Eagle 2.5: Boosting Long-Context Post-Training for Frontier
  Vision-Language Models

Guo Chen; Zhiqi Li; Shihao Wang; Jindong Jiang; Yicheng Liu; Lidong Lu; De-An Huang; Wonmin Byeon; Matthieu Le; Tuomas Rintamaki; Tyler Poon; Max Ehrlich; Tuomas Rintamaki; Tyler Poon; Tong Lu; Limin Wang; Bryan Catanzaro; Jan Kautz; Andrew Tao; Zhiding Yu; Guilin Liu

Paper · arXiv 2504.15271

Eagle 2.5: Boosting Long-Context Post-Training for Frontier Vision-Language Models

Guo Chen, Zhiqi Li, Shihao Wang, Jindong Jiang, Yicheng Liu, Lidong Lu, De-An Huang, Wonmin Byeon, Matthieu Le, Tuomas Rintamaki, Tyler Poon, Max Ehrlich, Tuomas Rintamaki, Tyler Poon, Tong Lu, Limin Wang, Bryan Catanzaro, Jan Kautz, Andrew Tao, Zhiding Yu, Guilin Liu

69 upvotesApril 21, 2025arXiv 预印本

AI 摘要

Eagle 2.5 improves long-context multimodal learning through Automatic Degrade Sampling and Image Area Preservation techniques, enhancing VLMs for video and image understanding, and matches state-of-the-art performance on benchmarks.

vision-language modelslong-context multimodal learninglong video comprehensionhigh-resolution image understandingAutomatic Degrade SamplingImage Area Preservationefficiency optimizationslong-context data trainingEagle-Video-110Kmultimodal benchmarksVideo-MME

Abstract

We introduce Eagle 2.5, a family of frontier vision-language models (VLMs) for long-context multimodal learning. Our work addresses the challenges in long video comprehension and high-resolution image understanding, introducing a generalist framework for both tasks. The proposed training framework incorporates Automatic Degrade Sampling and Image Area Preservation, two techniques that preserve contextual integrity and visual details. The framework also includes numerous efficiency optimizations in the pipeline for long-context data training. Finally, we propose Eagle-Video-110K, a novel dataset that integrates both story-level and clip-level annotations, facilitating long-video understanding. Eagle 2.5 demonstrates substantial improvements on long-context multimodal benchmarks, providing a robust solution to the limitations of existing VLMs. Notably, our best model Eagle 2.5-8B achieves 72.4% on Video-MME with 512 input frames, matching the results of top-tier commercial model such as GPT-4o and large-scale open-source models like Qwen2.5-VL-72B and InternVL2.5-78B.