Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.14229 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917088590823424 |
|---|---|
| author | Broadbent, Jim Cohen, Felix Hvilshøj, Frederik Landau, Eric Sasoglu, Eren |
| author_facet | Broadbent, Jim Cohen, Felix Hvilshøj, Frederik Landau, Eric Sasoglu, Eren |
| contents | We simplify space binding by focusing on two core components, a single encoder per modality and high-quality data; enabling training state-of-the-art models on a single GPU in a few hours as opposed to multiple days. We present EBind, an Easy, data-centric, and parameter-efficient method to Bind the embedding spaces of multiple contrastive models. We demonstrate that a simple 1.8B-parameter image-text-video-audio-3D model can outperform models 4 to 17x the size. The key to achieving this is a carefully curated dataset of three complementary data sources: i) 6.7M fully-automated multimodal quintuples sourced via SOTA retrieval models, ii) 1M diverse, semi-automated triples annotated by humans as negative, partial, or positive matches, and iii) 3.4M pre-existing captioned data items. We use 13 different evaluations to demonstrate the value of each data source. Due to limitations with existing benchmarks, we further introduce the first high-quality, consensus-annotated zero-shot classification benchmark between audio and PCs. In contrast to related work, we will open-source our code, model weights, and datasets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_14229 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | EBind: a practical approach to space binding Broadbent, Jim Cohen, Felix Hvilshøj, Frederik Landau, Eric Sasoglu, Eren Machine Learning We simplify space binding by focusing on two core components, a single encoder per modality and high-quality data; enabling training state-of-the-art models on a single GPU in a few hours as opposed to multiple days. We present EBind, an Easy, data-centric, and parameter-efficient method to Bind the embedding spaces of multiple contrastive models. We demonstrate that a simple 1.8B-parameter image-text-video-audio-3D model can outperform models 4 to 17x the size. The key to achieving this is a carefully curated dataset of three complementary data sources: i) 6.7M fully-automated multimodal quintuples sourced via SOTA retrieval models, ii) 1M diverse, semi-automated triples annotated by humans as negative, partial, or positive matches, and iii) 3.4M pre-existing captioned data items. We use 13 different evaluations to demonstrate the value of each data source. Due to limitations with existing benchmarks, we further introduce the first high-quality, consensus-annotated zero-shot classification benchmark between audio and PCs. In contrast to related work, we will open-source our code, model weights, and datasets. |
| title | EBind: a practical approach to space binding |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2511.14229 |