Update app.py

app.py

@@ -55,13 +55,20 @@ def generate_response(transcription):
         with torch.no_grad():
             output = model.generate(
                 input_ids,
-                max_new_tokens=100,
-                num_return_sequences=1,
-                temperature=0.1,
-                top_k=50,
-                top_p=0.5,
-                repetition_penalty=1.2,
-                do_sample=True
+                # max_new_tokens=100,
+                # num_return_sequences=1,
+                # temperature=0.1,
+                # top_k=50,
+                # top_p=0.5,
+                # repetition_penalty=1.2,
+                # do_sample=True
+                max_new_tokens=300,
+                do_sample=True,
+                top_k=5,
+                num_beams=1,
+                use_cache=False,
+                temperature=0.2,
+                repetition_penalty=1.1,
             )
         generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
         return find_last_sentence(generated_text)