fix

app.py

@@ -68,7 +68,7 @@ def load_model():
     tokenizer.pad_token = tokenizer.eos_token
     tokenizer.padding_side = "right"
     
-    base_model_instance = AutoModel.from_pretrained("orionweller/repllama-instruct-hard-positives-v2-joint-full-weights", max_memory={"cpu": "12GiB"}, torch_dtype=torch.float16, offload_state_dict=True)
+    model = AutoModel.from_pretrained("orionweller/repllama-instruct-hard-positives-v2-joint-full-weights", max_memory={"cpu": "12GiB"}, torch_dtype=torch.float16, offload_state_dict=True)
     # base_model_instance = AutoModel.from_pretrained(BASE_MODEL)
     # model = PeftModel.from_pretrained(base_model_instance, CUR_MODEL)
     model.eval()