Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,40 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
task_categories:
|
4 |
+
- token-classification
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
tags:
|
8 |
+
- TAM
|
9 |
+
- CAM
|
10 |
+
- MLLM
|
11 |
+
- VLLM
|
12 |
+
- Explainability
|
13 |
+
pretty_name: TAM
|
14 |
+
size_categories:
|
15 |
+
- 1B<n<10B
|
16 |
+
---
|
17 |
+
|
18 |
+
# Token Activation Map to Visually Explain Multimodal LLMs
|
19 |
+
We introduce the Token Activation Map (TAM), a groundbreaking method that cuts through the contextual noise in Multimodal LLMs. This technique produces exceptionally clear and reliable visualizations, revealing the precise visual evidence behind every word the model generates.
|
20 |
+
|
21 |
+
# Evaluation Datasets
|
22 |
+
This is a dataset repo to evaluate TAM. The involved datasets are formatted for easy useage.
|
23 |
+
|
24 |
+
# Paper and Code
|
25 |
+
[](https://arxiv.org/abs/2506.23270)
|
26 |
+
|
27 |
+
[🐙 GitHub Page](https://github.com/xmed-lab/TAM)
|
28 |
+
|
29 |
+
## Citation
|
30 |
+
```
|
31 |
+
@misc{li2025tokenactivationmapvisually,
|
32 |
+
title={Token Activation Map to Visually Explain Multimodal LLMs},
|
33 |
+
author={Yi Li and Hualiang Wang and Xinpeng Ding and Haonan Wang and Xiaomeng Li},
|
34 |
+
year={2025},
|
35 |
+
eprint={2506.23270},
|
36 |
+
archivePrefix={arXiv},
|
37 |
+
primaryClass={cs.CV},
|
38 |
+
url={https://arxiv.org/abs/2506.23270},
|
39 |
+
}
|
40 |
+
```
|