Update app.py

app.py

@@ -1,6 +1,7 @@
 from transformers import pipeline
 import gradio as gr
 import json
+import time
 
 # Initialize the pipeline with the new model
 pipe = pipeline("text-generation", model="Blexus/Quble_test_model_v1_INSTRUCT_v1")
@@ -42,15 +43,18 @@ def generate(prompt, system, history, temperature=0.9, max_new_tokens=4096, top_
     if formatted_prompt in database:
         response_text = database[formatted_prompt]
     else:
-        # Use the pipeline to generate the response
+        # Generate the response without streaming
         try:
-            # Stream the response
-            for response in pipe(formatted_prompt, max_new_tokens=max_new_tokens, temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty, return_full_text=False, streaming=True):
-                assistant_response = response["generated_text"].split("ASSISTANT:")[-1].strip()
-                yield assistant_response
+            response = pipe(formatted_prompt, max_new_tokens=max_new_tokens, temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty)[0]["generated_text"]
+            response_text = response.split("ASSISTANT:")[-1].strip()
 
-            # Save the generated response to the database after the stream
-            database[formatted_prompt] = assistant_response
+            # Simulate streaming by yielding parts of the response
+            for char in response_text:
+                yield char
+                time.sleep(0.05)  # Add a slight delay to simulate typing
+
+            # Save the generated response to the database after the response is generated
+            database[formatted_prompt] = response_text
             save_database(database)
         except Exception as e:
             print(f"Error generating response: {e}")