# Converting StyleGan to tensorflow js **URL:** \ **Category:** General Discussion\ **Tags:** keras, models, tfjs, help\_request\ **Created:** [June 8, 2021, 12:25am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335 "2021-06-08T00:25:39Z")\ **Posts on this page:** 9\ **Page:** 1
**Author:** ![Saswat\_Das](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/saswat_das/32/7438_2.png) [@Saswat\_Das](https://discuss.ai.google.dev/u/Saswat_Das)\ **Post date:** [June 8, 2021, 12:25am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/1 "2021-06-08T00:25:39Z")
Hello. This doubt may seem a little bit less researched, but here I go. I wanted to convert the stylegan generator to tensorflow js model. But this I understand that stylegan python (using TF2.x) has custom loops and objects in the generator object. I followed the steps mentioned below : I right now have a keras model stylegan generator with weights loaded and the class has a call function that takes in noise and gives images. The definition is same as “StyleGAN\_G” class in [StyleGAN-Tensorflow2/stylegan.py at master · ialhashim/StyleGAN-Tensorflow2 (github.com)](https://github.com/ialhashim/StyleGAN-Tensorflow2/blob/master/stylegan.py). The weights I loaded was using the tutorial given in the same repository [StyleGAN-Tensorflow2/stylegan.ipynb at master · ialhashim/StyleGAN-Tensorflow2 (github.com)](https://github.com/ialhashim/StyleGAN-Tensorflow2/blob/master/stylegan.ipynb). So which format do I save the generator object (before converting it to tfjs model) ? Tensorflow seems to have a lot of save options And after saving, what all flag values I have to set tfjsconvertor ? ---
**Author:** ![lgusm](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/lgusm/32/6230_2.png) [@lgusm](https://discuss.ai.google.dev/u/lgusm)\ **Post date:** [June 9, 2021, 11:23am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/2 "2021-06-09T11:23:28Z")
@Jason might be able to help ---
**Author:** ![Jason](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/jason/32/6193_2.png) [@Jason](https://discuss.ai.google.dev/u/Jason)\ **Post date:** [June 9, 2021, 6:49pm UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/3 "2021-06-09T18:49:56Z")
Let me have a chat with the team and get back as I have not done much in the GAN space myself yet, but welcome to the community Saswat! ---
**Author:** ![lina128](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/lina128/32/7138_2.png) [@lina128](https://discuss.ai.google.dev/u/lina128)\ **Post date:** [June 9, 2021, 7:13pm UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/4 "2021-06-09T19:13:14Z")
Hi @Saswat_Das , please save it as SavedModel, `model.save('my_model')` (doc: [Save, serialize, and export models  |  TensorFlow Core](https://www.tensorflow.org/guide/keras/save_and_serialize)). And then follow the tfjs-converter doc to convert the SavedModel to tfjs format: [tfjs/tfjs-converter at master · tensorflow/tfjs · GitHub](https://github.com/tensorflow/tfjs/tree/master/tfjs-converter) Basic cmd: ```auto tensorflowjs_converter \ --input_format=tf_saved_model \ --output_format=tfjs_graph_model \ --signature_name=serving_default \ --saved_model_tags=serve \ model_path/saved_model \ new_path/web_model ``` ---
**Author:** ![Saswat\_Das](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/saswat_das/32/7438_2.png) [@Saswat\_Das](https://discuss.ai.google.dev/u/Saswat_Das)\ **Post date:** [June 10, 2021, 8:35am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/5 "2021-06-10T08:35:49Z")
@lina128 Thank you . I converted the model to shard files and model.json. Then I loaded the graph model. When I tried to run the following code to generate the image, it throws an error . `const noise = tf.randomNormal([512]).expandDims(0); let generatedImage = await generator.executeAsync(noise)` The error is that: util\_base.ts:154 Uncaught (in promise) Error: Error in depthwiseConv2d: number of input channels (8) must match the inChannels dimension in filter 512. at Object.assert (util\_base.ts:154) at depthwiseConv2d\_ (depthwise\_conv2d.ts:99) at Object.depthwiseConv2d\_\_op (operation.ts:51) at Object.executeOp (convolution\_executor.ts:195) at operation\_executor.ts:69 at engine.ts:467 at Engine.scopedRun (engine.ts:478) at Engine.tidy (engine.ts:465) at Object.tidy (globals.ts:192) at operation\_executor.ts:68 ---
**Author:** ![Saswat\_Das](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/saswat_das/32/7438_2.png) [@Saswat\_Das](https://discuss.ai.google.dev/u/Saswat_Das)\ **Post date:** [June 10, 2021, 5:58pm UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/6 "2021-06-10T17:58:34Z")
I am guessing it is because of the fact that depthwise convolution in tensorflow doesn’t support ‘BCHW’ format as mentioned in the API reference. But in the actual python code the data\_format specified in the depthwise convolution was ‘BCHW’. Can anyone suggest what could be the error ? ---
**Author:** ![lina128](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/lina128/32/7138_2.png) [@lina128](https://discuss.ai.google.dev/u/lina128)\ **Post date:** [June 14, 2021, 4:52pm UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/7 "2021-06-14T16:52:10Z")
It’s possible. Please add a feature request to add NCHW format or considering contribute. Or change the python model to use NHWC. ---
**Author:** ![Saswat\_Das](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/saswat_das/32/7438_2.png) [@Saswat\_Das](https://discuss.ai.google.dev/u/Saswat_Das)\ **Post date:** [June 15, 2021, 5:53am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/8 "2021-06-15T05:53:13Z")
Yes thank you. I changed the model to NHWC and it worked like a charm. ---
**Author:** ![MadMan](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/madman/32/8665_2.png) [@MadMan](https://discuss.ai.google.dev/u/MadMan)\ **Post date:** [August 19, 2021, 2:55am UTC](https://discuss.ai.google.dev/t/converting-stylegan-to-tensorflow-js/17335/9 "2021-08-19T02:55:46Z")
Can you please share how did you convert it to NHWC format?