Paper summary
Paper summary: GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning
GLM-4.1V-Thinking and GLM-4.5V use a reasoning-centric training framework with Reinforcement Learning with Curriculum Sampling (RLCS); GLM-4.5V is reported to lead open models of its size on nearly all of 42 benchmarks.
- A strong pre-trained vision foundation "sets the upper bound"; RL with curriculum sampling unlocks it.
- The authors note that RL improved task completion but didn't always improve reasoning quality.