Add model card
#1
by nielsr HF Staff - opened
README.md
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pipeline_tag: image-feature-extraction
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# MixTok
|
| 6 |
+
|
| 7 |
+
**Adaptive Visual Tokenization with Global 1D Tokens and Sparse 2D Refinement**
|
| 8 |
+
|
| 9 |
+
MixTok is a hybrid visual tokenizer that combines a compact 1D global representation with sparse 2D tokens for local refinement. A content-adaptive router selects which spatial locations to refine and how many 2D tokens to use per image. On ImageNet 256×256, MixTok achieves highly competitive reconstruction and generation performance with about 128–131 tokens per image.
|
| 10 |
+
|
| 11 |
+
- Paper: [MixTok: Adaptive Visual Tokenization with Global 1D Tokens and Sparse 2D Refinement](https://huggingface.co/papers/2610.13181)
|
| 12 |
+
- Project page: [https://lcglichenge.github.io/MixTok-page/](https://lcglichenge.github.io/MixTok-page/)
|
| 13 |
+
- Code: [https://github.com/LCGLiChenge/MixTok](https://github.com/LCGLiChenge/MixTok)
|