Tuana commited on
Commit
fe7b517
1 Parent(s): 67f4a7d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -1
app.py CHANGED
@@ -34,7 +34,6 @@ def pdf_to_document_store(pdf_file):
34
 
35
  def summarize(file):
36
  pdf_to_document_store(file)
37
- st.write('Number of documents', document_store.get_document_count())
38
  summaries = summarizer.predict(documents=document_store.get_all_documents(), generate_single_summary=True)
39
  return summaries
40
 
@@ -45,6 +44,10 @@ def set_state_if_absent(key, value):
45
  set_state_if_absent("summaries", None)
46
  document_store, summarizer, preprocessor = start_haystack()
47
 
 
 
 
 
48
  uploaded_file = st.file_uploader("Choose a PDF file", accept_multiple_files=False)
49
 
50
  if uploaded_file is not None:
 
34
 
35
  def summarize(file):
36
  pdf_to_document_store(file)
 
37
  summaries = summarizer.predict(documents=document_store.get_all_documents(), generate_single_summary=True)
38
  return summaries
39
 
 
44
  set_state_if_absent("summaries", None)
45
  document_store, summarizer, preprocessor = start_haystack()
46
 
47
+ st.markdown( """
48
+ This Summarization demo uses a [Haystack TransformerSummarizer node](https://haystack.deepset.ai/pipeline_nodes/summarizer). You can upload a PDF file, which will be converted to text with the [Haystack PDFtoTextConverter](https://haystack.deepset.ai/reference/file-converters#pdftotextconverter). In this demo, we produce 1 summary for the whole file you upload. So, the TransformerSummarizer treats the while thing as one string, which means along with the model limitations, PDFs that have a lot of unneeded text at the beginning produce poor results.
49
+ """, unsafe_allow_html=True)
50
+
51
  uploaded_file = st.file_uploader("Choose a PDF file", accept_multiple_files=False)
52
 
53
  if uploaded_file is not None: