SVDF layer implementation compatible with TFLite’s SVDF operator

google-ai-edge / ai-edge-torch

Supporting PyTorch models with the Google AI Edge TFLite runtime.

Apache License 2.0

377 stars 51 forks source link

SVDF layer implementation compatible with TFLite’s SVDF operator #386

Open gaikwadrahul8 opened 1 day ago

gaikwadrahul8 commented 1 day ago

Although TFLite has a built-in SVDF operator (as listed here, and whose implementation can be seen here), currently, Tensorflow doesn't have a keras layer implementation for the SVDF operation. I was wondering what would be the sequence of operations (using Tensorflow's Python library) needed so that they could be fused and recognized as an SVDF operator after converting to TFLite.

pkgoogle commented 13 hours ago

Original Issue: https://github.com/tensorflow/tensorflow/issues/63035