diff --git a/scripts/float16.py b/scripts/float16.py
index 0be170f33..647c5df0e 100644
--- a/scripts/float16.py
+++ b/scripts/float16.py
@@ -135,13 +135,13 @@ def convert_tensor_float_to_float16(tensor, min_positive_val=1e-7, max_finite_va
         # convert raw_data (bytes type)
         if tensor.raw_data:
             # convert n.raw_data to float
-            float32_list = np.fromstring(tensor.raw_data, dtype="float32")
+            float32_list = np.frombuffer(tensor.raw_data, dtype="float32")
             # convert float to float16
             float16_list = convert_np_to_float16(
                 float32_list, min_positive_val, max_finite_val
             )
             # convert float16 to bytes and write back to raw_data
-            tensor.raw_data = float16_list.tostring()
+            tensor.raw_data = float16_list.tobytes()
     return tensor
 
 
diff --git a/scripts/quantize.py b/scripts/quantize.py
index 3f73f0916..42e92dd26 100644
--- a/scripts/quantize.py
+++ b/scripts/quantize.py
@@ -12,7 +12,7 @@
 from onnxruntime.quantization import QuantType, QuantizationMode
 from onnxruntime.quantization.onnx_quantizer import ONNXQuantizer
 from onnxruntime.quantization.registry import IntegerOpsRegistry
-from onnxruntime.quantization.matmul_4bits_quantizer import MatMul4BitsQuantizer
+from onnxruntime.quantization.matmul_nbits_quantizer import MatMulNBitsQuantizer
 from onnxruntime.quantization.matmul_bnb4_quantizer import MatMulBnb4Quantizer
 
 from . import float16
@@ -234,7 +234,8 @@ def quantize_q4(
     Quantize the weights of the model from float32 to 4-bit int
     """
 
-    quantizer = MatMul4BitsQuantizer(
+    # Default algo_config is 4 bits quantization
+    quantizer = MatMulNBitsQuantizer(
         model=model,
         block_size=block_size,
         is_symmetric=is_symmetric,
diff --git a/scripts/requirements.txt b/scripts/requirements.txt
index 6d18662f4..9adb65b35 100644
--- a/scripts/requirements.txt
+++ b/scripts/requirements.txt
@@ -1,7 +1,4 @@
-transformers[torch]==4.49.0
-onnxruntime==1.20.1
-optimum@git+https://github.com/huggingface/optimum.git@b04feaea78cda58d79b8da67dca3fd0c4ab33435
-onnx==1.17.0
-tqdm==4.67.1
-onnxslim==0.1.48
-numpy==2.2.6
+optimum-onnx==0.0.3
+onnxruntime==1.23.2
+onnxslim==0.1.78
+accelerate==1.12.0
\ No newline at end of file