Fixed ambiguous logging in DIETClassifier #5444

tttthomasssss · 2020-03-18T21:55:17Z

Proposed changes:

Improved logging as it was unclear whether a log message comes from DIETClassifier or a subclass.

Changes made:

added class name of caller to log message

Status (please check what you already did):

added some tests for the functionality
updated the documentation
updated the changelog (please check changelog for instructions)
reformat files using black (please check Readme for instructions)

…case this is executed from a subclass

CLAassistant · 2020-03-18T21:55:21Z

All committers have signed the CLA.

tabergma

Looks good 👍 Thanks for adding the class name!

(You need to sign the contributor license. Otherwise you cannot merge.)

tabergma · 2020-03-19T07:08:07Z

rasa/nlu/classifiers/diet_classifier.py

@@ -267,7 +267,7 @@ def __init__(

        if component_config is not None and EPOCHS not in component_config:
            logger.warning(
-                f"Please configure the number of '{EPOCHS}' in your configuration file."
+                f"Please configure the number of '{EPOCHS}' in your configuration file for {self.__class__.name}."


Suggested change

f"Please configure the number of '{EPOCHS}' in your configuration file for {self.__class__.name}."

f"Please configure the number of '{EPOCHS}' in your configuration file for '{self.__class__.name}'."

tabergma · 2020-03-19T07:08:35Z

rasa/nlu/classifiers/diet_classifier.py

@@ -647,8 +647,8 @@ def train(
    def _predict(self, message: Message) -> Optional[Dict[Text, tf.Tensor]]:
        if self.model is None:
            logger.debug(
-                "There is no trained model: component is either not trained or "
-                "didn't receive enough training data."
+                f"There is no trained model for {self.__class__.name}: component is either not trained or "


Suggested change

f"There is no trained model for {self.__class__.name}: component is either not trained or "

f"There is no trained model for '{self.__class__.name}': component is either not trained or "

tabergma · 2020-03-19T07:08:57Z

rasa/nlu/classifiers/diet_classifier.py

@@ -826,7 +826,7 @@ def load(

        if not model_dir or not meta.get("file"):
            logger.debug(
-                f"Failed to load model. "
+                f"Failed to load {cls.name}. "


Suggested change

f"Failed to load {cls.name}. "

f"Failed to load '{cls.name}'. "

sara-tagger · 2020-03-19T08:55:55Z

Thanks for submitting a pull request 🚀 @dakshvar22 will take a look at it as soon as possible ✨

