AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

DWIM: Towards Tool-aware Visual Reasoning via Discrepancy-aware Workflow Generation & Instruct-Masking Tuning

Fucai Ke, Vijay Kumar B G, Xingjian Leng, Zhixi Cai, Zaid Khan, Weiqing Wang, Pari Delir Haghighi, Hamid Rezatofighi, Manmohan Chandraker

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 5 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 discrepancy-aware training workflow generation (ablation)LLaMa3.1-8B generator · GQAdata utilization: 48.2 -> 68.3Improved
add discrepancy-aware training workflow generation (ablation)DWIM · GQAaccuracy: 57.9 -> 69.3Improved
add discrepancy-aware training workflow generation (ablation)LLaMa3.1-8B generator · SugarCREPEdata utilization: 56.5 -> 76.4Improved
add discrepancy-aware training workflow generation (ablation)DWIM · SugarCREPEaccuracy: 60.0 -> 74.6Improved
use GPT4O instead of LLaMa3.1 as standard workflow generator (ablation)standard generation · GQAaccuracy: 57.9 -> 58.0No clear effect
replace SFT with Instruct-Masking on standard workflows (ablation)LLaMa3.1-8B · GQAaccuracy: 53.6 -> 57.9Improved
replace SFT with Instruct-Masking on standard workflows (ablation)LLaMa3.1-8B · SugarCREPEaccuracy: 57.1 -> 60.0Improved
replace SFT with Instruct-Masking on discrepancy-aware workflows (ablation)LLaMa3.1-8B · GQAaccuracy: 54.8 -> 69.3Improved
replace SFT with Instruct-Masking on discrepancy-aware workflows (ablation)LLaMa3.1-8B · SugarCREPEaccuracy: 58.3 -> 74.6Improved
mask only effective actions instead of random masking (ablation)Instruct-Masking · GQAaccuracy: 65.1 -> 69.3Improved
mask only effective actions instead of also masking rethink actions (ablation)Instruct-Masking · SugarCREPEaccuracy: 70.7 -> 74.6Improved
add DWIM training to frozen 10-shot Mistral-v0.3-7BMistral-v0.3-7B · VCRaccuracy: 52.0 -> 67.5Improved

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