Skip to content

QDense_batchnorm? #64

@thesps

Description

@thesps

I see you have qconv2d_batchnorm layer which folds the weights of the two layers and then quantizes.

We're bringing support for that to hls4ml, and it should help us save some resources & latency.

I'm wondering, do you plan to add the equivalent combined QDense + BatchNormalization layer to QKeras?

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions