This repository provides the ViLaSR-data introduced in the paper:Reinforcing Spatial Reasoning in Vision-Language Models with Interwoven Thinking and Visual Drawing.
VILASR-ColdStart-33k: Initial data generated from cold-start prompts (cold_start_path*.zip)
VILASR-RRS-8k: Data refined using Reflective Rejection Sampling (reflective_rejection_sampling_part*.zip)
VILASR-RL-40k: Reinforcement Learning enhanced data… See the full description on the dataset page:
https://huggingface.co/datasets/AntResearchNLP/ViLaSR-data.
