Fundus-R1 is a fundus-reading MLLM trained exclusively on public data via RAG-generated reasoning traces and process-reward RLVR, outperforming its base model and a version trained without the traces.
arXiv preprint arXiv:2503.03987 (2025) A Prompt Degisn Details This section details the prompt design for training Mags-RL
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Mags-RL uses agentic RL and a super-resolution agent for two-round reasoning in MLLMs, claiming gains on VSR, TallyQA, and GQA with a curriculum needing only 40 samples.
citing papers explorer
-
Fundus-R1: Training a Fundus-Reading MLLM with Knowledge-Aware Reasoning on Public Data
Fundus-R1 is a fundus-reading MLLM trained exclusively on public data via RAG-generated reasoning traces and process-reward RLVR, outperforming its base model and a version trained without the traces.
-
Mags-RL: Wearing Multimodal LLMs a Magnifying Glass via Agentic Reinforcement Learning For Complex Scene Reasoning
Mags-RL uses agentic RL and a super-resolution agent for two-round reasoning in MLLMs, claiming gains on VSR, TallyQA, and GQA with a curriculum needing only 40 samples.