5 papers
MaxProof: Scaling Mathematical Proof with Generative-Verifier RL and Population-Level Test-Time Scaling
Jiacheng Chen, Xinyu Zhang, Shunkai Zhang +20
We present MaxProof, a population-level test-time scaling framework for competition-level mathematical proof in the MiniMax-M3 series. M3 first trains three proof-oriented capabili…
ClaimDiff-RL: Fine-Grained Caption Reinforcement Learning through Visual Claim Comparison
Tianle Li, Xuyang Shen, Yan Ma +7
Long-form image captioning exposes a reward granularity problem in RL: captions are judged as whole sequences, while the important errors occur at the level of individual visual cl…
Are Unified Vision-Language Models Necessary: Generalization Across Understanding and Generation
Jihai Zhang, Tianle Li, Linjie Li +2
Recent advancements in unified vision-language models (VLMs), which integrate both visual understanding and generation capabilities, have attracted significant attention. The under…
Unveiling the Compositional Ability Gap in Vision-Language Reasoning Model
Tianle Li, Jihai Zhang, Yongming Rao +1
While large language models (LLMs) demonstrate strong reasoning capabilities utilizing reinforcement learning (RL) with verifiable reward, whether large vision-language models (VLM…
BREEN: Bridge Data-Efficient Encoder-Free Multimodal Learning with Learnable Queries
Tianle Li, Yongming Rao, Winston Hu +1
Encoder-free multimodal large language models(MLLMs) eliminate the need for a well-trained vision encoder by directly processing image tokens before the language model. While this…