Update app.py

app.py

@@ -7,13 +7,13 @@ import sentencepiece
 
 model_id = "01-ai/Yi-34B-200K"
 
-os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:54'
-device = "cuda" if torch.cuda.is_available() else "cpu"
+# os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:54'
+# device = "cuda" if torch.cuda.is_available() else "cpu"
 
 tokenizer = YiTokenizer(vocab_file="./tokenizer.model")
-model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
-model = model.to(torch.bfloat16)
-model = model.to(device)
+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)
 
 def run(message, chat_history, max_new_tokens=4056, temperature=3.5, top_p=0.9, top_k=800):
     prompt = get_prompt(message, chat_history)