Update app.py

app.py

@@ -17,7 +17,7 @@ MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "8192"))
 
 model_id = "BramVanroy/fietje-2b-chat"
 avatar_url = "https://huggingface.co/spaces/BramVanroy/fietje-2b/resolve/main/img/fietje-2b-avatar.png"
-model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto")
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", load_in_8bit=True, attn_implementation="flash_attention_2")
 model = model.to("cuda" if torch.cuda.is_available() else "cpu")
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 tokenizer.pad_token_id = tokenizer.eos_token_id
@@ -116,7 +116,7 @@ chat_interface = gr.ChatInterface(
             minimum=0,
             maximum=20,
             step=1,
-            value=8,
+            value=0,
         ),
         gr.Checkbox(
             label="Do sample",