Update app.py

app.py

@@ -3,25 +3,23 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 import gradio as gr
 import concurrent.futures
 
-# Set device
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+device = torch.device("cpu")
+
 
-# Load models and tokenizers
 def load_model(name):
     tokenizer = AutoTokenizer.from_pretrained(name)
     model = AutoModelForCausalLM.from_pretrained(name)
     
-    # Define pad token explicitly
+    
     tokenizer.pad_token = tokenizer.eos_token
     model.config.pad_token_id = tokenizer.pad_token_id
     
     return tokenizer, model.to(device)
 
 tokenizer1, model1 = load_model("Gensyn/Qwen2.5-0.5B-Instruct")
-tokenizer2, model2 = load_model("deepset/roberta-base-squad2")
+tokenizer2, model2 = load_model("google/flan-t5-small")
 tokenizer3, model3 = load_model("microsoft/phi-1_5")
 
-# Generation function
 def generate_response(model, tokenizer, prompt):
     inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True).to(device)
     outputs = model.generate(
@@ -35,7 +33,7 @@ def generate_response(model, tokenizer, prompt):
     )
     return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
-# Multi-agent handler
+
 def multi_agent_chat(user_input):
     with concurrent.futures.ThreadPoolExecutor() as executor:
         futures = [
@@ -46,13 +44,13 @@ def multi_agent_chat(user_input):
         results = [f.result() for f in futures]
     return results
 
-# Gradio Interface
+
 interface = gr.Interface(
     fn=multi_agent_chat,
     inputs=gr.Textbox(lines=2, placeholder="Ask something..."),
     outputs=[
         gr.Textbox(label="Agent 1 (Gensyn/Qwen2.5-0.5B-Instruct)"),
-        gr.Textbox(label="Agent 2 (deepset/roberta-base-squad2)"),
+        gr.Textbox(label="Agent 2 (google/flan-t5-small)"),
         gr.Textbox(label="Agent 3 (microsoft/phi-1_5)")
     ],
     title="3-Agent AI Chatbot",