quoc-khanh commited on
Commit
df051d1
·
verified ·
1 Parent(s): bf72bbf

Update helpers.py

Browse files
Files changed (1) hide show
  1. helpers.py +2 -1
helpers.py CHANGED
@@ -153,7 +153,8 @@ def load_text_data(file_path):
153
  loader = DoclingLoader(
154
  file_path=file_path,
155
  export_type=ExportType.MARKDOWN, # Enable internal chunking,
156
- chunker = text_splitter
 
157
  )
158
  return loader.load()
159
 
 
153
  loader = DoclingLoader(
154
  file_path=file_path,
155
  export_type=ExportType.MARKDOWN, # Enable internal chunking,
156
+ chunker = text_splitter,
157
+ convert_kwargs={"input_format": "docx"} # Specify the input format
158
  )
159
  return loader.load()
160