Pith. sign in

REVIEW 30 cited by

SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvement

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 2504.07934 v3 pith:SQYLXTV4 submitted 2025-04-10 cs.CV

SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvement

classification cs.CV
keywords reasoningmodelssotadifficultyqwen2samplesamplesself-improvement
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We introduce ThinkLite-VL, a family of visual reasoning models that achieve state-of-the-art (SoTA) performance using an order of magnitude fewer training samples, relying purely on reinforcement fine-tuning (RFT) self-improvement without any knowledge distillation. Our central insight is that sample difficulty critically influences RFT effectiveness: appropriately challenging examples can drive substantial reasoning improvements, even in low-data regimes. However, quantifying sample difficulty in a reliable and scalable manner remains non-trivial. To address this, we repurpose Monte Carlo Tree Search (MCTS) to measure sample difficulty via the number of reasoning iterations a vision-language model (VLM) requires to solve each instance. This MCTS-based selection procedure identifies samples that induce deeper reasoning while remaining solvable, allowing us to filter a high-quality subset from 70k open-source examples spanning math, natural image understanding, and chart comprehension. Using this approach, we select just 11k challenging samples for RFT on Qwen2.5-VL-7B-Instruct and 7.5k samples for Qwen2.5-VL-72B-Instruct. The resulting models, ThinkLite-VL-7B and ThinkLite-VL-72B, significantly outperform their respective base models across eight visual reasoning benchmarks. In particular, ThinkLite-VL-7B improves the average performance of Qwen2.5-VL-7B-Instruct by 7\% and surpasses all existing 7B-level models, as well as much larger models such as GPT-4o, O1 and Qwen2.5-VL-72B, achieving a new SoTA score of 75.1 on MathVista. ThinkLite-VL-72B further advances the SoTA frontier, achieving an accuracy of 79.7 on MathVista and an average benchmark improvement of 4.42 over the open-source SOTA. These results demonstrate that MCTS-guided difficulty filtering provides a scalable and effective path toward data-efficient self-improvement in multimodal reasoning.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 30 Pith papers

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

  1. No Place to Hide: Benchmarking Video Hallucination with Background-Controlled Pairs

    cs.CV 2026-06 unverdicted novelty 7.0

    Introduces VidPair-Halluc benchmark of 1K background-controlled adversarial video pairs and 11K QA pairs generated via PairFlow pipeline to evaluate hallucination in LVMs.

  2. Are Tools Always Beneficial? Learning to Invoke Tools Adaptively for Dual-Mode Multimodal LLM Reasoning

    cs.CL 2026-05 conditional novelty 7.0

    AutoTool uses reinforcement learning with dual-mode rewards to train multimodal LLMs to adaptively choose between tool-assisted and text-centric reasoning, yielding accuracy and efficiency gains on V* and POPE benchmarks.

  3. CGC: Compositional Grounded Contrast for Fine-Grained Multi-Image Understanding

    cs.CV 2026-04 unverdicted novelty 7.0

    CGC improves fine-grained multi-image understanding in MLLMs by constructing contrastive training instances from existing single-image annotations and adding a rule-based spatial reward, achieving SOTA on MIG-Bench an...

  4. Understanding the Role of Hallucination in Reinforcement Post-Training of Multimodal Reasoning Models

    cs.LG 2026-04 unverdicted novelty 7.0

    RL post-training on hallucination-forced multimodal data improves reasoning performance and can outperform standard training.

  5. Learning Self-Correction in Vision-Language Models via Rollout Augmentation

    cs.CV 2026-02 conditional novelty 7.0

    Recombining pre/post-correction segments of existing rollouts yields dense self-correction training signals, giving an 8B VLM a 1.0-point average accuracy gain over GSPO on 7 benchmarks at 0.72x rollout time.

  6. Trace: A Taxonomy-Guided Environment for Multidomain Visual Reasoning

    cs.CV 2026-07 conditional novelty 6.0

    RLVR training on 64,000 procedurally generated Trace instances improves Qwen2.5-VL macro-average on 24 external visual reasoning benchmarks by 3.51 points at 3B and 4.06 points at 7B.

