Update app.py

app.py

@@ -1,6 +1,5 @@
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
-from PIL import Image
 import gradio as gr
 
 # Load the model and tokenizer
@@ -15,7 +14,7 @@ tokenizer = AutoTokenizer.from_pretrained("qresearch/llama-3.1-8B-vision-378", u
 # Define the function to process the image and instruction
 def describe_image(image, instruction):
  description = model.answer_question(
- image, instruction, tokenizer, max_new_tokens=128, do_sample=True, temperature=0.3
+ image, instruction, tokenizer, max_new_tokens=1000, do_sample=True, temperature=0.3
  )
  return description
 
@@ -24,12 +23,21 @@ interface = gr.Interface(
  fn=describe_image,
  inputs=[
  gr.Image(type="pil"), # Input for the image
- gr.Textbox(placeholder="Enter your instruction here...", label="Instruction") # Input for the instruction
+ gr.Textbox(
+ placeholder="Enter your instruction here...",
+ label="Instruction",
+ lines=10, # Increase number of lines for instruction input
+ max_lines=20 # Maximum number of lines for scrolling
+ ) 
  ],
- outputs="text", # Output is text (the description)
- title="LLaMA 3.1 with vision",
+ outputs=gr.Textbox(
+ label="Description",
+ lines=10, # Increase number of lines for output
+ max_lines=30 # Maximum number of lines for scrolling
+ ),
+ title="LLaMA 3.1 with Vision",
  description="Upload an image and enter an instruction to generate a description based on the provided instruction."
 )
 
 # Launch the Gradio app
-interface.launch()
+interface.launch()