@tabergma

* refactor loss and f1 helpers * review comments on featurizers * fix docstrings in components * review comments on lexical_syntactic_featuirzer. * review comments on convert * Update 5292.enhancement.rst * review comments on hugging face components * rename inverted tag and label dicts * remove _find_example_for_tag * remove setting numpy random seed in train * review comments * create no entity tag constant * add type to tf_layers * update constants comment * remove magic numbers probs * fix type of Data in model data * add axis= * add explanatory comments * add RabbitMQ message header to "rasa export" * fix isinstance() calls * check if responses are present. * fix test * review comments * Apply suggestions from code review * update docstring * update method * add comment and type * rename relative lengths * remove batch_tuple_sizes * review comments * review comments * add docstring * add comments to model data * create tmp dir for convert * update type * add comments * change comment * recalculate number of examples after balancing * reorganize methods in model_data * remove num_neg check from ted * update requirements * fix nlu comparison test * update requirements * removed whitespace * update version * Update alt_requirements/requirements_pretrained_embeddings_convert.txt Co-Authored-By: Tobias Wochinger <t.wochinger@rasa.com> * Fixed an issue with AWS persistor - It's possible now to specify region name for AWS S3 - Tests are setting mock region name now * Poetry for Rasa * Missing lockfile * Updated Dockerfile: rasa-sdk added * Travisfile and Dockerfile updates * Updated alt_requirements * Docs fixes * Updates for dependencies + github workflow * Update for documentation.yml * Updated dependencies * CI updates * Updated lockfile * CI update * Updated lockfile * Updates! * Dependencies fix * Updates for release.py and pika dependency * Fixes and updates * Updated dockerfiles * Dockerfiles updates and fixes * Utter actions (RasaHQ#5297) removed utterance actions from examples and documentation * Post-rebase fixes — updated dockerfiles and pyproject.toml * Updated dockerfiles * Updated dockerfiles * Returning back tensorflow_text * Typo fix * maybe make test_evaluation stable * make sure to commit pyproject toml as part of release script * Poetry integration fix: making kafka non-optional * fixed missing kafka dependency * made poetry version script executable * added poetry version to pyproject * [Rasa Open Source 1.8.0 Release] - Updated version of SDK * [Rasa Open Source 1.8.0 Release] - Updated release script - script now uses pyproject.toml to get version of sdk * prepared release of version 1.8.0 * updated dependencies * improved caching * updated the slack client * properly split text messages for socket io channel * mark test_end_to_end_evaluation_script as flaky * do cache rasabaster template * fixed docker builds by excluding configs from docker ignore * marked pretrained model test as flaky * added github token to slack publishing * Fixed dockerifles for 1.8.0 release * Make sure we install rasa in non-editable mode in Dockerfiles * Better way of building in non-editable mode with poetry * Returning pip<20 back * don't rely on restaurantbot * remove incorrect changelog entry * don't train nlu of formbot * update test story * remove deprecation warnings * add comment to story * beginnings of splitting up docker info, needs cleanup * clean up tracker store docs code formatting * fixes bug * fix failing test * added a test for bug as well * add check for failing test again * add changelog * build containers in gh * fix logging statement * update event broker docs * update oov warning * review comment * Apply suggestions from code review Co-Authored-By: ricwo <ric.wkr@gmail.com> * remove optional kafka install instructions * prepared release of version 1.8.1 * fixed tagging * fixed tagging * update wording * add details * updated dep * fix ujson as well * update based on copy suggestions * update code examples * made ujson flexible * docstrings to utils.hugging_face * fix typo * explain token id * add an example for unexpected user input; consitently call domain domain.yml * add an example for unexpected user input; consitently call domain domain.yml * some more examples * links to sections * updates for 1.8; checklists for tutorials; split up tutorials * merge back to one tutorial * move checklists to end * switch to newer 'async' syntax * clarify no-root user usage with sdk image * add changelog entry * fix override config * Update docs/user-guide/building-assistants.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * Update docs/user-guide/building-assistants.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * Update docs/user-guide/building-assistants.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * Update docs/index.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * rebase pr updates * set weight sparsity to 0 in embed intent classifier * update migration guide * Update docs/user-guide/building-assistants.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * Update docs/user-guide/building-assistants.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * change wording in checklists * review comments * fixed a few typos * simplify checklist * checklist updates * Apply suggestions from code review Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * implement some of my changes * rasa open source * review comments, improve links * suggest one-line deployment * edits to deploy in docker-compose page * small improvements to docs * no more docker compose * fix errors * pin tensorflow-estimator to 2.1.0 * add changelog entry * explain dependency pin in pyproject.toml * add class name of calling class to log message to avoid confusion in case this is executed from a subclass * add class name of calling class to log message to avoid confusion in case this is executed from a subclass * changelog for pr RasaHQ#5444 * prepared release of version 1.8.2 * update changelog with missing entries * fix some review comments * sort nlu/story files before training * more review comments * fix tracker store configuration docs * fix types * fix data tests * clarify choosing a tag * clarify actions folder for image building * use release insetad of latest in deploy in docker * use release and version instead of latest in build in docker * version every docker command and restructure a bit * fix build warnings * Update docs/user-guide/docker/building-in-docker.rst Co-Authored-By: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> * use -full tag * clarify how to direct endpoints to service, add a custom tracker store, and fix oracle documentation versioning * fix reference in changelog * fix changelog ... again * suggested changes by @tabergma on RasaHQ#5444 * set default value for weight_sparsity in response selector to 0 * prepared release of version 1.8.3 * more verbose changelog * Updated poetry.lock and dependencies Co-authored-by: Vova Vv <mr.voov@gmail.com> Co-authored-by: Tanja Bergmann <tabergma@gmail.com> Co-authored-by: Melinda Loubser <32034278+melindaloubser1@users.noreply.github.com> Co-authored-by: Tom Bocklisch <tom@rasa.com> Co-authored-by: ricwo <ric.wkr@gmail.com> Co-authored-by: Vladimir Vlasov <vladimir@rasa.com> Co-authored-by: Tobias Wochinger <t.wochinger@rasa.com> Co-authored-by: Alexander Khizov <a.khizov@rasa.com> Co-authored-by: alwx <alwxndr@gmail.com> Co-authored-by: MichaelMarien <marien.mich@gmail.com> Co-authored-by: Will Kearns <kearnsw@users.noreply.github.com> Co-authored-by: Alexander Khizov <degiz@users.noreply.github.com> Co-authored-by: Ella <erohmensing@gmail.com> Co-authored-by: Daksh <d.varshneya@rasa.com> Co-authored-by: melindaloubser1 <melinda.loubser@gmail.com> Co-authored-by: Akela Drissner-Schmid <32450038+akelad@users.noreply.github.com> Co-authored-by: akelad <akela@rasa.com> Co-authored-by: m-vdb <m.verger@rasa.com> Co-authored-by: thomas <t.kober@rasa.com> Co-authored-by: Maxime Vdb <me@maxvdb.com> Co-authored-by: Thomas <th0mas.ko6er@gmail.com>

