Update app.py

app.py

@@ -69,7 +69,7 @@ tokenizer.pad_token = tokenizer.eos_token
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_ID,
     device_map="auto",
-    quantization_config=quantization_config,
+    torch_dtype='auto',
     attn_implementation="flash_attention_2",
 )
 
@@ -144,7 +144,8 @@ def predict(message, history, system_prompt, temperature, max_new_tokens, top_k,
         'repetition_penalty':repetition_penalty, 
         'top_p':top_p, 
         'instruction':instruction, 
-        'output':"".join(outputs)
+        'output':"".join(outputs),
+        'precision': 'auto '+str(model.dtype),
     }
     hfapi.upload_file(
         path_or_fileobj=json.dumps(dd, indent=2, ensure_ascii=False).encode('utf-8'),