OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 37 Packages Projects Releases Wiki Activity

add warning to dnnlowp fc if quantization kind is not min_max

Browse Source 
Summary:
Print warning when using DNNLOWP dynamic int8 quant for FC and activation_quantization_kind != min_max.

Warning will display in console but not in Bento. Would have to use CAFFE_ENFORCE to alert in Bento.

Test Plan: buck run unit test forcing DNNLOWP FC with activation_quantization_kind = "l2" and saw warning printed in console.

Reviewed By: csummersea

Differential Revision: D17770921

fbshipit-source-id: b6532e4c9a86d74e3db4cb432735505d378a366e
This commit is contained in:
Amy Yang 2019-10-04 17:01:53 -07:00 committed by Facebook Github Bot
parent afbbe16f49
commit 381cf2bd24
1 changed files with 14 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
caffe2/quantization/server/caffe2_dnnlowp_utils.cc
View File

@ -71,7 +71,20 @@ TensorQuantizationParams GetInputTensorQuantizationParamsOf(
float min, max;
fbgemm::FindMinMax(
tensor->template data<float>(), &min, &max, tensor->numel());
auto activation_quantization_kind = qfactory->GetActivationKind();
if (activation_quantization_kind !=
QuantizationFactory::QuantizationKind::MIN_MAX_QUANTIZATION) {
LOG(WARNING)
<< "DNNLOWP dynamic int8 FC uses min_max as the only activation_quantization kind. Qparams will be assigned based on min_max regardless of activation_quantization_kind args.";
}
if (is_weight) {
auto weight_quantization_kind = qfactory->GetWeightKind();
if (weight_quantization_kind !=
QuantizationFactory::QuantizationKind::MIN_MAX_QUANTIZATION) {
LOG(WARNING)
<< "DNNLOWP dynamic int8 FC weight is not constant, assigning qparams to weight based on min_max, regardless of weight_quantization_kind args.";
}
}
return qfactory->ChooseQuantizationParams(min, max, is_weight);
}
}