|
|
--- |
|
|
base_model: |
|
|
- unsloth/Qwen3-VL-2B-Instruct |
|
|
tags: |
|
|
- text-generation-inference |
|
|
- transformers |
|
|
- unsloth |
|
|
- qwen3_vl |
|
|
- trl |
|
|
- sft |
|
|
license: apache-2.0 |
|
|
language: |
|
|
- en |
|
|
datasets: |
|
|
- lingcco/FakeClue |
|
|
pipeline_tag: visual-question-answering |
|
|
--- |
|
|
|
|
|
# Uploaded model |
|
|
|
|
|
- **Developed by:** soorism |
|
|
- **License:** apache-2.0 |
|
|
|
|
|
This is a "Qwen3-VL-2B-instruct" model finetuned on the training split of the FakeClue dataset to compare with FakeVLM. Here are the results on the test set: |
|
|
|
|
|
Accuracy: 92.36% |
|
|
|
|
|
| Class | Precision | Recall | F1-Score | |
|
|
|:------|:-----------|:--------|:----------| |
|
|
| Fake | 0.9225 | 0.9624 | 0.9421 | |
|
|
| Real | 0.9259 | 0.8532 | 0.8881 | |
