srpo-multimodal-reflection-rl

Teach multimodal language models to reflect on their reasoning and improve answers through structured RL training.

Independent GitHub snapshot · Automated classification · Original documentation governs