@martinavalogia I think that we currently don’t support Python 3.8 yet. It might help to use Python 3.7 for now as you’re exploring new Rasa. It also seems like you’re trying to install rasa==2.0.0.a1 which is our first alpha candidate. If you want to try out the latest version then you should also be able to run rasa==2.0.0.a2. It is my understanding that soon we’ll also have our 3rd alpha release candidate, but currently a2 is our most recent version.
2 Likes
Hey @koaning and @Emma -Thanks -Apologies for the oversight.Have used rasa==2.0.0.a2 and python version 3.7.9 on a windows 10 setup but now I am facing these issues -How should I go about
Thanks for the instant replies
@martinavalogia Just to check, could you try to install Rasa 1.10 and check if that works?
Also, could you run this in your python shell (from the same virtualenv):
import sys
sys.version
Could you let me know what you see?
Hi @koaning Vincent, I have been able to install version 1.10 and this is what I get
Ah. Then I think we’ve found a really important bug. @martinavalogia could you start a new question here on the forum? I’ll gladly learn what is going wrong here with you but to keep this thread on topic it might be good if we continue this conversation in another one.
You can tag me: @koaning and I’ll pick up the conversation there.
@koaning Yep sure Vincent!
Thanks , Regards Martina
Great. I’m curious to know if RASA can support dynamic loading of NLU based on API/website/user input. Otherwise its completely hardcoding.
Secondly, I would like to also find how we can re-train automatically when NLU is fed in the backend. please let me know. Thanks.
How smooth is the upgrade to Rasa 2.0? Like for example I am currently using the available Rasa version and once this 2.0 release available I would like to go for it so what all changes are required?
Hey everyone,
Just a little update, you can find the latest version here:
Alpha 3 - Version: 2.0.0a3
Rasa version==2.0.0a3. When I run the interactive mode with ‘rasa interactive’, it suggests to save the story in data/stories.yml. However, the console logs the following error: Please provide a valid export path for the stories, e.g. ‘stories.md’.Same goes for the nlu.yml. I can’t save the result of the interactive mode in the new data format. Is this feature missing in the current version or am I doing something wrong?
Anyone know how to deal with FAQ in 2.0? I couldn’t find any information from 2.0 doc and example.
Thanks!
Hey @yiouyou, have you seen the page on Chitchat and FAQs?
@akelad Thanks, my bad, I didn’t find the page.
A few questions about chitchat and FAQs in 2.0:
-
It seems support multiple answers for a single intent, correct me if I’m wrong
-
Since we’re using retrieval actoins to handel all FAQs, is there a way to distinguish the different confidence for each of them? The reason I’d like to get the information is to list the top 3 hits of FAQs, and let the user to choose any of or none of them.
-
If ResponseSelector couldn’t do the job properly, any way to tweak it with two stage fallback?
Thanks!
Has a date been set for Rasa 2 stable release yet?
Does the following confidence_threshold and nlu_fallback work for FAQs ResponseSelector?
And how to get the nlu fallback confidence list?
Thanks!
Hi, I am having trouble installing rasa rasa==2.0.0a3, it complains about not finding an compatible rasa-sdk version.
has anyone experience this problem?
@yiouyou sorry for the delay in answering you.
- Yes, you can specify responses for retrieval intents (chitchats/faqs) as you would normally for any other responses - meaning you can have multiple responses (which get chosen randomly), image attachments etc.
- What do you mean by letting the user choose any or none of them? Is this for the Fallback policy in particular, or is there a different use case?
- The answer here depends on the above
@tatianaf We have a webinar scheduled for the 14th of October, so it’ll be released at some point before then.
I am having problem with lookup tables. I can not find how to reference them in the new version. Old style does not seem to work. Does anyone had experience configuring them for rasa 2.0?