s3nh commited on
Commit
e0b18bc
1 Parent(s): 2d729bc

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +2 -2
app.py CHANGED
@@ -88,10 +88,10 @@ def answer_question(image, question, max_crops, num_tokens, sample, temperature,
88
  with gr.Blocks() as demo:
89
  gr.HTML("<h1 class='gradio-heading'><center>MC-LLaVA 3B</center></h1>")
90
  gr.HTML(
91
- "<center><p class='gradio-sub-heading'>MC-LLaVA 3B is a model that can answer questions about small details in high-resolution images. Check out the <a href='https://huggingface.co/visheratin/MC-LLaVA-3b'>model card</a> for more details. If you have any questions or ideas hot to make the model better, <a href='https://x.com/visheratin'>let me know</a>.</p></center>"
92
  )
93
  gr.HTML(
94
- "<center><p class='gradio-sub-heading'>There are two main parameters - max number of crops and number of image tokens. The first one controls into how many parts will the image be cut. This is especially useful when you are working with high-resolution images. The second parameter controls how many image features will be extracted for LLM to be processed. You can increase it if you are trying to extract info from a small part of the image, e.g., text.</p></center>"
95
  )
96
  with gr.Group():
97
  with gr.Row():
 
88
  with gr.Blocks() as demo:
89
  gr.HTML("<h1 class='gradio-heading'><center>MC-LLaVA 3B</center></h1>")
90
  gr.HTML(
91
+ "<center><p class='gradio-sub-heading'>MC-LLaVA 3B is a model that can answer questions about small details in high-resolution images. </p></center>"
92
  )
93
  gr.HTML(
94
+ "<center><p class='gradio-sub-heading'>The magic of LLM happened when we can combine them with different data sources. We are able to search for object on images and get answer prepared by Large Language Model.</p></center>"
95
  )
96
  with gr.Group():
97
  with gr.Row():