Add pipeline tag and fix image path in model card (#1)
Browse files- Add pipeline tag and fix image path in model card (1523c27143502a99ce97b59f5b2bb1758f13a7a7)
Co-authored-by: Niels Rogge <[email protected]>
README.md
CHANGED
|
@@ -1,27 +1,27 @@
|
|
| 1 |
---
|
| 2 |
-
license: other
|
| 3 |
-
license_name: other
|
| 4 |
-
license_link: https://github.com/TencentARC/TokLIP/blob/main/LICENSE
|
| 5 |
-
language:
|
| 6 |
-
- en
|
| 7 |
base_model:
|
| 8 |
- google/siglip2-so400m-patch16-384
|
| 9 |
- google/siglip2-so400m-patch16-256
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
tags:
|
| 11 |
- Tokenizer
|
| 12 |
- CLIP
|
| 13 |
- UnifiedMLLM
|
| 14 |
---
|
| 15 |
|
| 16 |
-
|
| 17 |
# TokLIP: Marry Visual Tokens to CLIP for Multimodal Comprehension and Generation
|
| 18 |
|
| 19 |
<h5 align="center">
|
| 20 |
|
| 21 |
[](https://arxiv.org/abs/2505.05422)
|
| 22 |
[](https://github.com/TencentARC/TokLIP)
|
| 23 |
-
[ tokens and **incorporating CLIP-level semantics** while enabling end-to-end multimodal autoregressive training with standard VQ tokens.
|
| 47 |
|
|
@@ -137,4 +137,4 @@ Please cite our work if you use our code or discuss our findings in your own res
|
|
| 137 |
journal={arXiv preprint arXiv:2505.05422},
|
| 138 |
year={2025}
|
| 139 |
}
|
| 140 |
-
```
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
base_model:
|
| 3 |
- google/siglip2-so400m-patch16-384
|
| 4 |
- google/siglip2-so400m-patch16-256
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
license: other
|
| 8 |
+
license_name: other
|
| 9 |
+
license_link: https://github.com/TencentARC/TokLIP/blob/main/LICENSE
|
| 10 |
+
pipeline_tag: image-text-to-text
|
| 11 |
tags:
|
| 12 |
- Tokenizer
|
| 13 |
- CLIP
|
| 14 |
- UnifiedMLLM
|
| 15 |
---
|
| 16 |
|
|
|
|
| 17 |
# TokLIP: Marry Visual Tokens to CLIP for Multimodal Comprehension and Generation
|
| 18 |
|
| 19 |
<h5 align="center">
|
| 20 |
|
| 21 |
[](https://arxiv.org/abs/2505.05422)
|
| 22 |
[](https://github.com/TencentARC/TokLIP)
|
| 23 |
+
[](https://huggingface.co/TencentARC/TokLIP)
|
| 24 |
+
[](https://github.com/TencentARC/TokLIP/blob/main/LICENSE)
|
| 25 |
<br>
|
| 26 |
|
| 27 |
</h5>
|
|
|
|
| 41 |
|
| 42 |
## 👀 Introduction
|
| 43 |
|
| 44 |
+
<img src="https://raw.githubusercontent.com/TencentARC/TokLIP/main/docs/TokLIP.png" alt="TokLIP" style="zoom:50%;" />
|
| 45 |
|
| 46 |
- We introduce TokLIP, a visual tokenizer that enhances comprehension by **semanticizing** vector-quantized (VQ) tokens and **incorporating CLIP-level semantics** while enabling end-to-end multimodal autoregressive training with standard VQ tokens.
|
| 47 |
|
|
|
|
| 137 |
journal={arXiv preprint arXiv:2505.05422},
|
| 138 |
year={2025}
|
| 139 |
}
|
| 140 |
+
```
|