Puzzle Curriculum GRPO for Vision-Centric Reasoning

• URL: http://arxiv.org/abs/2512.14944v1
• Date: Tue, 16 Dec 2025 22:17:25 GMT
• Title: Puzzle Curriculum GRPO for Vision-Centric Reasoning
• Authors: Ahmadreza Jeddi, Hakki Can Karaimer, Hue Nguyen, Zhongling Wang, Ke Zhao, Javad Rajabi, Ran Zhang, Raghav Goyal, Babak Taati, Radek Grzeszczuk,
• Abstract summary: We present Puzzle Curriculum GRPO (PC- GRPO), a supervision-free recipe for reinforcement learning (RL) with Verifiable Rewards (RLVR)<n>PC- GRPO replaces labels with three self-supervised puzzle environments: PatchFit, Rotation (with binary rewards) and Jigsaw (with graded partial credit mitigating reward sparsity)<n>To counter flat rewards and vanishing group-relative advantages, we introduce a difficulty-aware curriculum that dynamically weights samples and peaks at medium difficulty.
• Score: 7.117453620224407
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent reinforcement learning (RL) approaches like outcome-supervised GRPO have advanced chain-of-thought reasoning in Vision Language Models (VLMs), yet key issues linger: (i) reliance on costly and noisy hand-curated annotations or external verifiers; (ii) flat and sparse reward schemes in GRPO; and (iii) logical inconsistency between a chain's reasoning and its final answer. We present Puzzle Curriculum GRPO (PC-GRPO), a supervision-free recipe for RL with Verifiable Rewards (RLVR) that strengthens visual reasoning in VLMs without annotations or external verifiers. PC-GRPO replaces labels with three self-supervised puzzle environments: PatchFit, Rotation (with binary rewards) and Jigsaw (with graded partial credit mitigating reward sparsity). To counter flat rewards and vanishing group-relative advantages, we introduce a difficulty-aware curriculum that dynamically weights samples and peaks at medium difficulty. We further monitor Reasoning-Answer Consistency (RAC) during post-training: mirroring reports for vanilla GRPO in LLMs, RAC typically rises early then degrades; our curriculum delays this decline, and consistency-enforcing reward schemes further boost RAC. RAC correlates with downstream accuracy. Across diverse benchmarks and on Qwen-7B and Qwen-3B backbones, PC-GRPO improves reasoning quality, training stability, and end-task accuracy, offering a practical path to scalable, verifiable, and interpretable RL post-training for VLMs.

Related papers

• From Verifiable Dot to Reward Chain: Harnessing Verifiable Reference-based Rewards for Reinforcement Learning of Open-ended Generation [52.62655622099456]
  We propose reinforcement learning with verifiable reference-based rewards (RLVRR)<n>Instead of checking the final answer, RLVRR extracts an ordered linguistic signal from high-quality references (i.e., reward chain)<n>In this way, RLVRR decomposes rewards into two dimensions: content, which preserves deterministic core concepts, and style, which evaluates adherence to stylistic properties.
  arXiv  Detail & Related papers  (2026-01-26T14:39:58Z)
• Empowering Multi-Turn Tool-Integrated Reasoning with Group Turn Policy Optimization [20.004150645050537]
  Group Turn Policy Optimization (GTPO) is a novel reinforcement learning algorithm designed for training Large Language Models (LLMs) on multi-turn Tool-Integrated Reasoning tasks.<n>GTPO introduces three key innovations: turn-level reward assignment that provides fine-grained feedback for individual turns, return-based advantage estimation, and self-supervised reward shaping.<n>Our comprehensive evaluation demonstrates that GTPO outperforms GRPO by 3.0% on average across diverse reasoning benchmarks.
  arXiv  Detail & Related papers  (2025-11-18T19:01:16Z)
• GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning [53.894789613838654]
  We introduce SEED-Bench-R1, a benchmark with complex real-world videos requiring balanced perception and reasoning.<n>Using SEED-Bench-R1, we find that standard GRPO, while improving answer accuracy, often reduces logical coherence between reasoning steps and answers, with only a 57.9% consistency rate.<n>We propose GRPO-CARE, a consistency-aware RL framework optimizing both answer correctness and reasoning coherence without explicit supervision.
  arXiv  Detail & Related papers  (2025-06-19T08:49:13Z)
• Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning [87.7836502955847]
  We propose a novel self-rewarding reinforcement learning framework to enhance Large Language Model (LLM) reasoning.<n>Our key insight is that correct responses often exhibit consistent trajectory patterns in terms of model likelihood.<n>We introduce CoVo, an intrinsic reward mechanism that integrates Consistency and Volatility via a robust vector-space aggregation strategy.
  arXiv  Detail & Related papers  (2025-06-10T12:40:39Z)
• A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce [68.99924691391048]
  We revisit GRPO from a reinforce-like algorithm perspective and analyze its core components.<n>We find that a simple rejection sampling baseline, RAFT, yields competitive performance than GRPO and PPO.<n>Motivated by this insight, we propose Reinforce-Rej, a minimal extension of policy gradient that filters both entirely incorrect and entirely correct samples.
  arXiv  Detail & Related papers  (2025-04-15T16:15:02Z)
• Rethinking RL Scaling for Vision Language Models: A Transparent, From-Scratch Framework and Comprehensive Evaluation Scheme [36.34443944082215]
  This work introduces a transparent, from-scratch framework forReinforcement learning (RL) in vision-based models (VLMs)<n>It offers a minimal yet functional four-step pipeline validated across multiple models and datasets.<n>In addition, a standardized evaluation scheme is proposed to assess training dynamics and reflective behaviors.
  arXiv  Detail & Related papers  (2025-04-03T13:53:28Z)
• Exploring the Effect of Reinforcement Learning on Video Understanding: Insights from SEED-Bench-R1 [53.894789613838654]
  We introduce SEED-Bench-R1, a benchmark designed to evaluate post-training methods for MLLMs in video understanding.<n>It includes intricate real-world videos and complex everyday planning tasks in the format of multiple-choice questions.<n>Using Qwen2-VL-Instruct-7B as a base model, we compare RL with supervised fine-tuning (SFT)<n>Our detailed analysis reveals that RL enhances visual perception but often produces less coherent reasoning chains.
  arXiv  Detail & Related papers  (2025-03-31T17:55:23Z)
• GTR: Guided Thought Reinforcement Prevents Thought Collapse in RL-based VLM Agent Training [62.536191233049614]
  Reinforcement learning with verifiable outcome rewards (RLVR) has effectively scaled up chain-of-thought (CoT) reasoning in large language models (LLMs)<n>This work investigates this problem through extensive experiments on complex card games, such as 24 points, and embodied tasks from ALFWorld.<n>We find that when rewards are based solely on action outcomes, RL fails to incentivize CoT reasoning in VLMs, instead leading to a phenomenon we termed thought collapse.
  arXiv  Detail & Related papers  (2025-03-11T15:17:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.