Papers › VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with...

VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning

10 Apr 2025arXiv:2504.08837archive 2025-07-28

Haozhe Wang, Chao Qu, Zuming Huang, Wei Chu, Fangzhen Lin, Wenhu Chen

Recently, slow-thinking systems like GPT-o1 and DeepSeek-R1 have demonstrated great potential in solving challenging problems through explicit reflection. They significantly outperform the best fast-thinking models, such as GPT-4o, on various math and science benchmarks. However, their multimodal reasoning capabilities remain on par with fast-thinking models. For instance, GPT-o1's performance on benchmarks like MathVista, MathVerse, and MathVision is similar to fast-thinking models. In this paper, we aim to enhance the slow-thinking capabilities of vision-language models using reinforcement learning (without relying on distillation) to advance the state of the art. First, we adapt the GRPO algorithm with a novel technique called Selective Sample Replay (SSR) to address the vanishing advantages problem. While this approach yields strong performance, the resulting RL-trained models exhibit limited self-reflection or self-verification. To further encourage slow-thinking, we introduce Forced Rethinking, which appends a rethinking trigger token to the end of rollouts in RL training, explicitly enforcing a self-reflection reasoning step. By combining these two techniques, our model, VL-Rethinker, advances state-of-the-art scores on MathVista, MathVerse to achieve 80.4%, 63.5% respectively. VL-Rethinker also achieves open-source SoTA on multi-disciplinary benchmarks such as MathVision, MMMU-Pro, EMMA, and MEGA-Bench, narrowing the gap with OpenAI-o1. Our empirical results show the effectiveness of our approaches.

PaperPDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2504.08837")

Code

Syntology Ran 2 of 12 code samples harvested from 1 repository linked to this paper; 10 have no recorded run. Of those that ran: 1 ran · our draft was wrong; 1 ran with no contract checked.

By repository: community (archive-listed): 12 samples from 1 repository, 2 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

infly-ai/inf-mllm mentioned on GitHubpytorch report
tiger-ai-lab/vl-rethinker mentioned on GitHubpytorchApache-2.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

12 samples harvested; 2 ran; 0 honoured the contract we drafted; 10 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

1ran · our draft was wrong
1ran
10unverified

Licence: 0 of the 12 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from tiger-ai-lab/vl-rethinker. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

reward_fn tiger-ai-lab/vl-rethinker/openrlhf/cli/eval_ray.py community (archive-listed) ran · our draft was wrong Apache-2.0 (permissive) · 18ed61dcfadbbdf0 · report
zero_pad_sequences tiger-ai-lab/vl-rethinker/openrlhf/datasets/utils.py community (archive-listed) ran Apache-2.0 (permissive) · c48a5c183e44469e · report
compute_approx_kl tiger-ai-lab/vl-rethinker/openrlhf/models/utils.py community (archive-listed) unverified Apache-2.0 (permissive) · de2fefc774bb0362 · report
compute_reward tiger-ai-lab/vl-rethinker/openrlhf/models/utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 0d27f3b8c425d79f · report
convert_ring_attn_params tiger-ai-lab/vl-rethinker/openrlhf/models/ring_attn_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · d66f259aaf6cd9c7 · report
entropy_from_logits tiger-ai-lab/vl-rethinker/openrlhf/models/actor.py community (archive-listed) unverified Apache-2.0 (permissive) · c710eef68603bd62 · report
exist_and_not_none tiger-ai-lab/vl-rethinker/openrlhf/datasets/utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 71edf11f55693a58 · report
get_print tiger-ai-lab/vl-rethinker/openrlhf/models/loss.py community (archive-listed) unverified Apache-2.0 (permissive) · 27ab2676fee3ce6a · report
log_probs_from_logits tiger-ai-lab/vl-rethinker/openrlhf/models/utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 0612b61574b81472 · report
preprocess_data tiger-ai-lab/vl-rethinker/openrlhf/datasets/prompts_dataset.py community (archive-listed) unverified Apache-2.0 (permissive) · 8368fe60f5326faa · report
read_jsonl tiger-ai-lab/vl-rethinker/openrlhf/trainer/evaluator.py community (archive-listed) unverified Apache-2.0 (permissive) · 059f7ea3562486c2 · report
reset_ring_attn_position_ids tiger-ai-lab/vl-rethinker/openrlhf/models/ring_attn_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · e93117d99aa6b210 · report

Tasks

MathMultimodal Reasoning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections