mratanusarkar commited on
Commit
ff75fe0
·
1 Parent(s): 2691833

bugfix: prevent model load on every extraction

Browse files
medrag_multi_modal/document_loader/image_loader/marker_img_loader.py CHANGED
@@ -50,6 +50,7 @@ class MarkerImageLoader(BaseImageLoader):
50
 
51
  def __init__(self, url: str, document_name: str, document_file_path: str):
52
  super().__init__(url, document_name, document_file_path)
 
53
 
54
  async def extract_page_data(
55
  self, page_idx: int, image_save_dir: str, **kwargs
@@ -72,11 +73,9 @@ class MarkerImageLoader(BaseImageLoader):
72
  - "file_url": (str) the URL of the PDF file.
73
  - "image_file_path": (str) the local file path where the image is stored.
74
  """
75
- model_lst = load_all_models()
76
-
77
  _, images, out_meta = convert_single_pdf(
78
  self.document_file_path,
79
- model_lst,
80
  max_pages=1,
81
  batch_multiplier=1,
82
  start_page=page_idx,
 
50
 
51
  def __init__(self, url: str, document_name: str, document_file_path: str):
52
  super().__init__(url, document_name, document_file_path)
53
+ self.model_lst = load_all_models()
54
 
55
  async def extract_page_data(
56
  self, page_idx: int, image_save_dir: str, **kwargs
 
73
  - "file_url": (str) the URL of the PDF file.
74
  - "image_file_path": (str) the local file path where the image is stored.
75
  """
 
 
76
  _, images, out_meta = convert_single_pdf(
77
  self.document_file_path,
78
+ self.model_lst,
79
  max_pages=1,
80
  batch_multiplier=1,
81
  start_page=page_idx,