aiqbench analysis
NeoMME introduces a novel single-tower architecture for multimodal document retrieval, eliminating the parameter overhead typically associated with repurposing generative vision-language models. By pretraining from scratch with a masked discrete-diffusion objective, it achieves superior throughput and compression ratios while maintaining high accuracy on benchmark datasets. This advancement offers a more efficient foundation encoder specifically tailored for non-generative visual document tasks.
NeoMME: A Single-Tower Multimodal-Native Multilingual Foundation Encoder for Efficient Fine-Tuning and Inference
arXiv:2609.01657v1 Announce Type: cross Abstract: Multimodal models often build on architectures designed for generative vision-language modeling, typically combining separately pretrained vision encoders with causal language models. Visual document retrievers such as ColPali repurpose these models as encoders, carrying over the parameter and compute overhead of a VLM for a non-generative task. We introduce NeoMME, a family of 260M and 800M-parameter Multimodal and Multilingual bidirectional Encoders that process multilingual text and raw image patches in a single bidirectional Transformer encoder. Both models are pretrained from scratch with a masked discrete-diffusion text objective, conditioned on visible image patches for multimodal examples. Both support a 16,384-token context, enough to encode up to two standard 4K UHD images. To demonstrate its downstream capabilities, we fine-tune NeoMME with jointly trained dense and late-interaction heads. On the ViDoRe v3 benchmark, the resulting NeoMME-Retriever 260M outperforms all evaluated models strictly below 800M parameters with 0.523 nDCG@10, while NeoMME-Retriever 800M reaches 0.556. At a matched 2048x2048 image input size on an NVIDIA L40S, NeoMME-260M encodes pages with about 2x the throughput of ColModernVBERT. Hierarchical token pooling and asymmetric quantization compress late-interaction multimodal document embeddings by 255x while preserving over 95% of baseline nDCG@10. We contribute NeoMME to Hugging Face Transformers and release the pretrained backbone and retrieval-compatible checkpoints under Apache 2.0 at https://hf.co/collections/Hcompany/neomme.
Abstract:Multimodal models often build on architectures designed for generative vision-language modeling, typically combining separately pretrained vision encoders with causal language models. Visual document retrievers such as ColPali repurpose these models as encoders, carrying over the parameter and compute overhead of a VLM for a non-generative task. We introduce NeoMME, a family of 260M and 800M-parameter Multimodal and Multilingual bidirectional Encoders that process multilingual text and raw image patches in a single bidirectional Transformer encoder. Both models are pretrained from scratch with a masked discrete-diffusion text objective, conditioned on visible image patches for multimodal examples. Both support a 16,384-token context, enough to encode up to two standard 4K UHD images. To demonstrate its downstream capabilities, we fine-tune NeoMME with jointly trained dense and late-interaction heads. On the ViDoRe v3 benchmark, the resulting NeoMME-Retriever 260M outperforms all evaluated models strictly below 800M parameters with 0.523 nDCG@10, while NeoMME-Retriever 800M reaches 0.556. At a matched 2048x2048 image input size on an NVIDIA L40S, NeoMME-260M encodes pages with about 2x the throughput of ColModernVBERT. Hierarchical token pooling and asymmetric quantization compress late-interaction multimodal document embeddings by 255x while preserving over 95% of baseline nDCG@10. We contribute NeoMME to Hugging Face Transformers and release the pretrained backbone and retrieval-compatible checkpoints under Apache 2.0 at this https URL.
Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2609.01657 [cs.IR]
(or arXiv:2609.01657v1 [cs.IR] for this version)
https://doi.org/10.48550/arXiv.2609.01657 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 315 words
Related coverage
NeoMME: an efficient Multimodal-native and Multilingual Encoder
Hugging Face · Sep 3, 2026
H Company Releases NeoMME: A Family of 260M and 800M Single-Tower Multimodal Encoders That Drop the Vision Tower and Causal Decoder
MarkTechPost · Sep 6, 2026
[AINews] Agents disprove Polysemanticity; greatly improve Interpretability & Alignment
Latent Space · Oct 16, 2026
[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model
Latent Space · Oct 13, 2026