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

Add assertion when loading cpu and cuda kernel fails

This commit is contained in:
songxxzp 2023-04-14 19:58:42 +08:00
parent fe0674f86d
commit bcc35f08b4
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
quantization.py
View File

@ -441,10 +441,10 @@ def quantize(model, weight_bit_width, use_quantization_cache=False, empty_init=F
try:
load_cpu_kernel(**kwargs)
except:
print("Cannot load cpu kernel, don't use quantized model on cpu.")
if kernels is None: # CUDA kernels failed
print("Cannot load cuda kernel, quantization failed.")
return model
print("Cannot load cpu or cuda kernel, quantization failed:")
assert kernels is None
print("Cannot load cpu kernel, don't use quantized model on cpu.")
current_device = model.device