SMIR: Efficient Synthetic Data Pipeline To Improve Multi-Image Reasoning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Li, Andrew, Thapa, Rahul, Chalamala, Rahul, Wu, Qingyang, Chen, Kezhen, Zou, James
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916616712749056
author Li, Andrew
Thapa, Rahul
Chalamala, Rahul
Wu, Qingyang
Chen, Kezhen
Zou, James
author_facet Li, Andrew
Thapa, Rahul
Chalamala, Rahul
Wu, Qingyang
Chen, Kezhen
Zou, James
contents Vision-Language Models (VLMs) excel at understanding single images, aided by high-quality instruction datasets. However, multi-image reasoning remains underexplored in the open-source community due to two key challenges: (1) scaling datasets with correlated images and complex reasoning instructions is resource-intensive, and (2) robust evaluation benchmarks for multi-image tasks are lacking. To address this, we introduce SMiR, a synthetic data-generation pipeline for multi-image reasoning, along with a high-quality dataset generated using this pipeline. SMiR efficiently extracts correlated images via multimodal embeddings, integrates visual and descriptive information, and leverages open-source LLMs to generate quality instructions. Using this approach, we produce 160K synthetic training samples, offering a cost-effective alternative to closed-source solutions. Additionally, we present SMiR-Bench, a multi-image reasoning benchmark comprising 200 diverse examples across seven complex reasoning tasks. SMiR-Bench is multi-turn and employs a VLM judge to evaluate free-form responses, providing a comprehensive assessment of model expressiveness and reasoning capability across modalities. We demonstrate the effectiveness of SMiR by fine-tuning open-source VLMs and evaluating them on SMiR-Bench.
format Preprint
id arxiv_https___arxiv_org_abs_2501_03675
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SMIR: Efficient Synthetic Data Pipeline To Improve Multi-Image Reasoning
Li, Andrew
Thapa, Rahul
Chalamala, Rahul
Wu, Qingyang
Chen, Kezhen
Zou, James
Computer Vision and Pattern Recognition
Vision-Language Models (VLMs) excel at understanding single images, aided by high-quality instruction datasets. However, multi-image reasoning remains underexplored in the open-source community due to two key challenges: (1) scaling datasets with correlated images and complex reasoning instructions is resource-intensive, and (2) robust evaluation benchmarks for multi-image tasks are lacking. To address this, we introduce SMiR, a synthetic data-generation pipeline for multi-image reasoning, along with a high-quality dataset generated using this pipeline. SMiR efficiently extracts correlated images via multimodal embeddings, integrates visual and descriptive information, and leverages open-source LLMs to generate quality instructions. Using this approach, we produce 160K synthetic training samples, offering a cost-effective alternative to closed-source solutions. Additionally, we present SMiR-Bench, a multi-image reasoning benchmark comprising 200 diverse examples across seven complex reasoning tasks. SMiR-Bench is multi-turn and employs a VLM judge to evaluate free-form responses, providing a comprehensive assessment of model expressiveness and reasoning capability across modalities. We demonstrate the effectiveness of SMiR by fine-tuning open-source VLMs and evaluating them on SMiR-Bench.
title SMIR: Efficient Synthetic Data Pipeline To Improve Multi-Image Reasoning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2501.03675