Update app.py

app.py

@@ -42,7 +42,7 @@ def extend(input_text, max_size=20, top_k=50, top_p=0.95):
     top_k=top_k, 
     top_p=top_p, 
     do_sample=True,
-    repetition_penalty=25.0,
+    repetition_penalty=2.5,
     num_return_sequences=1)
 
     # Remove the batch dimension when returning multiple sequences
@@ -99,7 +99,7 @@ if __name__ == "__main__":
 
     st.sidebar.subheader("Configurable parameters")
 
-    max_len = st.sidebar.slider("Max-Length", 0, 256, 192,help="The maximum length of the sequence to be generated.")
+    max_len = st.sidebar.slider("Max-Length", 0, 128, 192,help="The maximum length of the sequence to be generated.")
     top_k = st.sidebar.slider("Top-K", 0, 100, 40, help="The number of highest probability vocabulary tokens to keep for top-k-filtering.")
     top_p = st.sidebar.slider("Top-P", 0.0, 1.0, 0.92, help="If set to float < 1, only the most probable tokens with probabilities that add up to top_p or higher are kept for generation.")