Tuana commited on
Commit
2a5639d
1 Parent(s): 9c1fb8f

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +2 -1
app.py CHANGED
@@ -14,7 +14,7 @@ def start_haystack():
14
  split_by="word",
15
  split_length=100,
16
  split_respect_sentence_boundary=True,
17
- split_overlap=3
18
  )
19
  summarizer = TransformersSummarizer(model_name_or_path="google/pegasus-xsum")
20
  return document_store, summarizer, preprocessor
@@ -29,6 +29,7 @@ def pdf_to_document_store(pdf_files):
29
  documents.append(doc)
30
  st.write(len(documents))
31
  preprocessed_docs = preprocessor.process(documents)
 
32
  document_store.write_documents(preprocessed_docs)
33
  st.write('Document count: ', document_store.get_document_count())
34
 
 
14
  split_by="word",
15
  split_length=100,
16
  split_respect_sentence_boundary=True,
17
+ split_overlap=0
18
  )
19
  summarizer = TransformersSummarizer(model_name_or_path="google/pegasus-xsum")
20
  return document_store, summarizer, preprocessor
 
29
  documents.append(doc)
30
  st.write(len(documents))
31
  preprocessed_docs = preprocessor.process(documents)
32
+ st.write('Preprocessed count: ', len(preprocessed_docs))
33
  document_store.write_documents(preprocessed_docs)
34
  st.write('Document count: ', document_store.get_document_count())
35