EffGen: Enabling Small Language Models as Capable Autonomous Agents

EffGen is an open-source framework designed to optimize small language models (SLMs) for autonomous agent applications, addressing limitations of large models like high token costs and privacy concerns. Key features include prompt optimization that reduces input size by up to 80%, intelligent task decomposition, complexity-based routing, and a unified memory system. Benchmark results indicate that EffGen outperforms existing frameworks such as LangChain and AutoGen in terms of success rates and execution speed, making it a valuable tool for practitioners seeking efficient local deployment of SLMs.