whybeyoung
/
chatglm-6b-int4
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove assert in load_cpu_kernel

This commit is contained in:
duzx16 2023-04-29 10:34:45 +08:00
parent f55a1089a2
commit 63d66b0572
1 changed files with 2 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
quantization.py
View File

@ -442,7 +442,6 @@ class QuantizedEmbedding(Embedding): # TODO: backward, check empty_init
def load_cpu_kernel(**kwargs): def load_cpu_kernel(**kwargs):
global cpu_kernels global cpu_kernels
cpu_kernels = CPUKernel(**kwargs) cpu_kernels = CPUKernel(**kwargs)
assert cpu_kernels.load
def quantize(model, weight_bit_width, use_quantization_cache=False, empty_init=False, **kwargs): 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_h_to_4h_quantization_cache = None
dense_4h_to_h_quantization_cache = None dense_4h_to_h_quantization_cache = None
try: load_cpu_kernel(**kwargs)
load_cpu_kernel(**kwargs) if not cpu_kernels.load:
except:
if kernels is None: # CUDA kernels failed if kernels is None: # CUDA kernels failed
print("Cannot load cpu or cuda kernel, quantization failed:") print("Cannot load cpu or cuda kernel, quantization failed:")
assert kernels is not None assert kernels is not None