AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Boosting the Power of Small Multimodal Reasoning Models to Match Larger Models with Self-Consistency Training

Cheng Tan, Jingxuan Wei, Zhangyang Gao, Linzhuang Sun, Siyuan Li, Ruifeng Guo, BiHui Yu, Stan Z. Li

ECCV 2024 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 22 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add rationale voting during training (ablation)MC-CoT Base · ScienceQAaverage accuracy: 84.70 -> 90.64Improved
add answer voting during training (ablation)MC-CoT Base · ScienceQAaverage accuracy: 90.19 -> 90.64Improved
vote during training instead of only at inference (ablation)MC-CoT Base · ScienceQAaverage accuracy: 84.58 -> 90.64Improved
combine mean and weighted-mean logits instead of either alone (ablation)MC-CoT Base · ScienceQAaverage accuracy: 90.03 -> 90.64 (mean only); 89.79 -> 90.64 (weighted only)Improved
apply MC-CoT self-consistency training to Multimodal-CoT BaseMultimodal-CoT Base (223M) · ScienceQAaverage accuracy: 84.91 -> 90.64Improved
apply MC-CoT self-consistency training to Multimodal-CoT LargeMultimodal-CoT Large (738M) · ScienceQAaverage accuracy: 91.68 -> 93.37Improved
use FLAN-T5 Large instead of UnifiedQA Large as base modelMC-CoT · ScienceQAaverage accuracy: 93.37 -> 94.88Improved
apply MC-CoT to A-OKVQA multi-choice with base modelMultimodal-CoT Base · A-OKVQA multi-choiceaccuracy: 50.6 -> 71.0Improved
rationale quality gain from MC-CoT versus answer accuracy gainMC-CoT Base · ScienceQARougeL of rationale and answer accuracy: about 1% RougeL, about 6% answer accuracyImproved
answer voting removed: category effects (ablation)MC-CoT Base · ScienceQA NAT and TXTaccuracy: NAT 91.87 -> 92.36; TXT 92.28 -> 92.42 without answer votingMixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy