We recently switched over to the tensorflow_embedding pipeline (mostly due to multiple intent support and less memory consumption). However (due to the architecture of the pipeline), the model doesn’t generalize over new examples at all. In many cases the models returns an incorrect intent with very high confidence and the correct intent (along with many others) is returned as zero. Is it possible to have a structured way of solving this problem? I tired the pipeline specified in this thread [image] Pre-load vectors for Tensorflow ranking classifier pipeline Rasa Open Source I was looking this exact thing. Has someone figured out a standard way of doing this that works? It would be extremely helpful to us since our data set is quite small but it didn’t work and I got the same results.

Pre-Trained vectors for tensorflow pipeline

adrianhumphrey111 (Adrian Humphrey) September 2, 2018, 6:40pm 3

hey man, could you take a look at this for me?

Topic		Replies	Views
Decision about using a pre-trained words embeddings or not Getting Started with Rasa	3	136	June 3, 2020
Improve Rasa NLU model Rasa Open Source	5	2007	October 15, 2019
spaCy pretrained models break chatbot NLU capacities Rasa Open Source	6	716	October 16, 2019
New Language support docs inconsistency Rasa Open Source	4	474	August 12, 2020
What features does Rasa NLU use from spacy? Rasa Open Source	0	618	February 21, 2019