Update app.py

app.py

@@ -52,10 +52,6 @@ def generate_prompt(instruction, input=None):
     else:
         return PROMPT_DICT["prompt_no_input"].format(instruction=instruction)
 
-
-def getIds(inputs):
-    return inputs["input_ids"].cuda()
-
 def generator(input_ids, generation_config, max_new_tokens):
     # Without streaming
     with torch.no_grad():
@@ -79,17 +75,16 @@ def loadModel():
                     cache_dir=cache_dir,
                     device_map="auto",
                 )
-        model = PeftModel.from_pretrained(
-                    model,
-                    lora_weights,
-                    device_map="auto",
-                    cache_dir='',
-                    torch_dtype=torch.float16,
-                    is_trainable=False,
-                )
+        #model = PeftModel.from_pretrained(
+        #            model,
+        #            lora_weights,
+        #            device_map="auto",
+        #            cache_dir='',
+        #            torch_dtype=torch.float16,
+        #            is_trainable=False,
+        #        )
         tokenizer =  AutoTokenizer.from_pretrained(base_model,use_fast=False,cache_dir=cache_dir)
         tokenizer.pad_token = tokenizer.unk_token
-        model = model.to("cuda")
     return model
 
 model, tokenizer = loadModel()