tttthomasssss added 2 commits March 18, 2020 22:31

add class name of calling class to log message to avoid confusion in …

df53df8

…case this is executed from a subclass

add class name of calling class to log message to avoid confusion in …

de7fc15

…case this is executed from a subclass

tttthomasssss added type:bug 🐛 Inconsistencies or issues which will cause an issue or problem for users or implementors. area:rasa-oss 🎡 Anything related to the open source Rasa framework labels Mar 18, 2020

tttthomasssss added this to the Rasa 1.9 milestone Mar 18, 2020

tttthomasssss requested a review from tabergma March 18, 2020 21:55

tttthomasssss self-assigned this Mar 18, 2020

changelog for pr #5444

c394333

tabergma approved these changes Mar 19, 2020

View reviewed changes

sara-tagger requested a review from dakshvar22 March 19, 2020 08:55

dakshvar22 removed their request for review March 19, 2020 13:18

tabergma mentioned this pull request Mar 23, 2020

Fixed ambiguous logging in DIETClassifier #5443

Closed

4 tasks

Ghostvv mentioned this pull request Mar 23, 2020

Warning about epochs is confusing #5414

Closed

suggested changes by @tabergma on #5444

c0f236d

tttthomasssss merged commit 2b0e516 into 1.8.x Mar 24, 2020

tmbo deleted the logger-bugger-1.8.x branch May 1, 2020 10:13

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixed ambiguous logging in DIETClassifier #5444

Fixed ambiguous logging in DIETClassifier #5444

tttthomasssss commented Mar 18, 2020

CLAassistant commented Mar 18, 2020 •

edited

Loading

tabergma left a comment •

edited

Loading

tabergma Mar 19, 2020

tabergma Mar 19, 2020

tabergma Mar 19, 2020

sara-tagger commented Mar 19, 2020

	f"Please configure the number of '{EPOCHS}' in your configuration file for {self.__class__.name}."
	f"Please configure the number of '{EPOCHS}' in your configuration file for '{self.__class__.name}'."

	f"There is no trained model for {self.__class__.name}: component is either not trained or "
	f"There is no trained model for '{self.__class__.name}': component is either not trained or "

	f"Failed to load {cls.name}. "
	f"Failed to load '{cls.name}'. "

Fixed ambiguous logging in DIETClassifier #5444

Fixed ambiguous logging in DIETClassifier #5444

Conversation

tttthomasssss commented Mar 18, 2020

CLAassistant commented Mar 18, 2020 • edited Loading

tabergma left a comment • edited Loading

Choose a reason for hiding this comment

tabergma Mar 19, 2020

Choose a reason for hiding this comment

tabergma Mar 19, 2020

Choose a reason for hiding this comment

tabergma Mar 19, 2020

Choose a reason for hiding this comment

sara-tagger commented Mar 19, 2020

CLAassistant commented Mar 18, 2020 •

edited

Loading

tabergma left a comment •

edited

Loading