Update app.py

app.py

@@ -1,101 +1,45 @@
 import gradio as gr
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
 import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
 
-# ------------------ MODEL SETUP ------------------
-MODEL_NAME = "microsoft/Phi-3-mini-4k-instruct"
+model_name = "facebook/blenderbot-400M-distill"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 
-tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
-model = AutoModelForCausalLM.from_pretrained(
-    MODEL_NAME,
-    torch_dtype=torch.float32,
-    device_map="cpu"
-)
-model.eval()
+history = []
 
-SYSTEM_PROMPT = """You are a calm, intelligent assistant.
-Respond ONLY to the user's last message.
-Do NOT assume intent or emotions.
-If the message is unclear, ask for clarification.
-Keep responses natural and concise.
-"""
-
-# ------------------ CHAT FUNCTIONS ------------------
-
-def build_prompt(history):
-    prompt = SYSTEM_PROMPT + "\n\n"
-    for msg in history:
-        role = "User" if msg["role"] == "user" else "Assistant"
-        prompt += f"{role}: {msg['content']}\n"
-    prompt += "Assistant:"
-    return prompt
-
-def respond(message, history):
-    # Add user message
+def respond(message):
+    global history
     history.append({"role": "user", "content": message})
-    
-    # Add loading placeholder
-    history.append({"role": "assistant", "content": "Thinking..."})
-    yield history, history
-
-    # Build prompt for model
-    prompt = build_prompt(history[:-1])
-
-    inputs = tokenizer(
-        prompt,
-        return_tensors="pt",
-        truncation=True,
-        max_length=2048
-    )
-
-    # Generate response
-    with torch.no_grad():
-        output = model.generate(
-            **inputs,
-            max_new_tokens=120,
-            do_sample=False,
-            temperature=0.0,
-            top_p=1.0,
-            repetition_penalty=1.1
-        )
-
-    text = tokenizer.decode(output[0], skip_special_tokens=True)
-    # Extract model response
-    response = text.split("Assistant:")[-1].strip()
-
-    # Replace loading with actual response
-    history[-1]["content"] = response
-
-    # Trim history if too long
-    if len(history) > 10:
-        history = history[-8:]
-
-    yield history, history
+    last_msgs = history[-3:]
+    input_text = " ".join([m["content"] for m in last_msgs])
+    inputs = tokenizer(input_text, return_tensors="pt")
+    outputs = model.generate(**inputs, max_new_tokens=100)
+    response_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    history.append({"role": "assistant", "content": response_text})
+    return history
 
 def reset_chat():
-    return [], []
-
-# ------------------ GRADIO UI ------------------
-
-with gr.Blocks(css="""
-body { background-color: #000; color: #fff; }
-.gr-chatbot { background-color: #111; border-radius: 12px; }
-.gr-chatbot .message.user { border-color: #0ff; background-color: transparent; }
-.gr-chatbot .message.bot { border-color: #aaa; background-color: transparent; }
-.gr-textbox textarea { background-color: transparent; color: #fff; border: 1px solid #555; }
-.gr-textbox textarea::selection { background-color: #0ff; color: #000; }
-.gr-button { background-color: #0ff; color: #000; border-radius: 8px; }
-footer { display: none; }
-""") as demo:
+    global history
+    history = []
+    return history
+
+css = """
+body {background-color: #000 !important; color: #fff !important;}
+.gr-chatbot {background-color: #111 !important; border-radius: 12px; height: 100% !important;}
+.gr-chatbot .message.user {border-color: #0ff; background-color: transparent !important;}
+.gr-chatbot .message.bot {border-color: #aaa; background-color: transparent !important;}
+.gr-textbox textarea {background-color: transparent !important; color: #fff !important; border: 1px solid #555 !important;}
+.gr-textbox textarea::selection {background-color: #0ff !important; color: #000 !important;}
+.gr-button {background-color: #0ff !important; color: #000 !important; border-radius: 8px;}
+footer {display: none !important;}
+"""
 
-    state = gr.State([])
+with gr.Blocks(css=css) as demo:
     chatbot = gr.Chatbot(label="DevMegaBlack")
     msg = gr.Textbox(placeholder="Say something...")
-    reset = gr.Button("Reset Chat")
-
-    # Submit message
-    msg.submit(respond, [msg, state], [chatbot, state]).then(lambda: "", None, msg)
-    reset.click(reset_chat, [], [chatbot, state])
+    reset_btn = gr.Button("Reset Chat")
+    msg.submit(respond, msg, chatbot)
+    reset_btn.click(reset_chat, [], chatbot)
 
-demo.queue()
 demo.launch()