RASA NLU - Can we set the model name to a static name in the latest release

In a prior version (not sure what version number) you could make the model names static. In the latest version I can’t find a way to provide the file name for the model. The note on the doc indicates the server now only accepts command line arguments and the config is only for the default pipeline. Can anyone tell me how I can set the model file name

Hey @ronmichaud. You should be able to set the model name by providing the --fixed_model_name parameter inside the training command. For example:

!python -m rasa_nlu.train -c nlu_config.yml --data nlu.md -o models --fixed_model_name nlu --project current --verbose

In the example above the fixed mode name will be nlu.

Thank you, but we are using the web API for training. Is there a way to do it in the HTTP train request? We are using the docker deployment and the Rest services for all interactions.

I will test your suggestion as well, just to make sure I have a fallback if we can’s do it from the APIs

@Juste the command line does work, but does not make it available to the server. If I restart the server it does show as available, but still will not get loaded for some reason.

I need to be able to train and make it available to the server.

Thoughts?

@Juste - I found a trail where there is an issue if you have a fixed model name you must also include that name in the parse request {…, model: “nlu”} . That now appears to work.

1 Like

Awesome. Glad to hear that :slight_smile:

This is not work for me. :frowning: