Skip to content

[Pass] fix quant_dequant_fuse_pass when there is no dequant op before weight #10093

Merged
hong19860320 merged 5 commits intoPaddlePaddle:developfrom
mjp9527:fix_conv2d_transpose
Mar 16, 2023
Merged

[Pass] fix quant_dequant_fuse_pass when there is no dequant op before weight #10093
hong19860320 merged 5 commits intoPaddlePaddle:developfrom
mjp9527:fix_conv2d_transpose

Conversation

@mjp9527
Copy link
Collaborator

@mjp9527 mjp9527 commented Mar 13, 2023

PR devices

others

PR types

Bug fixes

PR changes

PASS

Description

fix quant_dequant_fuse_pass when there is no dequant op before weight

@paddle-bot
Copy link

paddle-bot bot commented Mar 13, 2023

Thanks for your contribution!

bool enable_int8_cond = false;
for (auto& inlink_node : quantized_node->inlinks) {
enable_int8_cond = true;
// no op before weight
Copy link
Collaborator

@hong19860320 hong19860320 Mar 15, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

建议加上类似

/* fuse logit block: y=log(x/(1-x)),inverse function of sigmoid */
示意图

Copy link
Collaborator

@hong19860320 hong19860320 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@hong19860320 hong19860320 merged commit 4ee3f6d into PaddlePaddle:develop Mar 16, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants