Pith. sign in

REVIEW 13 cited by

Boosting the Generalization and Reasoning of Vision Language Models with Curriculum Reinforcement Learning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2503.07065 v1 pith:Z7NUCBYK submitted 2025-03-10 cs.CV

classification cs.CV
keywords modelscurr-reftlearningreinforcementvlmscapabilitiescurriculumlanguage
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

While state-of-the-art vision-language models (VLMs) have demonstrated remarkable capabilities in complex visual-text tasks, their success heavily relies on massive model scaling, limiting their practical deployment. Small-scale VLMs offer a more practical alternative but face significant challenges when trained with traditional supervised fine-tuning (SFT), particularly in two aspects: out-of-domain (OOD) generalization and reasoning abilities, which significantly lags behind the contemporary Large language models (LLMs). To address these challenges, we propose Curriculum Reinforcement Finetuning (Curr-ReFT), a novel post-training paradigm specifically designed for small-scale VLMs. Inspired by the success of reinforcement learning in LLMs, Curr-ReFT comprises two sequential stages: (1) Curriculum Reinforcement Learning, which ensures steady progression of model capabilities through difficulty-aware reward design, transitioning from basic visual perception to complex reasoning tasks; and (2) Rejected Sampling-based Self-improvement, which maintains the fundamental capabilities of VLMs through selective learning from high-quality multimodal and language examples. Extensive experiments demonstrate that models trained with Curr-ReFT paradigm achieve state-of-the-art performance across various visual tasks in both in-domain and out-of-domain settings. Moreover, our Curr-ReFT enhanced 3B model matches the performance of 32B-parameter models, demonstrating that efficient training paradigms can effectively bridge the gap between small and large models.

Discussion (0). Sign in to comment.

Forward citations

Cited by 13 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. S-GRPO: Unified Post-Training for Large Vision-Language Models

    cs.LG 2026-04 unverdicted novelty 7.0 of 10

    S-GRPO unifies SFT and RL for LVLMs via conditional ground-truth injection that supplies a maximal-reward anchor when group exploration fails completely.

  2. SyncLoop: A Multimodal Dual-Loop Framework for Self-Improving Mathematical Reasoning

    cs.CV 2025-07 conditional novelty 6.0 of 10

    SyncLoop jointly evolves multimodal training data and model capability through alternating SFT and RL, selecting error-prone samples to improve geometry reasoning.

  3. MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval

    cs.AI 2025-06 conditional novelty 6.0 of 10

    MM-R5, a 7B multimodal re-ranker trained with SFT and GRPO, achieves state-of-the-art page-level recall on MMDocIR by generating per-page reasoning chains.

  4. ZeroGUI: Automating Online GUI Learning at Zero Human Cost

    cs.AI 2025-05 conditional novelty 6.0 of 10

    ZeroGUI uses VLM-generated tasks and VLM-estimated rewards with two-stage GRPO to improve GUI agent success rates on OSWorld and AndroidLab without human annotations.

  5. Align and Surpass Human Camouflaged Perception: Visual Refocus Reinforcement Fine-Tuning

    cs.CV 2025-05 conditional novelty 6.0 of 10

    A curriculum reinforcement-learning framework with in-context refocus examples improves camouflaged object classification and detection for a vision-language model, and the authors report surpassing human performance ...

  6. Rethinking Reward Signals in Video GRPO: When Scores Become Targets

    cs.CV 2025-11 reject novelty 5.0 of 10

    A self-paced GRPO variant that adaptively reweights visual, temporal, and text-alignment reward components as the generator improves reports small VBench gains over static-reward baselines.

  7. ChartReasoner: Code-Driven Modality Bridging for Long-Chain Reasoning in Chart Question Answering

    cs.CL 2025-06 conditional novelty 5.0 of 10

    ChartReasoner converts charts into executable ECharts code, distills long-chain reasoning traces from that code, and trains a 7B multimodal model with SFT and GRPO to improve chart question answering.

  8. WeThink: Toward General-purpose Vision-Language Reasoning via Reinforcement Learning

    cs.CV 2025-06 conditional novelty 5.0 of 10

    WeThink, a 120K-image QA dataset with AI-generated reasoning paths, combined with hybrid-reward reinforcement learning, improves a 7B vision-language model across 14 benchmarks.

  9. UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning

    cs.CV 2025-05 conditional novelty 5.0 of 10

    UniVG-R1 uses CoT supervised fine-tuning plus GRPO with difficulty-aware reweighting to make Qwen2-VL substantially better at multi-image, reasoning-based visual grounding.

  10. Reinforcement Learning Meets Large Language Models: A Survey of Advancements and Applications Across the LLM Lifecycle

    cs.CL 2025-09 conditional novelty 3.0 of 10

    A survey that maps reinforcement learning methods, datasets, benchmarks, and open-source tools across the full training lifecycle of large language models, focusing on verifiable-reward reasoning.

  11. Empowering Multimodal LLMs with External Tools: A Comprehensive Survey

    cs.CV 2025-08 unverdicted novelty 2.0 of 10

    A survey paper maps how external tools are used to augment multimodal large language models across data, tasks, evaluation, and future directions.

  12. Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models

    cs.CL 2025-05 conditional novelty 2.0 of 10

    A survey-style position paper claims that reinforcement fine-tuning powers reasoning in multimodal LLMs, summarizing over a hundred recent works and proposing five future research directions.

  13. Aligning Large Vision-Language Models by Deep Reinforcement Learning and Direct Preference Optimization

    cs.LG 2025-09 unverdicted

    A survey categorizing deep reinforcement learning and direct preference optimization methods for aligning large vision-language models, with tables of studies and datasets and no new experimental result.

Pith tools