Eagle: Exploring The Design Space for Multimodal LLMs with Mixture of Encoders
pith:KBR4GOJOopen to challenge →
read the original abstract
The ability to accurately interpret complex visual information is a crucial topic of multimodal large language models (MLLMs). Recent work indicates that enhanced visual perception significantly reduces hallucinations and improves performance on resolution-sensitive tasks, such as optical character recognition and document analysis. A number of recent MLLMs achieve this goal using a mixture of vision encoders. Despite their success, there is a lack of systematic comparisons and detailed ablation studies addressing critical aspects, such as expert selection and the integration of multiple vision experts. This study provides an extensive exploration of the design space for MLLMs using a mixture of vision encoders and resolutions. Our findings reveal several underlying principles common to various existing strategies, leading to a streamlined yet effective design approach. We discover that simply concatenating visual tokens from a set of complementary vision encoders is as effective as more complex mixing architectures or strategies. We additionally introduce Pre-Alignment to bridge the gap between vision-focused encoders and language tokens, enhancing model coherence. The resulting family of MLLMs, Eagle, surpasses other leading open-source models on major MLLM benchmarks.
This paper has not been read by Pith yet.
Forward citations
Cited by 18 Pith papers
-
FLARE: Fully Integration of Vision-Language Representations for Deep Cross-Modal Understanding
FLARE is a vision-language model family using text-guided vision encoding, context-aware alignment decoding, dual-semantic mapping loss, and text-driven VQA synthesis to achieve deep cross-modal integration, outperfor...
-
OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning
OCRBench v2 is a new benchmark with four times more tasks than prior versions that reveals most large multimodal models score below 50 out of 100 on visual text tasks and share five specific weaknesses.
-
Beyond Encoder Accumulation: Measuring Encoder Roles in Multi-Encoder VLMs
Retraining all 31 subsets of five vision encoders shows Capacity and Necessity are distinct, pre-projector effective rank predicts residual performance at fixed parameter count, and high-Capacity plus adaptive complem...
-
Investigating Adversarial Robustness of Multi-modal Large Language Models
Robust vision encoders from multimodal adversarial pretraining transfer to MLLMs and deliver large gains in adversarial captioning and VQA performance, while test-time stochastic transformations provide an effective b...
-
PRISM: Synergizing Vision Foundation Models via Self-organized Expert Specialization
PRISM is a two-stage MoE framework that achieves new state-of-the-art results on PASCAL-Context and NYUD-v2 by enabling self-organized expert specialization across diverse vision foundation models.
-
ROVER: Routing Object-Centric Visual Evidence for Grounded Multi-Image Reasoning
ROVER introduces a learnable routing plugin for object-centric visual evidence in MLLMs via token triplets and differential attention, reporting gains on MM-GCoT and VideoEspresso when integrated into Qwen2.5-VL-7B.
-
20/20 Vision Language Models: A Prescription for Better VLMs through Data Curation Alone
Data curation alone raises VLM accuracy by 11+ points on average, improves reliability and OOD generalization, and achieves near-frontier results at far lower training and inference cost.
-
20/20 Vision Language Models: A Prescription for Better VLMs through Data Curation Alone
Data curation alone raises VLM accuracy by more than 11 points on average across many benchmarks while cutting required training compute by up to 87 times.
-
Boosting Visual Instruction Tuning with Self-Supervised Guidance
Mixing 3-10% of visually grounded self-supervised instructions into visual instruction tuning consistently boosts MLLM performance on vision-centric benchmarks.
-
CoME-VL: Scaling Complementary Multi-Encoder Vision-Language Learning
CoME-VL fuses contrastive and self-supervised vision encoders via entropy-guided multi-layer aggregation and RoPE cross-attention to improve vision-language model performance on benchmarks.
-
Visual Compositional Tuning
COMPACT synthesizes compositional visual instruction data to reduce VIT training data by 90% while achieving 100.2% of full performance across eight multimodal benchmarks.
-
InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models
InternVL3-78B sets a new open-source SOTA of 72.2 on MMMU via native joint multimodal pre-training, V2PE, MPO, and test-time scaling while remaining competitive with proprietary models.
-
Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
InternVL 2.5 is the first open-source MLLM to surpass 70% on the MMMU benchmark via model, data, and test-time scaling, with a 3.7-point gain from chain-of-thought reasoning.
-
From Structure to Synergy: A Survey of Vision-Language Perception Paradigm Evolution in Multimodal Large Language Models
The survey formalizes MLLM perception as a unified vision-language capability and traces its evolution via a new five-stage taxonomy while outlining future challenges.
-
VaaWIT: Visual-Aware Adaptation of Large Language Models for Multilingual Web Image Translation
VaaWIT proposes DSAM and VAA modules to adapt LLMs for multilingual web image translation, claiming outperformance over open-source baselines on benchmarks.
-
EvoDriveVLA: Evolving Driving VLA Models via Collaborative Perception-Planning Distillation
EvoDriveVLA uses collaborative perception-planning distillation with self-anchor and future-aware teachers to fix perception degradation and long-term instability in driving VLA models, reaching SOTA on nuScenes and NAVSIM.
-
ThinkAct: Vision-Language-Action Reasoning via Reinforced Visual Latent Planning
ThinkAct introduces reinforced visual latent planning in a dual VLA system to enable better long-horizon reasoning and adaptation for embodied tasks.
-
NVILA: Efficient Frontier Visual Language Models
NVILA improves on VILA with a scale-then-compress visual token strategy and full-lifecycle efficiency optimizations, matching or exceeding leading VLMs on image and video benchmarks while reducing training cost 1.9-5....
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.