Josebert commited on
Commit
3d27426
·
verified ·
1 Parent(s): fb37c01

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +13 -7
app.py CHANGED
@@ -1,9 +1,9 @@
1
  from vllm import LLM
2
  from vllm.sampling_params import SamplingParams
3
  from datetime import datetime, timedelta
4
- from huggingface_hub import hf_hub_download, login
5
- import os
6
- import gradio as gr
7
  from pdf2image import convert_from_path
8
  import easyocr
9
 
@@ -12,7 +12,7 @@ reader = easyocr.Reader(['en'])
12
 
13
  # ... existing SYSTEM_PROMPT and load_system_prompt definitions ...
14
 
15
- def process_pdf_or_image(file_path):
16
  # Handle PDF files
17
  if file_path.lower().endswith('.pdf'):
18
  images = convert_from_path(file_path)
@@ -28,8 +28,8 @@ def process_pdf_or_image(file_path):
28
  return extracted_text
29
 
30
  def generate_response(file_path):
31
- # Extract text from PDF/image
32
- extracted_text = process_pdf_or_image(file_path)
33
 
34
  # Prepare messages for the LLM
35
  messages = [
@@ -46,7 +46,7 @@ def generate_response(file_path):
46
  ]
47
 
48
  # Initialize the LLM
49
- llm = LLM(model="mistralai/Mistral-Small-3.1-24B", tokenizer_mode="mistral")
50
 
51
  # Define sampling parameters
52
  sampling_params = SamplingParams(max_tokens=512, temperature=0.15)
@@ -56,6 +56,12 @@ def generate_response(file_path):
56
 
57
  return outputs[0].outputs[0].text
58
 
 
 
 
 
 
 
59
  # Gradio interface
60
  with gr.Blocks() as demo:
61
  gr.Markdown("# Document Processing with Mistral")
 
1
  from vllm import LLM
2
  from vllm.sampling_params import SamplingParams
3
  from datetime import datetime, timedelta
4
+ from huggingface_hub import hf_hub_download
5
+ import requests
6
+ import json
7
  from pdf2image import convert_from_path
8
  import easyocr
9
 
 
12
 
13
  # ... existing SYSTEM_PROMPT and load_system_prompt definitions ...
14
 
15
+ def process_document(file_path):
16
  # Handle PDF files
17
  if file_path.lower().endswith('.pdf'):
18
  images = convert_from_path(file_path)
 
28
  return extracted_text
29
 
30
  def generate_response(file_path):
31
+ # Extract text from document
32
+ extracted_text = process_document(file_path)
33
 
34
  # Prepare messages for the LLM
35
  messages = [
 
46
  ]
47
 
48
  # Initialize the LLM
49
+ llm = LLM(model="mistralai/Mistral-Small-3.1-24B-Instruct-2503", tokenizer_mode="mistral")
50
 
51
  # Define sampling parameters
52
  sampling_params = SamplingParams(max_tokens=512, temperature=0.15)
 
56
 
57
  return outputs[0].outputs[0].text
58
 
59
+ # Example usage
60
+ if __name__ == "__main__":
61
+ document_path = "path/to/your/document.pdf" # or .jpg/.png
62
+ response = generate_response(document_path)
63
+ print(response)
64
+
65
  # Gradio interface
66
  with gr.Blocks() as demo:
67
  gr.Markdown("# Document Processing with Mistral")