Transition from v0.x.x to v1.0 confusing

Rasa Open Source
1

Hey there!

First of all, great job for Rasa X it is a very nice feature !

I’ve been working with Rasa NLU/Core v0.x.x for over a year now not as a pip module but directly using the source code because I need to slightly change it for my applications (GPU, custom message broker, etc.) .

I’m used to play directly with the python code for launching my servers, training my models, etc. With the new version Rasa 1.0, I don’t know where to find such entry points. I know your are transiting towards a new CLI features but I’m confused how to use it directly using the python code.

Could someone enlighten me ?

Regards,

Alex

PS - here are some examples

PYTHONPATH=…/…/ python …/…/rasa_nlu/evaluate.py --data data/nlu_data.md --config …/…/sample_configs/tf_embedding.yml --mode crossvalidation

PYTHONPATH=…/…/ python …/…/rasa_core/train.py interactive -o models/demo -d domain.yml -s data/stories.md --config policy_embedding.yml --endpoints endpoints.yml --augmentation 0 -u default/

1 Like
2

Hey Alex,

Check out the migration guide – the scripts aren’t directly executable anymore, the CLI should be used. Migration Guide

3

@asokolow You can still run your modified source code, by running the library module as a script:

python3 -m rasa train -o models/demo -d domain.yml -s data/stories.md …

1 Like