AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

FlowGen: Synthesizing Diverse Flowcharts to Enhance and Benchmark MLLM Reasoning

Kaiwen Shi, Sichen Liu, Ziyue Lin, Hangrui Guo, Gong Cheng

ICLR 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Multimodal LLM benchmarks

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Introduces FlowGen, a flowchart synthesis framework with fine-grained control over graph complexity and rendering style, used to train and benchmark multimodal LLM flowchart reasoning.

Problem

Interpreting flowcharts is hard due to structural complexity and visual diversity, and existing flowchart datasets lack fine-grained control over graph complexity and rendering style, limiting training and testing of multimodal LLMs.

Approach

A controllable flowchart generator that synthesizes diverse flowcharts with explicit control over graph complexity and rendering style, producing data both to enhance and to benchmark MLLM flowchart reasoning.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fine-tune on FlowGen synthetic flowchartsQwen2.5-VL-3B · FlowVQA parsingstrict F1: 12.7 -> 51.3Improved
fine-tune on FlowGen synthetic flowchartsQwen2.5-VL-7B · FlowLearn parsingstrict F1: 43.2 -> 60.1Improved
fine-tune on FlowGen synthetic flowchartsLlava-V1.6-Mistral-7B · parsing benchmarks (FlowVQA)strict F1: 2.6 -> 4.9No clear effect
fine-tune on FlowGen synthetic flowchartsInternVL3-2B · FlowLearn parsingstrict F1: 21.8 -> 13.2Hurt
mix FlowGen parsing data into QA SFTQwen2.5-VL-7B · FlowLearn QAaccuracy: 85.3 -> 89.1Improved
mix FlowGen parsing data into QA SFTQwen2.5-VL-3B · FlowLearn QAaccuracy: 85.1 -> 84.7No clear effect
mix FlowGen parsing data into MMMU SFTQwen2.5-VL-7B · MMMUaccuracy: 62.6 -> 65.3Improved
prompt QA with triplets extracted by FlowGen-tuned Qwen2.5-VL-7B instead of self-extractedQwen2.5-VL-3B · FlowLearn QAaccuracy: 77.1 -> 88.2Improved
add FlowGen-extracted triplets to promptQwen2.5-VL-7B · FlowLearn QAaccuracy: 71.1 -> 68.6Hurt
train on multiple renderers instead of only Mermaid (ablation)FlowGen SFT (Qwen2.5-VL-7B) · CBD / FC B parsingstrict F1: 53.3 -> 55.4 (CBD); 38.1 -> 40.8 (FC B)Improved
train on multiple renderers instead of only Mermaid (ablation)FlowGen SFT (Qwen2.5-VL-7B) · FlowVQA / FlowLearn parsingstrict F1: 76.2 (Mermaid only) vs 70.9 (FlowVQA)Hurt
add nested subgraphs to training data (ablation)FlowGen SFT (Qwen2.5-VL-7B) · hdBPMN parsingstrict F1: 16.6 -> 18.3Improved

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