Example output of just the huwiki models.
ActivePublic
Actions

Authored by Halfak on Feb 13 2019, 8:34 PM.

Tags

Referenced Files

	F28211266: Example output of just the huwiki models.
	Feb 13 2019, 8:34 PM

Subscribers

None

	$ ./utility generate_make --config test-config
	# This file is built automatically using cg.py file and Makefile.j2
	# Any change you make on this file will be lost in the next run.

	# Remove target files after command failure.
	.DELETE_ON_ERROR:

	models: \
	huwiki_models

	tuning_reports: \
	huwiki_tuning_reports

	touch:
	touch datasets/*
	touch models/*

	include Makefile.manual


	############################# Hungarian Wikipedia ################################
	datasets/huwiki.human_labeled_revisions.raw.5k_2016.json:
	./utility fetch_labels \
	https://labels.wmflabs.org/campaigns/huwiki/12/ > $@
	datasets/huwiki.sampled_revisions.40k_2016.json:
	wget -qO- http://quarry.wmflabs.org/run/79645/output/0/json-lines?download=true > $@

	datasets/huwiki.autolabeled_revisions.40k_2016.json: \
	datasets/huwiki.sampled_revisions.40k_2016.json
	cat $< \| \
	./utility autolabel --host=https://hu.wikipedia.org \
	--trusted-groups=sysop,oversight,trusted,bot,rollbacker,checkuser,abusefilter,bureaucrat,editor,templateeditor,interface-editor \
	--trusted-edits=1000 \
	--revert-radius=3 \
	--revert-window=48 \
	--verbose > $@

	datasets/huwiki.revisions_for_review.5k_2016.json: \
	datasets/huwiki.autolabeled_revisions.40k_2016.json
	( \
	cat datasets/huwiki.autolabeled_revisions.40k_2016.json \| grep '"needs_review": (true\|"True") \| \
	shuf -n 2500; \
	cat datasets/huwiki.autolabeled_revisions.40k_2016.json \| grep '"needs_review": (false\|"False") \| \
	shuf -n 2500 \
	) \| shuf > $@

	datasets/huwiki.labeled_revisions.40k_2016.json: \
	datasets/huwiki.autolabeled_revisions.40k_2016.json \
	datasets/huwiki.human_labeled_revisions.5k_2016.json ./utility merge_labels $^ > $@

	datasets/huwiki.labeled_revisions.w_cache.40k_2016.json: \
	datasets/huwiki.labeled_revisions.20k_2016.json
	revscoring extract \
	editquality.feature_lists.huwiki.damaging \
	editquality.feature_lists.huwiki.goodfaith \
	--host https://hu.wikipedia.org \
	--extractors $(max_extractors) \
	--verbose > $@

	tuning_reports/huwiki.damaging.md: \
	datasets/huwiki.labeled_revisions.w_cache.40k_2016.json
	cat $< \| \
	revscoring tune \
	config/classifiers.params.yaml \
	editquality.feature_lists.huwiki.damaging \
	damaging \
	roc_auc.labels.true \
	--label-weight $(damaging_label_weight) \
	--pop-rate "true=0.01" \
	--pop-rate "false=0.99" \
	--center --scale \
	--cv-timeout 60 \
	--debug > $@

	models/huwiki.damaging.gradient_boosting.model: \
	datasets/huwiki.labeled_revisions.w_cache.40k_2016.json
	cat $< \| \
	revscoring cv_train
	damaging \
	--version=$(damaging_major_minor). \
	-p 'learning_rate=0.01' \
	-p 'max_depth=7' \
	-p 'max_features="log2"' \
	-p 'n_estimators=700' \
	--label-weight $(damaging_label_weight) \
	--pop-rate "true=0.01" \
	--pop-rate "false=0.99" \
	--center --scale > $@

	revscoring model_info $@ > model_info/huwiki.damaging.md

	tuning_reports/huwiki.goodfaith.md: \
	datasets/huwiki.labeled_revisions.w_cache.40k_2016.json
	cat $< \| \
	revscoring tune \
	config/classifiers.params.yaml \
	editquality.feature_lists.huwiki.goodfaith \
	goodfaith \
	roc_auc.labels.true \
	--label-weight $(goodfaith_label_weight) \
	--pop-rate "true=0.99" \
	--pop-rate "false=0.010000000000000009" \
	--center --scale \
	--cv-timeout 60 \
	--debug > $@

	models/huwiki.goodfaith.gradient_boosting.model: \
	datasets/huwiki.labeled_revisions.w_cache.40k_2016.json
	cat $< \| \
	revscoring cv_train
	goodfaith \
	--version=$(goodfaith_major_minor). \
	-p 'learning_rate=0.01' \
	-p 'max_depth=7' \
	-p 'max_features="log2"' \
	-p 'n_estimators=700' \
	--label-weight $(goodfaith_label_weight) \
	--pop-rate "true=0.99" \
	--pop-rate "false=0.010000000000000009" \
	--center --scale > $@

	revscoring model_info $@ > model_info/huwiki.goodfaith.md


	huwiki_models: \
	models/huwiki.goodfaith.gradient_boosting.model \
	models/huwiki.damaging.gradient_boosting.model

	huwiki_tuning_reports: \
	tuning_reports/huwiki.goodfaith.md \
	tuning_reports/huwiki.damaging.md

Event Timeline

Halfak created this paste.Feb 13 2019, 8:34 PM

Halfak mentioned this in T190968: Simplify and modularize the Makefile template.Feb 13 2019, 8:36 PM

Example output of just the huwiki models. ActivePublicActions

Event Timeline

Example output of just the huwiki models.
ActivePublic
Actions