Live page · Day archive

Reframe: Observation-Aligned Multi-Teacher On-Policy Distillation for Agentic Visual Reasoning

Read the original at HF Daily Papers

Summary

Reframe is a post-training framework for agentic visual reasoning that aligns observation states using Ground-Drop-GRPO and Observation-Gated-MOPD.

Carried by: HF Daily Papers. First seen: .