Embedding policy 'active_form_sales_form_form' could not be found in feature map

Rasa Open Source
,
#1 
Feature 'active_form_sales_form_form' could not be found in feature map.

 rasa_core.featurizers - Feature 'active_form_sales_form_form' (value: '1.0') could not be found in feature map. Make sure you added all intents and entities to the domain
#2

Hi Azizullah, Did you find an answer to these messages? I am starting to see those in my bot and not sure what it is, and what should I correct. Any help, much appreciated

Thanks

#3

@azizullah2017 I should probably reference with the user tag, my mistake Hi Azizullah, Did you find an answer to these messages? I am starting to see those in my bot and not sure what it is, and what should I correct. Any help, much appreciated

Thanks

#4

I do face the same warning

#5

Having the same problem

#6

try to check the domain.json in the model created after make train-core to check that it exists. you might find it in models/dialogue/domain.json

#7

Thank you @hossa95 in my case. I can see ‘active_form_agenda_form’ as the very last item on the domain.json

Shaii I delete that? or having ‘active_form_agenda_form’ in the domain.json is the expected behavior?

I am not sure I am understanding the reason of this warning, and by the way it is printed twice on the log screen.

WARNING rasa_core.featurizers - Feature ‘active_form_agenda_form’ could not be found in feature map.

Thanks again

#8

I am also facing the same issue (training time) for all the forms defined in my domain.yml. Additionally I get the same warnings during run time as well (http api) and seems to impact the responses I get back from the bot. Has anyone faced the same issue at the run time as well?

#9

@akelad do you have any thoughts on this one? explain and chitchat sample here GitHub - RasaHQ/conversational-ai-workshop-18: Example showing generalisation is not working or get a wrong response from bot on my data set when I use EmbeddingPolicy. Also any pointer in general to implement explain intent - how does nlu identify explain intent?

#10

@apurva I would recommend to use keras policy without using the parameter of – argument 0

#11

Any update on this one?

"Thank you @hossa95 in my case. I can see ‘active_form_agenda_form’ as the very last item on the domain.json

Shall I delete that? or having ‘active_form_agenda_form’ in the domain.json is the expected behavior?

I am not sure I am understanding the reason of this warning, and by the way it is printed twice on the log screen.

WARNING rasa_core.featurizers - Feature ‘active_form_agenda_form’ could not be found in feature map.

Thanks again"

#12

@azizullah2017 Thanks for your input but I would like to use EmbeddingPolicy or at least would like to understand why I am getting that warning. In either case trying to see how i can implement ‘explain/chitchat’ intents. Based on the article it seems attention model in EmbeddingPolicy should perform better functionally. @Juste do you have any thoughts on the issue I have reported earlier on this thread?

1 Like
#13

I really don’t know what’s wrong … you should see it in the domain.json … can you post ur nlu pipeline and your train_dialogue function may be we can find something missing

#14

thank you @hossa95 for your answer and I am not sure I am understanding all the information you are requesting Below is an extract of my domain.json where you can see at the end the “active_form_agenda_form”

“states”: [ “intent_affirm”, … … “prev_agenda_form”, “active_form_agenda_form” ] }

Here the NLU pipeline

pipeline:

  • name: tokenizer_whitespace
  • name: ner_crf

- name: ner_synonyms

  • name: intent_featurizer_count_vectors token_pattern: (?u)\b\w+\b
  • name: intent_classifier_tensorflow_embedding epochs: 100
  • name: ‘ner_duckling_http’ url: ‘http://localhost:8000’ locale: “es_MX” dimensions: [‘time’,‘PhoneNumber’]

the train dialogue I am using is in the makefile train-nlu: python -m rasa_nlu.train -c config_tf.yml --fixed_model_name current --data data/nlu.md -o models --project nlu/default --verbose

train-core: python -m rasa_core.train -s data/stories.md -d domain.yml -o models/dialogue -c policy_config.yml --augmentation 0 --debug --dump_stories --debug_plots

hope this helps to clarify @Juste hope you can add some light here

Thanks everybody

#15

Hi @amsolu, Have you resolved the issue?