diff --git a/server/server.py b/server/server.py
index 7fbe833..0544070 100644
--- a/server/server.py
+++ b/server/server.py
@@ -1,8 +1,7 @@
-from .. import autograd_4bit
 import time
 import torch
-from ..autograd_4bit import load_llama_model_4bit_low_ram, Autograd4bitQuantLinear
-from alpaca_lora_4bit.model_attn_mlp_patch import make_quant_attn, make_fused_mlp, inject_lora_layers
+from autograd_4bit import load_llama_model_4bit_low_ram, Autograd4bitQuantLinear
+from model_attn_mlp_patch import make_quant_attn, make_fused_mlp, inject_lora_layers
 import zmq
 from transformers import StoppingCriteria, StoppingCriteriaList
 from io import BytesIO
diff --git a/text-generation-webui/generate_monkey_patch.py b/text-generation-webui/generate_monkey_patch.py
index d3c59b1..3848840 100644
--- a/text-generation-webui/generate_monkey_patch.py
+++ b/text-generation-webui/generate_monkey_patch.py
@@ -1,6 +1,6 @@
 import modules.text_generation
 from modules.text_generation import *
-from alpaca_lora_4bit.server import _SentinelTokenStoppingCriteria
+from server import _SentinelTokenStoppingCriteria
 
 def generate_reply_patched(question, state, eos_token=None, stopping_strings=[]):
     if shared.model_name == 'None' or shared.model is None: