Hi, I have developed a chatbot that I want to deploy on another computer. Can I use the ‘models’ saved on my computer to be used on another computer? Or I need to do ‘rasa train’ everytime I change a computer?
I’m not sure how comfortable you are managing files, but the trained models can be copied around between computers or even shared over HTTP: Model Storage (rasa.com). By default, you’ll find your trained models stored in a subdirectory called
models/ from wherever you trained your model on the command line.
Is that enough to solve your problem?
@aman31kmr Yes, you can import the old train models in models folder, but make sure it have only one model the latest one or by running specific model using this command or ref this Model Storage.
rasa run --model models/20190536-100418.tar.gz
Tips: Train models also depends on the Rasa Versions on which the last train model is trained, in that case, you will see the error saying that the model is trained on Rasa 1.0 etc. In this case you need to trained the model again based on your current Rasa Version or environments. Further, if your model is dependent on action server the you have to met all the dependencies required to run the action.py file for the same.
If you have same cloned environment in other system. I will encourage to create requirement file from old environment while using the below command:
pip freeze > requirements.txt
Note: The most common command is pip freeze > requirements . txt , which records an environment’s current package list into requirements . txt . If you want to install the dependencies in a virtual environment, create and activate that environment first, then use the Install from requirements .
run: pip install -r requirements .txt
I hope all these points will help you. Good Luck!