1 articles with this tag
Together AI's Aurora framework uses RL to continuously adapt speculative decoding for faster LLM inference, outperforming static models.
