Update app.py

app.py

@@ -59,8 +59,8 @@ tokenizer = AutoTokenizer.from_pretrained(
 
 model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
                                                  # torch_dtype= torch.uint8, 
-                                             torch_dtype=torch.float16,
-                                              load_in_16bit=True,
+                                             torch_dtype=torch.float32,
+                                              # load_in_16bit=True,
                                              # #  # torch_dtype=torch.fl,
                                              # attn_implementation="flash_attention_2",
                                              # low_cpu_mem_usage=True,