update readme
Browse files
README.md
CHANGED
@@ -1,3 +1,31 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
+
pipeline_tag: image-to-text
|
4 |
---
|
5 |
+
|
6 |
+
# MMAlaya2
|
7 |
+
|
8 |
+
MMAlaya2 fine-tunes 20 LoRA modules based on the InternVL-Chat-V1-5 model. These fine-tuned LoRA modules are then merged with the InternVL-Chat-V1-5 model using the PEFT model merging method, TIES.
|
9 |
+
|
10 |
+
You can find the inference code [here](https://github.com/open-compass/VLMEvalKit/blob/main/vlmeval/vlm/mmalaya.py#L8).
|
11 |
+
|
12 |
+
The [MMBench](https://mmbench.opencompass.org.cn/) benchmark contains 20 categories in the `mmbench_dev_cn_20231003.tsv` dataset. For each category, we first use CoT (Chain of Thought) consistency with the InternVL-Chat-V1-5 model to prepare the training dataset. For specific categories like nature_relation, image_emotion, image_scene, action_recognition, and image_style, we analyze the bad cases made by the InternVL-Chat-V1-5 model. We then prepare images and QA text from online sources to address these issues.
|
13 |
+
|
14 |
+
After fine-tuning the 20 LoRAs, they are merged with the InternVL-Chat-V1-5 model using the TIES method. The average score on the `mmbench_test_cn_20231003.tsv` benchmark reached 82.2, which we found noteworthy. As a result, we are sharing this model publicly.
|
15 |
+
|
16 |
+
# License
|
17 |
+
|
18 |
+
This project is released under the MIT license, in alignment with the InternVL-Chat-V1-5 model's license. InternLM2, however, is licensed under the Apache-2.0 license.
|
19 |
+
|
20 |
+
# Citation
|
21 |
+
|
22 |
+
If you find this project useful in your research, please consider citing:
|
23 |
+
|
24 |
+
```bibtex
|
25 |
+
@misc{datacanvas2024mmalaya2,
|
26 |
+
author = {DataCanvas Ltd.},
|
27 |
+
title = {MMAlaya2},
|
28 |
+
year = {2024},
|
29 |
+
howpublished = {\url{https://huggingface.co/DataCanvas/MMAlaya2}},
|
30 |
+
}
|
31 |
+
```
|