All - Activity - timurrakhimzhan - Rasa Community Forum

Add some new layers to Rasa

Rasa CALM Aug 30, 2024

two others, actually. But overall, yeah, that’s what I am trying to do

Rasa CALM Aug 27, 2024

Assistant is using CALM. I am trying to add a translation of user’s input to English and bot’s response to user language. I achieved that using my own InputChannel and OutputChannel, however testing is not possible using rasa inspect.

Rasa CALM Aug 25, 2024

Currently, I am trying to add a translation layer to Rasa. Firstly, I need to translate user’s message to English and after that translate bot’s message back to needed language. However current API of rasa does not allow to do it. Can you please consider adding some new layer, that we can use to man…

Translation Layer

Rasa CALM Jul 30, 2024

Mainly, because you use embeddings for flow retrieval. And yes there are multilanguage embeddings but they perform worse for languages other than English. And for some not so popular languages they perform terrible. Secondly, cheaper models (gpt-3.5) perform far worse when dealing with languages ot…

Rasa CALM Jul 29, 2024

But how to integrate translation of user’s message to English with the current architecture? Creating a channel for this disables a possibility to test a bot from shell/inspect. Creating a GraphComponent which will intercept the message and change its text is not changing it in tracker events, there…

Rasa CALM Jul 28, 2024

Currently Rasa PRO with CALM does not support translation out of the box. I think it would be a good idea to make it possible to add a translation layer to translate user messages to English, save the original language in metadata and translate bot’s message back to the original message using NLG ma…

Possibility to control number of last messages sent to Command Generator

Rasa CALM Jul 28, 2024

Hi! Currently the number of tokens increases with each message since the whole conversation history adds up to the prompt of command generator. I think making it possible to somehow control how many messages actually will be sent is crucial to control tokens. Or maybe I did not see something in docs…