500 error when training a model with version 0.23.0

Hi, After updating rasa to version 0.23.0, I get the following error when I train a new model:

Training Core model ...
Core model training completed.
Training NLU model ...
2019-11-28 12:43:28 ERROR rasax.community.api.blueprints.models - 500, Internal Server Error, body = 'b' {"version": "1.5.0", "status": "failure" , "message": "An unexpected error occurred during training. Error: argument of type \ 'NoneType \' is not iterable", "reason": "TrainingError", "details": {}, "help": null, " code ": 500} ''

How can I solve that? Thank you!

@rcucala can you try upgrading to rasa==1.5.1? We’re working on releasing a 0.23.1 for rasa-x today that uses that version

1 Like

Hello @akelad thanks for answering :slight_smile:

I just made the update to version 0.23.2 and the problem still persists having the rasa version 1.5.1 I attach updates and errors logs.

root@e71108ac3956:/app# pip install -U rasa-x --extra-index-url https://pypi.rasa.com/simple
Looking in indexes: https://pypi.org/simple, https://pypi.rasa.com/simple
Collecting rasa-x
  Downloading https://pypi.rasa.com/api/package/rasa-x/rasa-x-0.23.2.tar.gz (1.6MB)
     |β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 1.6MB 11.9MB/s
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
    Preparing wheel metadata ... done

root@e71108ac3956:/app# rasa --version
Rasa 1.5.1

The error in rasa docker container its :

Training Core model...
Core model training completed.
Training NLU model...
2019-12-03 16:18:10 ERROR    rasax.community.api.blueprints.models  - 500, Internal Server Error, body='b'{"version":"1.5.1","status":"failure","message":"An unexpected error occurred during training. Error: argument of type \'NoneType\' is not iterable","reason":"TrainingError","details":{},"help":null,"code":500}'' 

I also attach a screenshot

I really appreciate the help you can give us!!

Thank you very much!

Hello, I have finally managed to fix the error. The problem was in the following configuration of the config.yml file:

- name: CountVectorsFeaturizer
- name: EmbeddingIntentClassifier
- intent_tokenization_flag: true
- intent_split_symbol: +

I have commented on these lines for multi-intentions and the error is no longer generated when doing the training.

Could you tell me where the error could be?

Thank you!

1 Like

@rcucala you need to get rid of the - in front of the intent parameters so it looks like this:

- name: CountVectorsFeaturizer
- name: EmbeddingIntentClassifier
  intent_tokenization_flag: true
  intent_split_symbol: +