Update app.py

app.py

@@ -20,13 +20,13 @@ LICENSE = """
 <p/>
 ---
 Chat with Buzz-small!
-only 3b,  this demo runs on the fp16 weights of the model in safetensors format (converting to cpp soon)
+only 3b,  this demo runs on the fp8 weights of the model in pytorch format, its brains are probably significantly damaged, converting to cpp soon, dont worry!
 """
 
 device = 0 if torch.cuda.is_available() else -1
 
 model_id = "H-D-T/Buzz-3b-small-v0.6.3"
-chatbot = pipeline(model=model_id, device=device, task="conversational")
+chatbot = pipeline(model=model_id, device=device, task="conversationa",model_kwargs={"load_in_8bit": True})
 
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 bos_token = "<|begin_of_text|>"