Update app.py

app.py

@@ -36,8 +36,7 @@ def respond(
     max_tokens,
     temperature,
     top_p,
-    top_k,
-    model,
+    top_k
 ):
     chat_template = get_messages_formatter_type(model)
 
@@ -57,8 +56,6 @@ def respond(
     settings.top_p = top_p
     settings.max_tokens = max_tokens
     settings.stream = True
-    settings.num_beams = 10
-    settings.num_return_sequences=10
 
     messages = BasicChatHistory()
 
@@ -82,22 +79,10 @@ def respond(
         print_output=False
     )
 
-    outputs = set()  # Use a set to store unique outputs
-    unique_responses = []
-    prompt_length = len(message)  # Assuming `message` is the prompt
-
-    for index, output in enumerate(stream, start=1):
-        if output not in outputs:
-            outputs.add(output)
-
-            # Post-process the output
-            output1 = output[prompt_length:]
-            first_inst_index = output1.find("[/INST]")
-            second_inst_index = output1.find("[/IN", first_inst_index + len("[/INST]") + 1)
-            predicted_selfies = output1[first_inst_index + len("[/INST]") : second_inst_index].strip()
-            predicted_smiles = sf.decoder(predicted_selfies)
-            unique_responses.append(f"Predict {index}: {predicted_smiles}")
-            yield "\n".join(unique_responses)
+    outputs = ""
+        for output in stream:
+            outputs += output
+            yield outputs
 
 
 PLACEHOLDER = """