Xnhyacinth/Resadapt-Mix
Resadapt-Mix: RL dataset on Hugging Face. Resadapt-Mix is the mixed multimodal training set used by the ResAdapt/VisionThink training launchers in both visionthink and visionthink-feature-new-feature. The default training and validation parquet files are: data/train lens.parquet…