Skip to content

[Quantization/Tensor] 0 should be represented as an offset for quantized numbers #2082

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
rdzhabarov opened this issue Nov 27, 2018 · 0 comments · Fixed by #2084
Closed

[Quantization/Tensor] 0 should be represented as an offset for quantized numbers #2082

rdzhabarov opened this issue Nov 27, 2018 · 0 comments · Fixed by #2084

Comments

@rdzhabarov
Copy link
Contributor

zero and isZero are not properly handled for quantized types.

Taking quantized int8 for example with {Scale, Offset} parameters.
Given that dequantization is calculated as Scale * (x_quantized - Offset), actual zero value should be represented as Offset and not just 0.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

[quantization] fix Tensor zero() and isZero() to respect quantized offset nickgg/glow
1 participant
@rdzhabarov