Which models would you like to see on TensorFlow Hub?

General Discussion
,
#1

TFHub have +1000 models but given the fast development of ML, there’s always a cool new one that you might want to try.
Which one are you missing?

5 Likes
#2

Can we add more audio tasks? More specifically in music projects (pitch detection, chord detection, source separation).

Example of a super recent model (published today) that I’m excited about: TensorFlow Hub

5 Likes
#3

Thanks for the suggestion peddy.
We’re working on that specifically! Let’s we can improve that!
thanks

2 Likes
#4

I agree with Peddy. We have lots of good models for text and image tasks, but audio tasks and models are mostly missing.
I’ve recently published TTS models in German on TF Hub (tfhub.dev/monatis), and I’m getting prepared to publish more models for TTS and ASR in several languages. We have good end-to-end models for these tasks, and training them might be challenging and time consuming. So I believe that publishing them will be good contributions to the community, and I’m ready to collaborate with anyone willing to do so :rocket:

2 Likes
#5

I would like to go back to Image Classification. The following two would be my top priority:

  • RegNet
  • Vision Transformers (ViT) in TensorFlow
3 Likes
#6

those are good suggestions!
thanks for contributing!

1 Like
#7

Those are nice too! I want to play with ViT too!

1 Like
#8

It could be nice to have Google’s Pay Attention to MLPs.

2 Likes
#9

Thanks @Bhack , that’s a good idea!

1 Like
#10

As a side node and more in general, I think that we could find a way to better interact with paperswithcode with TFHUB and the Model Garden SIG now that it is integrated with ArXiv

2 Likes
#11

I would like to see:

  • even more speeded-up variants of existing (mostly mobile/embedded focused) models: quantized variants down to 8 bit integers
  • Everything having support for 16 bit float32 / amp
  • yolo
3 Likes
#12

That’s a good idea @Bhack , for Model Garden I can take a look and see if we can improve some processes!

1 Like
#13

Thanks for the feedback @ntakouris

for the quantized versions, that’s something we can always tell the publishers that developers want it!

Yolo specifically was mentioned some days ago in another thread and TFHub is open for non-Googlers to publish their models. So if the Yolo maintainers want to publish their models, that would be awesome!

1 Like
#14

@lgusm It would be great.

Take a look at some recent stats:

2 Likes
#15

I hope that in addition to the trained model, we can also rebuild the model structure implemented with TensorFlow 2.x and use the corresponding dataset for learning and experiencing the training process.

1 Like
#16

Hi Nan Zheng,

For the full code, you can usually find it on TensorFlow Model Garden.

The corresponding dataset might be much harder since all of them have a specific license and hosting them is much harder than a technical issue. On TFHub and Model Garden the dataset the model was trained on is mentioned in the documentation but not hosted.

2 Likes
#17

Thanks for sharing @Bhack

2 Likes
#18

Probably we can improve the collaboration and integration with TFDS, when possibile, to have an improved ecosystem experience.

2 Likes
#19

Yes, that would be a good thing to do.
This kind of feedback really help the team to prioritize next steps so please keep them coming!

3 Likes
#20

Thanks for your reply @lgusm ,

Unfortunately not all models in TensorFlow Hub can I find the source code. For example, I’m interested in " handskeleton" model, what only I know is the structure is SSD, but I can’t find the TensorFlow code in TensorFlow Model Garden, will someone post it in there in the future?

1 Like