diff --git a/quantization.py b/quantization.py
index 49b53de..d739c90 100644
--- a/quantization.py
+++ b/quantization.py
@@ -442,7 +442,6 @@ class QuantizedEmbedding(Embedding):  # TODO: backward, check empty_init
 def load_cpu_kernel(**kwargs):
     global cpu_kernels
     cpu_kernels = CPUKernel(**kwargs)
-    assert cpu_kernels.load
 
 
 def quantize(model, weight_bit_width, use_quantization_cache=False, empty_init=False, **kwargs):
@@ -453,9 +452,8 @@ def quantize(model, weight_bit_width, use_quantization_cache=False, empty_init=F
     dense_h_to_4h_quantization_cache = None
     dense_4h_to_h_quantization_cache = None
 
-    try:
-        load_cpu_kernel(**kwargs)
-    except:
+    load_cpu_kernel(**kwargs)
+    if not cpu_kernels.load:
         if kernels is None:  # CUDA kernels failed
             print("Cannot load cpu or cuda kernel, quantization failed:")
             assert kernels is not None