Update app.py

app.py

@@ -2,46 +2,32 @@ import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
-# Model name
-model_name = "deepseek-ai/DeepSeek-V3-0324"
+model_id = "tiiuae/falcon-rw-1b"
 
-# Load tokenizer and model in float32 for CPU
-tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    trust_remote_code=True,
-    torch_dtype=torch.float32,  # CPU-compatible precision
-    device_map={"": "cpu"}      # Force CPU
-)
+# Load tokenizer and model for CPU
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float32)
 
-# Chat function
+# Chat logic
 def chat_with_bot(user_input, history):
     history = history or []
     prompt = ""
     for user, bot in history:
-        prompt += f"<|user|>\n{user}\n<|assistant|>\n{bot}\n"
-    prompt += f"<|user|>\n{user_input}\n<|assistant|>\n"
+        prompt += f"{user}\n{bot}\n"
+    prompt += f"{user_input}\n"
 
     inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
-    outputs = model.generate(
-        **inputs,
-        max_new_tokens=256,
-        do_sample=True,
-        temperature=0.7,
-        top_p=0.9,
-        eos_token_id=tokenizer.eos_token_id,
-    )
-
+    outputs = model.generate(**inputs, max_new_tokens=200, do_sample=True)
     decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    response = decoded.split("<|assistant|>\n")[-1].strip()
 
+    response = decoded[len(prompt):].strip()
     history.append((user_input, response))
     return response, history
 
 # Gradio UI
 gr.ChatInterface(
     fn=chat_with_bot,
-    title="DeepSeek CPU Chatbot",
+    title="Chatbot (CPU-Friendly)",
     theme="soft",
-    examples=["Tell me a joke", "What's the capital of Italy?", "What is a black hole?"]
+    examples=["What's Falcon?", "Tell me something about space.", "What is time travel?"]
 ).launch()