Hi , I´m new with rasa …I have a doubt…there is a limit characters/words when you use nlu for indentify the intent? …for example :
curl 192.168.2.144:5000/model/parse -d ‘{“text”:“text with 1000 words”}’
Hi @GiovanniSalazar. If you test your assistant with rasa shell nlu you will not be able to pass 815 character limit. Very important: it’s impossible to build a bot that can handle very long user inputs simply because NLU model will fail to classify them correctly. The user inputs you train your assistant to understand should be short and correspond to a specific intent label which is impossible to achieve with very long inputs.
Can you tell me more about your use case?
Thank , I´m interested in to use rasa nlu for to detect intents respect a several historial conversations (not real time) …there is some feature that I can use?..
@Juste i’d like to learn more about RASAs character limits, but i couldn’t find any other info about them online. I have a case where users may submit long voice notes, which i transcribe and sent to the NLU (these voice notes may be 10 minutes long and contain over 1000 characters). How can i find out if this will be a problem?
Thanks
Hi @travelmail26. The main problem of long messages is that it might be very difficult for your NLU model to process those messages correctly (extract the intent). Could you provide a little bit more context on what the conversations would be about?
I am having an issue that when i hit my intent it then hit an api in flask after that when it has to show output which is a base 16 file then rasa is showing error “Line too long”. any idea?
I’d like to know more about RASA character limits as well. In my use case I’d like to process a long message which would likely contain more than 815 characters. However this message will not be used for intent identification. It’ll be designed to follow certain rules (meaning the conversation flow is pre-defined, intent-irrelevant). I’m wondering whether there’s a workaround for the 815 char limits in this case?
Other than creating an intent with NLU samples of just a variety of long messages (would that even work?), is there a way to handle user messages over a certain character limit? The only way I can think of doing it is having a custom action that runs in every story after every intent to check the tracker.latest_message[‘text’] length, then have a Boolean slot for “too long”.
That seems far too cumbersome, though. Any suggestions?