Hello. I would like to use the multi-intent pipeline for a problem of mine and decided to get a better understanding using first the example from https://rasa.com/blog/how-to-handle-multiple-intents-per-input-using-rasa-nlu-tensorflow-pipeline/. I adapted the various data files to rasa version requirement and trained only for the nlu problem.
It seems to me that setting the flags:
intent_split_symbol: "+"
intent_tokenization_flag: true
does not have any effect on the rasa shell nlu
output, i.e. whether I am using true
or false
for the intent tokenization flag I get very similar prediction confidences.
The example from the blog is fairly outdated and I was wondering I somebody could comment on the expected behavior with recent versions of rasa.
Thanks
Rasa Version : 2.8.21
Minimum Compatible Version: 2.8.9
Rasa SDK Version : 2.8.3
Rasa X Version : 1.0.1
Python Version : 3.8.10
Operating System : Windows-10-10.0.18363-SP0
Python Path : C:\ProgramData\miniforge3\envs\ihi\python.exe
config.yml (568 Bytes) domain.yml (1.2 KB) nlu.yml (2.7 KB) stories.yml (2.9 KB)