Unable to find image ‘rasa/duckling:latest’ locally docker: Error response from daemon: Get [https://registry-1.docker.io/v2/] no such host. See ‘docker run --help’.
Can you give me a few more details? I guess you executed
docker run -p 8000:8000 rasa/duckling. Is that right? What docker version do you have? What OS do you use?
ow ya…i didnt pull the image first ya i got it…i have another question why is in formaction that action_ask_slot name is not working…i did include it in the action part but didnt get it…do i have to put it in the template?
Please share more details. What Rasa version do you use. How does your domain look like? It would also be helpful if you share an example story that includes
when i start the form it will jump the first two questions that is from custom action (action_ask_slot name) and starts from the third one that is from utter_ask_slot name
What Rasa version are you using? I think you can use a custom action, e.g.
action_ask_department instead of
utter_ask_department, just since Rasa 2.0.x.
You need to update to version Rasa Open Source 2.0.0 in order to forms this way. If you are using Rasa 1.x the form is just looking for
utter_ask_department which you do not have defined.
so how should i upgrade the the existing on or shall i download a new one?
You should update your existing assistant. Please take a look here: Migrating your Rasa 1.x assistant to Rasa 2.0
okay thank you
as u suggested i used or upgrade to rasa 2.0…and i try to fill the form from custom action where it is “action_ask_slot name”…but i have encounterd another error which is when i try to run the next action which takes slot from previous action the sql fails while filling the form
The error is a SQL syntax error, so nothing on Rasa side. Did you already print out the SQL query you try to make and see how it looks? Might be something related to the quotes. Or maybe you need to explicitly cast
get_dep_name to string. I would just print the SQL query and try it directly on the database to see what is going on.
It fails to extract the slot department as the intent you are looking for is
choose_department but it got
physician_intro. How does your NLU data look like? You might need to add more examples or adapt you slot mapping to include
What is the difference between from_text and from_entity…and also while we are doing form filling for example the asked slot is name and if a user inputs my name is tanja instade of tanja why does the form takes all the input as my name is tanja instade of only taking the name to store ??
Please take a look at Forms
In short: Let’s assume you want to fill the slot
person_name. Te slot mapping
from_text would take the complete user utterance, e.g. “my name is tanja”.
from_entity would just take “tanja” as slot value, if “tanja” was detected as “name” entity and the slot mapping of the form would be, for example:
person_name: - type: from_entity entity: name
In that case do I have to put all possible names in the nlu data???
No, that is why you are using entity extractors They should be able to generalize from a few examples. But you might want to improve your results by using lookup tables or regexes (see NLU Training Data).
Okay I see it thanks …but what is the difference between lookup table and regex…and for regex what pipeline am I suppose to use ?? And also if I write lookup or regex do I have also need to write intent first??