Update app.py

app.py

@@ -5,12 +5,12 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 # Define the BLOOM model name
 model_name = "CreitinGameplays/bloom-3b-conversational"
 
-# Load tokenizer and model
+# Load tokenizer and model (outside the function for efficiency)
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForCausalLM.from_pretrained(model_name)
 
 def generate_text(user_prompt):
-  """Generates text using the BLOOM model from Hugging Face Transformers and removes the user prompt."""
+  """Generates text using the pre-loaded BLOOM model and removes the user prompt."""
   # Construct the full prompt with system introduction, user prompt, and assistant role
   prompt = f"<|system|> You are a helpful AI assistant. </s> <|prompter|> {user_prompt} </s> <|assistant|>"
 
@@ -24,7 +24,7 @@ def generate_text(user_prompt):
       num_beams=1,
       num_return_sequences=1,  
       do_sample=True,  
-      top_k=0,  
+      top_k=50,  
       top_p=0.95,  
       temperature=0.2,  
       repetition_penalty=1.155
@@ -40,7 +40,7 @@ def generate_text(user_prompt):
   
   return assistant_response
 
-# Define the Gradio interface
+# Define the Gradio interface with live=True for real-time updates
 interface = gr.Interface(
   fn=generate_text,
   inputs=[
@@ -48,6 +48,7 @@ interface = gr.Interface(
   ],
   outputs="text",
   description="Interact with BLOOM-3b-conversational (Loaded with Hugging Face Transformers)",
+  live=True
 )
 
 # Launch the Gradio interface