Pretrained fusion layers

[adrienchaton]

Hi !

Thanks for providing OpenL3 as a package. I would like to ask if one could access the pretrained fusion layers, which would classify whether a pair of image/audio embeddings are matched or not.

Otherwise I would train it myself, on top of the pretrained fixed embedding extractors you provide. But as a starting point, I would take your classifier if possible.

Best

[justinsalamon]

We don't provide the fusion layers as of now. This could be something we could add in the future potentially, but right now our focus is on adding support for TF 2.x and other usability improvements, so it's hard to say whether/when we'd get to this.

[adrienchaton]

thanks for your reply