  7. DyCo-RL: Dynamic Cross-Modal Coordination for Visual Reasoning

    cs.CV 2026-06 unverdicted novelty 6.0

    DyCo-RL improves four RLVR algorithms on seven visual and math reasoning benchmarks by assigning tokens visual or text roles via Fisher-Rao geodesic distance on attention and reweighting advantages by role-alignment score.

  8. AnE: Pushing the Reasoning Frontier of Multimodal LLMs via Anchor Evolution

    cs.CV 2026-05 unverdicted novelty 6.0

    AnE combines Truth Anchor Expansion and Scaffold-Stripping to deliver 10.3% gains on eight multimodal reasoning benchmarks for MLLMs.

  9. Learning Spatiotemporal Sensitivity in Video LLMs via Counterfactual Reinforcement Learning

    cs.CV 2026-05 unverdicted novelty 6.0

    CRPO applies counterfactual videos and a cross-branch relation reward in RL post-training to reduce shortcut reliance in Video LLMs, with gains shown on the new DyBench paired benchmark.

  10. Are Tools Always Beneficial? Learning to Invoke Tools Adaptively for Dual-Mode Multimodal LLM Reasoning

    cs.CL 2026-05 unverdicted novelty 6.0

    AutoTool uses dual-mode RL to let MLLMs adaptively choose tool use or text-only reasoning, reporting 21.8% accuracy gain on V* and 44.9% efficiency gain on POPE versus baselines.

  11. CAVE: A Structured Credit Assignment Approach for Fragmented Visual Evidence Reasoning

    cs.CV 2026-05 unverdicted novelty 6.0

    CAVE is a GRPO-based process-reward method that improves VLMs on fragmented visual reasoning by crediting intermediate actions via belief update, evidence acquisition, and adaptive focus, shown on TRACER-Bench and pub...

  12. Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs

    cs.CV 2026-05 unverdicted novelty 6.0

    PVM adds a parallel branch to LVLMs that directly supplies visual embeddings to prevent attention decay over long generated sequences, yielding accuracy gains on reasoning tasks with minimal overhead.

  13. SSL-R1: Self-Supervised Visual Reinforcement Post-Training for Multimodal Large Language Models

    cs.CV 2026-04 unverdicted novelty 6.0

    SSL-R1 reformulates visual SSL tasks into verifiable puzzles to supply rewards for RL post-training of MLLMs, yielding gains on multimodal benchmarks without external supervision.

  14. Building a Precise Video Language with Human-AI Oversight

    cs.CV 2026-04 unverdicted novelty 6.0

    CHAI framework pairs AI pre-captions with expert human critiques to produce precise video descriptions, enabling open models to outperform closed ones like Gemini-3.1-Pro and improve fine-grained control in video gene...

  15. DR-MMSearchAgent: Deepening Reasoning in Multimodal Search Agents

    cs.CV 2026-04 unverdicted novelty 6.0

    DR-MMSearchAgent derives batch-wide trajectory advantages and uses differentiated Gaussian rewards to prevent premature collapse in multimodal agents, outperforming MMSearch-R1 by 8.4% on FVQA-test.

  16. Cognitive Pivot Points and Visual Anchoring: Unveiling and Rectifying Hallucinations in Multimodal Reasoning Models

    cs.AI 2026-04 conditional novelty 6.0

    Hallucinations in multimodal reasoners cluster at high-entropy pivot tokens due to intermediate-layer visual anchoring failure, and V-STAR mitigates this with attention rewards plus forced reflection.

  17. Deeper Thought, Weaker Aim: Understanding and Mitigating Perceptual Impairment during Reasoning in Multimodal Large Language Models

    cs.CV 2026-03 unverdicted novelty 6.0

    Attention dispersion during extended reasoning impairs MLLM perception on images, and a training-free VRGA framework mitigates it by selecting and reweighting visual attention heads using an entropy-focus criterion.

