Spaces:
Running
on
Zero
Running
on
Zero
update
Browse files
app.py
CHANGED
|
@@ -70,6 +70,10 @@ MARKDOWN = """
|
|
| 70 |
\[[arXiv Paper](https://www.arxiv.org/pdf/2502.13130)\] \[[Project Page](https://microsoft.github.io/Magma/)\] \[[Github Repo](https://github.com/microsoft/Magma)\] \[[Hugging Face Model](https://huggingface.co/microsoft/Magma-8B)\]
|
| 71 |
|
| 72 |
This demo is powered by [Gradio](https://gradio.app/) and uses [OmniParserv2](https://github.com/microsoft/OmniParser) to generate [Set-of-Mark prompts](https://github.com/microsoft/SoM).
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
</div>
|
| 74 |
"""
|
| 75 |
|
|
|
|
| 70 |
\[[arXiv Paper](https://www.arxiv.org/pdf/2502.13130)\] \[[Project Page](https://microsoft.github.io/Magma/)\] \[[Github Repo](https://github.com/microsoft/Magma)\] \[[Hugging Face Model](https://huggingface.co/microsoft/Magma-8B)\]
|
| 71 |
|
| 72 |
This demo is powered by [Gradio](https://gradio.app/) and uses [OmniParserv2](https://github.com/microsoft/OmniParser) to generate [Set-of-Mark prompts](https://github.com/microsoft/SoM).
|
| 73 |
+
|
| 74 |
+
The model can do two things simutaneously:
|
| 75 |
+
1. **UI Grounding and Navigation**: Ask the model to generate a click action.
|
| 76 |
+
2. **Visual Question Answering**: Ask the model with any verbal questions by appending a prefix "Q:" to the question.
|
| 77 |
</div>
|
| 78 |
"""
|
| 79 |
|