Skip to content
This repository has been archived by the owner on Sep 18, 2024. It is now read-only.

NNI can't quantize bias? #2903

Closed
wZuck opened this issue Sep 17, 2020 · 1 comment · Fixed by #2914
Closed

NNI can't quantize bias? #2903

wZuck opened this issue Sep 17, 2020 · 1 comment · Fixed by #2914
Assignees
@linbinskn
Labels
model compression user raised

Comments

@wZuck
Copy link

wZuck commented Sep 17, 2020

Recently, i tried to apply NNI to my model(with biases), but i found that NNI only support ['weight', 'input', 'output'] quantitation, is that NNI do not support bias quantization(in QAT)?
@QuanluZhang QuanluZhang self-assigned this Sep 17, 2020
@QuanluZhang
Copy link
Contributor

@wZuck thanks for pointing it out. bias has not been supported in the current implementation of quantization. We will support it in the next release soon.

@QuanluZhang QuanluZhang assigned linbinskn and unassigned QuanluZhang Sep 21, 2020
@QuanluZhang QuanluZhang linked a pull request Sep 23, 2020 that will close this issue
Add bias quantization in QAT and refactor the code of weight quantization #2914
Merged
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees

@linbinskn linbinskn

Labels
model compression user raised
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Add bias quantization in QAT and refactor the code of weight quantization linbinskn/nni
4 participants
@wZuck @QuanluZhang @scarlett2018 @linbinskn