Update app.py

app.py

@@ -8,10 +8,10 @@ import sentencepiece
 model_id = "01-ai/Yi-34B-200K"
 
 os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:50'
-# device = "cuda" if torch.cuda.is_available() else "cpu"
+device = "cuda" if torch.cuda.is_available() else "cpu"
 
 tokenizer = YiTokenizer(vocab_file="./tokenizer.model")
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="cuda", load_in_8bit=True, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", load_in_8bit=True, trust_remote_code=True)
 # model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
 # model = model.to(device)