Yash Jain

The paper demonstrates that supervised fine-tuning significantly outperforms frontier zero-shot large language models for screen-conditioned action prediction on the PiSAR benchmark, highlighting the importance of model-specific tuning recipes.