Deploying trained model in production

Archives DeepSpeech
1

Hi all,

I have a trained model and I want to deploy this model for production. I am not sure how to do this. But with some research I ended up with a conclusion to use TensorFlow servings. am I in a right track? or any other suggestions would be really helpful.

Thanks,
Megha

2

We removed support for TensorFlow serving a long-time ago, but it could work, maybe with some hacking. How you want to deploy will depend on your usecase …

3

Once example: https://gitlab.com/deepspeech/ds-srv/tree/master/

1 Like
4

I’m just curious, why was the TF serving support removed?

What’s the best practice of serving a DS model as part of a docker or kubernetes environment nowadays?

1 Like