Can I convert a LSTM Model of tensorflow to tensorFlowjs ?

I created a sentiment model in tensorflow using python.In which I used LSTM Layers.I wanted to use that in web browser so, I converted that model to TensorFlow js format using tensorflow js converter. I got three files with names

1. weights_manifest.json
2. tensorflowjs_model
3. group1-shard1of1

where 2,3 are binary files & json file contains information like [{"weights": [{"shape": [], "dtype": "int32", "name": "ArgMax_1/dimension"}], "paths": ["group1-shard1of1"]}] on seeing this I can say this model doesn’t work because it does not have any information of network.

In FAQ’s i found that Models with control flow ops (e.g. RNNs) are not yet supported.I just want to make sure does LSTM’s prepared model works in Tensorflowjs ?

I looked at example of sentiment analysis and translator of tensorflowjs. They are just making use of predefined models from the cloud.Can I get the examples of training those models with the data or conversion of those models from tensorflow to tensorflowjs format ?

[rthadur]

Automatically closing due to lack of recent activity. Please update the issue when new information becomes available, and we will reopen the issue. Thanks!

[bileschi]

Have you seen this new lstm-text-generation example? https://github.com/tensorflow/tfjs-examples/tree/master/lstm-text-generation

On Mon, Jul 2, 2018 at 4:52 PM, Gautam Mittal [e-mail] wrote:

+1 for text preprocessing examples

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/tensorflow/tfjs/issues/458#issuecomment-401931457, or mute the thread https://github.com/notifications/unsubscribe-auth/AAhZTtZz6gvp403M7_sm9156tqtGPzS8ks5uCogYgaJpZM4UyCk3 .

– Stan Bileschi Ph.D. | SWE | [e-mail] | 617-230-8081