Update app.py

app.py

@@ -18,12 +18,11 @@ quantization_config = BitsAndBytesConfig(
 )
 
 model = AutoModelForCausalLM.from_pretrained(
-    "meta-llama/Meta-Llama-3-8B-Instruct", quantization_config=quantization_config, token=token
+    "microsoft/Phi-3-mini-128k-instruct", quantization_config=quantization_config, token=token
 )
-tok = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct", token=token)
+tok = AutoTokenizer.from_pretrained("microsoft/Phi-3-mini-128k-instruct", token=token)
 terminators = [
     tok.eos_token_id,
-    tok.convert_tokens_to_ids("<|eot_id|>")
 ]
 
 if torch.cuda.is_available():
@@ -37,7 +36,7 @@ else:
 # Dispatch Errors
 
 
-@spaces.GPU(duration=150)
+@spaces.GPU(duration=60)
 def chat(message, history, temperature,do_sample, max_tokens):
     start_time = time.time()
     chat = []
@@ -49,7 +48,7 @@ def chat(message, history, temperature,do_sample, max_tokens):
     messages = tok.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
     model_inputs = tok([messages], return_tensors="pt").to(device)
     streamer = TextIteratorStreamer(
-        tok, timeout=10.0, skip_prompt=True, skip_special_tokens=True
+        tok, timeout=20.0, skip_prompt=True, skip_special_tokens=True
     )
     generate_kwargs = dict(
         model_inputs,
@@ -105,6 +104,6 @@ demo = gr.ChatInterface(
     ],
     stop_btn="Stop Generation",
     title="Chat With LLMs",
-    description="Now Running [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) in 4bit"
+    description="Now Running [microsoft/Phi-3-mini-128k-instruct](https://huggingface.com/microsoft/Phi-3-mini-128k-instruct) in 4bit"
 )
 demo.launch()