Benchmark and improve VLMs capabilities in complex, evolving markets as sequential traders - LLM post training, Reinforcement Learning Fine Tuning and Agentic AI