in notable AI research venues like NeurIPS 2024. It includes comprehensive documentation, example code snippets, and community support via GitHub Issues and Discord. Trace's unique capability to optimize AI systems end-to-end with general feedback makes it a powerful tool for researchers and developers aiming to build self-adapting AI agents and optimize complex computational tasks, such as parallel programming code optimization. Overall, Trace represents a significant advancement in AI training methodologies by generalizing gradient-based optimization to broader contexts, enabling new possibilities in AI agent development and performance enhancement.
https://github.com/microsoft/Trace
Also in End-to-End Optimizers
topoteretes/cognee
Cognee is an open-source project that provides scalable and modular memory solutions for AI agents using ECL pipelines, enabling efficient data ingestion, knowledge graph generation, and query capabilities to enhance AI performance and reduce hallucinations.
letta-ai/letta
Letta is an open-source framework for building stateful AI agents with advanced reasoning, transparent long-term memory, and context management, supporting multiple LLM backends and providing a graphical development environment.
langchain-ai/opengpts
OpenGPTs is an open-source platform that replicates and extends OpenAI's GPTs and Assistants API, offering customizable cognitive architectures, extensive tooling, and support for multiple language models and vector databases.
aiwaves-cn/agents
Agents 2.0 is an open-source framework that enables the training and self-evolution of autonomous language agents through symbolic learning inspired by neural network methodologies.
getzep/zep
Zep is an AI memory foundation that uses a temporal Knowledge Graph to enable continuous learning and personalized experiences for AI agents by efficiently managing and retrieving contextual chat and data artifacts.
algorithmicsuperintelligence/optillm
OptiLLM is an optimizing inference proxy that significantly improves the accuracy and performance of large language models on reasoning tasks without requiring any model training, by applying over 20 advanced optimization techniques during inference.
mll-lab-nu/RAGEN
A reinforcement learning framework for training multi-turn reasoning LLM agents on the StarPO algorithm, shipping ten Gym-compatible environments plus diagnostics for echo-trap and reasoning-collapse failure modes.