Update app.py

app.py

@@ -1,36 +1,48 @@
-import torch
-from transformers import GPT2Tokenizer, GPT2LMHeadModel
+
+import os
 import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-# Load the custom model and tokenizer
-model_path = 'redael/model_udc'
-tokenizer = GPT2Tokenizer.from_pretrained(model_path)
-model = GPT2LMHeadModel.from_pretrained(model_path)
+# Load your model and tokenizer from Hugging Face
+model_name = 'redael/model_udc'
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
 
-# Check if CUDA is available and use GPU if possible, enable FP16 precision
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 model.to(device)
-if device.type == 'cuda':
-    model = model.half()  # Use FP16 precision
 
-def generate_response(prompt, model, tokenizer, max_length=100, num_beams=1, temperature=0.7, top_p=0.9, repetition_penalty=2.0):
-    # Prepare the prompt
-    prompt = f"User: {prompt}\nAssistant:"
+# Function to generate response
+def generate_response(message, history, system_message, max_tokens, temperature, top_p):
+    # Prepare the conversation history
+    messages = [{"role": "system", "content": system_message}]
+    
+    for user_msg, bot_msg in history:
+        if user_msg:
+            messages.append({"role": "user", "content": user_msg})
+        if bot_msg:
+            messages.append({"role": "assistant", "content": bot_msg})
+    
+    messages.append({"role": "user", "content": message})
+    
+    # Tokenize and prepare the input
+    prompt = "\n".join([f"{msg['role'].capitalize()}: {msg['content']}" for msg in messages])
     inputs = tokenizer(prompt, return_tensors='pt', padding=True, truncation=True, max_length=512).to(device)
+    
+    # Generate the response
     outputs = model.generate(
         inputs['input_ids'],
-        max_length=max_length,
+        max_length=max_tokens,
         num_return_sequences=1,
         pad_token_id=tokenizer.eos_token_id,
-        num_beams=num_beams,
         temperature=temperature,
         top_p=top_p,
-        repetition_penalty=repetition_penalty,
-        early_stopping=True
+        early_stopping=True,
+        do_sample=True  # Enable sampling
     )
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
     
-    # Post-processing to clean up the response
+    # Clean up the response
     response = response.split("Assistant:")[-1].strip()
     response_lines = response.split('\n')
     clean_response = []
@@ -38,29 +50,27 @@ def generate_response(prompt, model, tokenizer, max_length=100, num_beams=1, tem
         if "User:" not in line and "Assistant:" not in line:
             clean_response.append(line)
     response = ' '.join(clean_response)
-    return response.strip()
-
-def respond(message, history):
-    # Prepare the prompt from the history and the new message
-    system_message = "You are a friendly chatbot."
-    conversation = system_message + "\n"
-    for user_message, assistant_response in history:
-        conversation += f"User: {user_message}\nAssistant: {assistant_response}\n"
-    conversation += f"User: {message}\nAssistant:"
-
-    # Fixed values for generation parameters
-    max_tokens = 100  # Adjusted max tokens
-    temperature = 0.7
-    top_p = 0.9
-
-    response = generate_response(conversation, model, tokenizer, max_length=max_tokens, temperature=temperature, top_p=top_p)
     
-    return response
+    return [(message, response)]
 
-# Gradio Chat Interface
+# Create the Gradio chat interface
 demo = gr.ChatInterface(
-    respond
+    fn=generate_response,
+    additional_inputs=[
+        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(
+            minimum=0.1,
+            maximum=1.0,
+            value=0.95,
+            step=0.05,
+            label="Top-p (nucleus sampling)",
+        ),
+    ],
+    title="Chatbot",
+    description="Ask anything to the chatbot."
 )
 
-if _name_ == "_main_":
+if __name__ == "__main__":
     demo.launch()