  18. ReMoT: Reinforcement Learning with Motion Contrast Triplets

    cs.CV 2026-02 conditional novelty 6.0

    Training a 4B vision-language model on rule-generated motion-contrast triplets with GRPO lifts spatio-temporal QA accuracy by about 17 points on the authors' own benchmark and by smaller margins on standard benchmarks.

  19. DeepEyesV2: Toward Agentic Multimodal Model

    cs.CV 2025-11 unverdicted novelty 6.0

    DeepEyesV2 uses a two-stage cold-start plus reinforcement learning pipeline to produce an agentic multimodal model that adaptively invokes tools and outperforms direct RL on real-world reasoning benchmarks.

  20. Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models

    cs.LG 2025-10 conditional novelty 6.0

    A dynamic replay and reweighting scheduler (RECAP) preserves general capabilities during RLVR while keeping reasoning performance at least as good as reasoning-only finetuning.

  21. Adaptive Chain-of-Focus Reasoning via Dynamic Visual Search and Zooming for Efficient VLMs

    cs.CV 2025-05 unverdicted novelty 6.0

    Chain-of-Focus enables VLMs to adaptively search and zoom on important image areas via a two-stage SFT and RL pipeline on a custom 3K-sample dataset, yielding 5% gains on the V* benchmark across resolutions from 224 to 4K.

  22. OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles

    cs.CV 2025-03 conditional novelty 6.0

    Iterative SFT-RL cycles enable a 7B LVLM to develop sophisticated visual chain-of-thought reasoning and improve performance on math and general reasoning benchmarks.

  23. Latent Visual States for Efficient Multimodal Reasoning

    cs.CV 2026-06 unverdicted novelty 5.0

    EVA generates adaptive Latent_slot tokens as internal visual thoughts, trained end-to-end with text tokens via D-GSPO on the EVA-230K dataset, claiming performance gains and better inference efficiency.

  24. VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct

    cs.AI 2026-06 unverdicted novelty 5.0

    VeriEvol decouples prompt difficulty evolution from answer reliability verification to scale verified data for visual math reasoning, lifting benchmark accuracy from 35.42 to 54.73 and adding +3.88 in GRPO RL.

  25. Perceptual Flow Network for Visually Grounded Reasoning

    cs.CV 2026-05 unverdicted novelty 5.0

    PFlowNet decouples perception from reasoning, integrates multi-dimensional rewards with vicinal geometric shaping via variational RL, and reports new SOTA results on V* Bench (90.6%) and MME-RealWorld-lite (67.0%).

  26. Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs

    cs.CV 2026-05 unverdicted novelty 5.0

    PVM adds a parallel learnable branch to LVLMs that supplies visual embeddings on demand to structurally prevent attention decay and visual signal dilution during deep autoregressive generation.

  27. Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding

    cs.LG 2026-04 unverdicted novelty 5.0

    A co-evolving proposer-critic RL framework improves GUI grounding accuracy by letting the model critique its own proposals rendered on screenshots.

  28. Cognitive Pivot Points and Visual Anchoring: Unveiling and Rectifying Hallucinations in Multimodal Reasoning Models

    cs.AI 2026-04 unverdicted novelty 5.0

    Multimodal reasoning models hallucinate at high-entropy cognitive bifurcation points due to loss of visual semantic anchoring, and the V-STAR training paradigm with HVAR rewards and FRM reflection mitigates this by re...

  29. Act Wisely: Cultivating Meta-Cognitive Tool Use in Agentic Multimodal Models

    cs.CV 2026-04 unverdicted novelty 5.0

    HDPO reframes tool efficiency as a conditional objective within accurate trajectories, enabling Metis to reduce tool invocations by orders of magnitude while raising reasoning accuracy.

  30. Smart Picks in the Dark: Towards Efficient RLVR for Reasoning via Tracing Metacognitive Pivots

    cs.LG 2026-06 unverdicted novelty 4.0

    PivotTrace selects unlabeled data for RLVR by quantifying uncertainty via pivot density from attention dynamics, outperforming full supervision using only 29.3% annotations and converging 2.75 times faster.