Update app.py

app.py

@@ -9,7 +9,7 @@ model_name = "sarvamai/sarvam-m"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
 
-@spaces.GPU
+@spaces.GPU(duration=120)
 def generate_response(prompt, chat_history):
     
     chat_history.append(dict(role="user", content=prompt ))
@@ -17,8 +17,7 @@ def generate_response(prompt, chat_history):
 
     print(chat_history)
     
-    messages = [{"role": "user", "content": prompt}]
-    text = tokenizer.apply_chat_template(messages, tokenize=False, enable_thinking=True)
+    text = tokenizer.apply_chat_template(chat_history, tokenize=False, enable_thinking=True)
 
     model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
 
@@ -28,9 +27,7 @@ def generate_response(prompt, chat_history):
     # Conduct text generation with streaming
     generation_kwargs = dict(
         input_ids=model_inputs.input_ids,
-        max_new_tokens=8192,
-        do_sample=True,
-        temperature=0.7,
+        max_new_tokens=4096,
         streamer=streamer,
     )
 
@@ -62,9 +59,8 @@ def generate_response(prompt, chat_history):
 # Create the Gradio interface
 with gr.Blocks() as demo:
     gr.Markdown("# Sarvam M Demo")
-    chatbot = gr.Chatbot(height=600, type="messages")
+    chatbot = gr.Chatbot(height=500, type="messages")
     msg = gr.Textbox(label="Your Message")
-
     msg.submit(generate_response, [msg, chatbot], [chatbot])
 
 if __name__ == "__main__":