RLlama is an open-source framework that enhances LLMs with memory-augmented reinforcement learning. Built on LlamaGym, it introduces episodic and working memory systems, allowing language models to learn from past experiences and maintain context across interactions.
RLlama is an open-source framework that enhances LLMs with memory-augmented reinforcement learning. Built on LlamaGym, it introduces episodic and working memory systems, allowing language models to learn from past experiences and maintain context across interactions.