Add metadata and improve model card
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,19 +1,21 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
datasets:
|
| 3 |
- VLM2Vec/MMEB-V2
|
| 4 |
language:
|
| 5 |
- en
|
| 6 |
-
|
| 7 |
-
|
| 8 |
---
|
|
|
|
| 9 |
# PLUME-Qwen2-VL-2B
|
| 10 |
|
| 11 |
**PLUME: Latent Reasoning Based Universal Multimodal Embedding**
|
| 12 |
|
| 13 |
-
PLUME is a latent reasoning framework for universal multimodal embedding (UME). It replaces explicit chain-of-thought (CoT) generation with a short autoregressive rollout of continuous latent states, achieving stronger retrieval
|
| 14 |
-
performance while delivering **over 30x faster inference** compared to explicit-CoT methods.
|
| 15 |
|
| 16 |
-
**[Project Page](https://haoxiangzhao12138.github.io/PLUME/)** | **[Paper](https://arxiv.org/abs/
|
| 17 |
|
| 18 |
## Highlights
|
| 19 |
|
|
@@ -43,5 +45,18 @@ huggingface-cli download CUDAOUTOFMEMORY/PLUME-Qwen2-VL-2B --local-dir /path/to/
|
|
| 43 |
# Option 2: git clone (requires git-lfs)
|
| 44 |
git lfs install
|
| 45 |
git clone https://huggingface.co/CUDAOUTOFMEMORY/PLUME-Qwen2-VL-2B
|
| 46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
```
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen2-VL-2B-Instruct
|
| 4 |
datasets:
|
| 5 |
- VLM2Vec/MMEB-V2
|
| 6 |
language:
|
| 7 |
- en
|
| 8 |
+
library_name: transformers
|
| 9 |
+
pipeline_tag: feature-extraction
|
| 10 |
---
|
| 11 |
+
|
| 12 |
# PLUME-Qwen2-VL-2B
|
| 13 |
|
| 14 |
**PLUME: Latent Reasoning Based Universal Multimodal Embedding**
|
| 15 |
|
| 16 |
+
PLUME is a latent reasoning framework for universal multimodal embedding (UME). It replaces explicit chain-of-thought (CoT) generation with a short autoregressive rollout of continuous latent states, achieving stronger retrieval performance while delivering **over 30x faster inference** compared to explicit-CoT methods.
|
|
|
|
| 17 |
|
| 18 |
+
**[Project Page](https://haoxiangzhao12138.github.io/PLUME/)** | **[Paper](https://arxiv.org/abs/2604.02073)** | **[Code](https://github.com/haoxiangzhao12138/PLUME)**
|
| 19 |
|
| 20 |
## Highlights
|
| 21 |
|
|
|
|
| 45 |
# Option 2: git clone (requires git-lfs)
|
| 46 |
git lfs install
|
| 47 |
git clone https://huggingface.co/CUDAOUTOFMEMORY/PLUME-Qwen2-VL-2B
|
| 48 |
+
```
|
| 49 |
+
|
| 50 |
+
## Citation
|
| 51 |
+
|
| 52 |
+
```bibtex
|
| 53 |
+
@misc{he2026plumelatentreasoningbased,
|
| 54 |
+
title={PLUME: Latent Reasoning Based Universal Multimodal Embedding},
|
| 55 |
+
author={Chenwei He and Xiangzhao Hao and Tianyu Yang and Yuxiang Ma and Yuheng Jia and Lingxiang Wu and Chaoyang Zhao and Haiyun Guo and Jinqiao Wang},
|
| 56 |
+
year={2026},
|
| 57 |
+
eprint={2604.02073},
|
| 58 |
+
archivePrefix={arXiv},
|
| 59 |
+
primaryClass={cs.CV},
|
| 60 |
+
url={https://arxiv.org/abs/2604.02073},
|
| 61 |
+
}
|
| 62 |
```
|