Update alpacatest.py

alpacatest.py

@@ -9,7 +9,8 @@ if torch.cuda.is_available():
 else:
     device = "cpu"
 
-model = RwkvForCausalLM.from_pretrained("rwkv-alpaca",device_map='auto') #仅500MB,自训练，使用alpaca
+model = RwkvForCausalLM.from_pretrained("rwkv-alpaca") 
+model = model.to(device)
 
 tokenizer = GPTNeoXTokenizerFast.from_pretrained("rwkv-alpaca", add_special_tokens=True)
 
@@ -33,7 +34,7 @@ def evaluate(
     max_new_tokens=128,
 ):
     prompt = generate_prompt(instruction)
-    input_ids = tokenizer.encode(prompt, return_tensors='pt')
+    input_ids = tokenizer.encode(prompt, return_tensors='pt').to(device)
     out = model.generate(input_ids=input_ids,temperature=temperature,top_p=top_p,top_k=top_k,max_new_tokens=max_new_tokens)
     answer = tokenizer.decode(out[0])
     return answer.split("### Response:")[1].strip()