fix comment on dnnlowp op arguments (#14265)
authorJongsoo Park <jongsoo@fb.com>
Wed, 21 Nov 2018 17:37:58 +0000 (09:37 -0800)
committerFacebook Github Bot <facebook-github-bot@users.noreply.github.com>
Wed, 21 Nov 2018 17:39:57 +0000 (09:39 -0800)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/14265

Fix comment

Reviewed By: hx89

Differential Revision: D13152106

fbshipit-source-id: fbe98906963cbd5cb20a583a737a792fbc38292e

caffe2/quantization/server/dnnlowp_op.h

index 36aeb6a..49b15fe 100644 (file)
@@ -55,14 +55,15 @@ namespace caffe2 {
  *          this option is intended for debugging accuracy issues.
  *
  *        For the following quantization method related options, please refer
- *        to deeplearning/quantization/dnnlowp/dnnlowp.cc for more details.
+ *        to caffe2/quantization/server/dnnlowp.cc for more details.
  *
  *        - activation_quantization_precision (default=8)
  *        - weight_quantization_precision (default=8)
  *        - requantization_multiplier_precision (default=32)
  *        - eltwise_quantization_precision (default=16)
  *        - force_scale_power_of_two (default=0)
- *        - preserve_sparsity (default=0)
+ *        - preserve_activation_sparsity (default=0)
+ *        - preserve_weight_sparsity (default=0)
  *        - activation_quantization_kind (default=min_max)
  *        - weight_quantization_kind (default=min_max)
  */