Update app.py

app.py

@@ -18,6 +18,7 @@ tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
 #streamlit stuff
 tags = st.text_input("Input tags separated by commas")
 text = st.text_input("Input text to classify")
+ = st.slider("Number of key phrases to extract from text", 10,30,20)
 #Methods for tag processing
 def pool_embeddings(out, tok):
   embeddings = out["hidden_states"][-1]
@@ -116,7 +117,7 @@ def extract_tokens(text,top_kp=30):
 #Process text and classify it
 if text and tags:
   text = text.lower()
-  t1_tokens = extract_tokens(text)
+  t1_tokens = extract_tokens(text, topkp)
   t1_kps = t1_tokens.pop("KPS")
   with torch.no_grad():
     outputs = model(**t1_tokens)