# Omni-Embed-Mini: Binding Modalities Without Forgetting via Dense Distillation

arXiv: https://arxiv.org/abs/2610.02148 · HTML: https://mlsanity.com/paper/2610.02148/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 32 of its 54 references that are in the map.
- First posted on arXiv 1 Oct 2026.

## Builds on (most cited first, 30 of 32)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models](https://mlsanity.com/paper/2506.05176/) (2025)
- [BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Models](https://mlsanity.com/paper/2104.08663/) (2021)
- [ImageBind: One Embedding Space To Bind Them All](https://mlsanity.com/paper/2305.05665/) (2023)
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://mlsanity.com/paper/2007.00808/) (2020)
- [Dense-Captioning Events in Videos](https://mlsanity.com/paper/1705.00754/) (2017)
- [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://mlsanity.com/paper/2211.06687/) (2022)
- [Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation](https://mlsanity.com/paper/2004.09813/) (2020)
- [C-Pack: Packed Resources For General Chinese Embeddings](https://mlsanity.com/paper/2309.07597/) (2023)
- [Clotho: An Audio Captioning Dataset](https://mlsanity.com/paper/1910.09387/) (2019)
- [AudioCLIP: Extending CLIP to Image, Text and Audio](https://mlsanity.com/paper/2106.13043/) (2021)
- [WavCaps: A ChatGPT-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research](https://mlsanity.com/paper/2303.17395/) (2023)
- [VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents](https://mlsanity.com/paper/2410.10594/) (2024)
- [Generative Representational Instruction Tuning](https://mlsanity.com/paper/2402.09906/) (2024)
- [Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking](https://mlsanity.com/paper/2601.04720/) (2026)
- [TED-LIUM 3: twice as much data and corpus repartition for experiments on speaker adaptation](https://mlsanity.com/paper/1805.04699/) (2018)
- [EmbeddingGemma: Powerful and Lightweight Text Representations](https://mlsanity.com/paper/2509.20354/) (2025)
- [Jasper and Stella: distillation of SOTA embedding models](https://mlsanity.com/paper/2412.19048/) (2024)
- [Natural Language Supervision for General-Purpose Audio Representations](https://mlsanity.com/paper/2309.05767/) (2023)
- [Scaling up masked audio encoder learning for general audio classification](https://mlsanity.com/paper/2406.06992/) (2024)
- [Think Then Embed: Generative Context Improves Multimodal Embedding](https://mlsanity.com/paper/2510.05014/) (2025)
- [RzenEmbed: Towards Comprehensive Multimodal Retrieval](https://mlsanity.com/paper/2510.27350/) (2025)
- [ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios](https://mlsanity.com/paper/2601.08620/) (2026)
- [Scaling Language-Centric Omnimodal Representation Learning](https://mlsanity.com/paper/2510.11693/) (2025)
- [HeySQuAD: A Spoken Question Answering Dataset](https://mlsanity.com/paper/2304.13689/) (2023)
- [e5-omni: Explicit Cross-modal Alignment for Omni-modal Embeddings](https://mlsanity.com/paper/2601.03666/) (2026)
- [GSQA: An End-to-End Model for Generative Spoken Question Answering](https://mlsanity.com/paper/2312.09781/) (2023)

## Abstract (arXiv)

Extending a text embedding model to new modalities typically degrades text retrieval quality, and existing omni-modal embedders compensate with multi-billion parameters. We present Omni-Embed-Mini, a 0.9B-parameter model that maps text, speech, audio, images, video, and visually-rich documents into a single shared cosine space without updating any text-side parameter. Our key insight is that the teacher signal requires no separate embedding model: each media sample is paired with a dense cascaded caption, and the teacher target is simply the frozen backbone's own embedding of that caption. Because teacher and student share the same backbone weights, they inhabit byte-identical geometry, and lightweight projectors plus phased LoRA adapters on the modality encoders suffice for alignment. Training combines a Matryoshka SigLIP contrastive loss with an online hybrid hard-negative miner whose negatives sharpen as the encoder improves. The recipe carries over to a 2.3B variant by swapping in a native vision-language backbone. Omni-Embed-Mini-0.9B keeps its text weights bit-identical to the backbone, so training cannot regress text retrieval (49.57 nDCG@10 on MTEB-v2 BEIR-8), while extending it to five additional modalities, and is ~2.7x to 9.5x smaller than every open omni embedder we compare against. The 2.3B variant is competitive with the closed gemini-embedding-2, edging ahead of it on the overall-modality average. Models, code, data and evaluation harness are on our project page: https://omniembed.cvmbzuai.com