|
--- |
|
license: cc-by-nc-4.0 |
|
datasets: |
|
- zhongshsh/CLoT-Oogiri-GO |
|
language: |
|
- zh |
|
--- |
|
### FaDianInternVL2-2B |
|
|
|
This is a repo for FaDianInternVL2-2B model, it finetune with dataset CLoT-Oogiri-GO. |
|
|
|
We split all the I2T datasets and filter, the final datasets is 24000+. Then we split it to 3000 images and QLoRA it. |
|
|
|
### before finetune |
|
|
|
![Image before](https://huggingface.co/SmartFlowAI/FaDianInternVL2-2B/resolve/main/31911e5a-5800-4871-9a7c-fe494cda33e7.jpeg) |
|
|
|
### after finetune |
|
|
|
![Image after](https://huggingface.co/SmartFlowAI/FaDianInternVL2-2B/resolve/main/a5c101f5-3a4d-4f49-9b9f-d03cc8290ba5.jpeg) |
|
|
|
### others |
|
|
|
first |
|
![Image after](https://huggingface.co/SmartFlowAI/FaDianInternVL2-2B/resolve/main/38e95a07-3bf2-4769-91e5-61c540ddbf0f.jpeg) |
|
|
|
second |
|
![Image after](https://huggingface.co/SmartFlowAI/FaDianInternVL2-2B/resolve/main/b6002e9d-dcdd-4ae0-90c8-d5e0cb1271aa.jpeg) |