riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-02 19:21:31 +06:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

Raise error when quantizing a quantized model (#27500)

Browse Source 
add error msg
This commit is contained in:
Marc Sun 2023-11-16 16:35:40 +01:00 committed by GitHub
parent fd65aa9818
commit 6b39470b74
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/transformers/modeling_utils.py
View File

@ -2781,7 +2781,12 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
quantization_method_from_config = config.quantization_config.get(
"quant_method", QuantizationMethod.BITS_AND_BYTES
)
if quantization_method_from_config is not None and quantization_method_from_args is not None:
if quantization_method_from_config != quantization_method_from_args:
raise ValueError(
f"The model is already quantized with {quantization_method_from_config}. "
f"You can't quantize it again with {quantization_method_from_args}"
)
if quantization_method_from_config == QuantizationMethod.GPTQ and quantization_method_from_args is not None:
loading_attr_dict = quantization_config.get_loading_attributes()
for attr, val in loading_attr_dict.items():