Pith Number

pith:27P4QCOH

pith:2025:27P4QCOHDNSQZP674SDDZJAAFE

not attested not anchored not stored refs resolved

NoisyGRPO: Incentivizing Multimodal CoT Reasoning via Noise Injection and Bayesian Estimation

Jiaxuan Sun, Longtian Qiu, Shan Ning, Xuming He

Noise injection into visual inputs and Bayesian advantage estimation improve generalization in multimodal chain-of-thought reasoning.

arxiv:2510.21122 v3 · 2025-10-24 · cs.CV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{27P4QCOHDNSQZP674SDDZJAAFE}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Experiments on standard CoT quality, general capability, and hallucination benchmarks demonstrate that NoisyGRPO substantially improves generalization and robustness, especially in RL settings with small-scale MLLMs such as Qwen2.5-VL 3B.

C2weakest assumption

The assumption that the injected Gaussian noise level can be directly used as a prior in a Bayesian model whose posterior advantage estimate reliably prefers visually grounded trajectories over those that succeed only under noise; this premise is invoked in the description of Bayesian Advantage Estimation without further justification of the likelihood model or prior calibration.

C3one line summary

NoisyGRPO is an RL framework that perturbs visual inputs with Gaussian noise for exploration and computes trajectory advantages via Bayesian posterior fusion of noise prior and reward likelihood to improve multimodal CoT generalization.

References

63 extracted · 63 resolved · 34 Pith anchors

[1] Gpt-4 technical report 2023

[2] Flamingo: a visual language model for few-shot learning.Advances in Neural Information Processing Systems, 35:23716–23736 2022

[3] Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond 2023 · arXiv:2308.12966

[4] Qwen2.5-VL Technical Report 2025 · arXiv:2502.13923

[5] Are We on the Right Way for Evaluating Large Vision-Language Models? 2024 · arXiv:2403.20330

Formal links

2 machine-checked theorem links

Cited by

2 papers in Pith

WikiCLIP: An Efficient Contrastive Baseline for Open-domain Visual Entity Recognition

PDCR: Perception-Decomposed Confidence Reward for Vision-Language Reasoning

Receipt and verification

First computed	2026-06-08T01:03:50.303122Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

d7dfc809c71b650cbfdfe4863ca4002937cc5242ed4283c9829873f921addd41

Aliases

arxiv: 2510.21122 · arxiv_version: 2510.21122v3 · doi: 10.48550/arxiv.2510.21122 · pith_short_12: 27P4QCOHDNSQ · pith_short_16: 27P4QCOHDNSQZP67 · pith_short_8: 27P4QCOH

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/27P4QCOHDNSQZP674SDDZJAAFE \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: d7dfc809c71b650cbfdfe4863ca4002937cc5242ed4283c9829873f921addd41

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "a4f0131f9e76647b87de5c63a815a71e55aef40f8f739a33f990cc727e40a340",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2025-10-24T03:23:34Z",
    "title_canon_sha256": "6d195fa51f33faadd7c1379b4d88688ddf3cbbf68b8dcd5aedc87f7aad6e88f8"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2510.21122",
    "kind": "arxiv",
    "version": 3
  }
}