# Productionalize the model

**URL:** <https://discuss.huggingface.co/t/productionalize-the-model/2567>\
**Category:** 🤗Transformers\
**Created:** [December 4, 2020, 6:50pm UTC](https://discuss.huggingface.co/t/productionalize-the-model/2567 "2020-12-04T18:50:28Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![chandrak](https://avatars.discourse-cdn.com/v4/letter/c/a9a28c/32.png) [@chandrak](https://discuss.huggingface.co/u/chandrak)\
**Post date:** [December 4, 2020, 6:50pm UTC](https://discuss.huggingface.co/t/productionalize-the-model/2567/1 "2020-12-04T18:50:28Z")

</div>

My use case is to serve re-trained BERT model from huggingface transformers in browser to end users  
(users will type in an editor in browser and they will see next word recommendations - like gmail auto compose).

Is there any example to productionalize “transformers” model ? I am having trouble with converting .bin file to tensorflow.js or onnx.js.

How to save transformers model to with .pt or .h5 extension?