garak.detectors.perspective

Perspective API Detectors

Detectors using the Perspective API

Requires a key set as an environment variable PERSPECTIVE_API_KEY

Get a key here, https://developers.google.com/codelabs/setup-perspective-api#0

class garak.detectors.perspective.Attack_On_Author(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for ATTACK_ON_AUTHOR - Attack on the author of an article or post.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Attack on the author of an article or post.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'ATTACK_ON_AUTHOR'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Attack_On_Commenter(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for ATTACK_ON_COMMENTER - Attack on fellow commenter.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Attack on fellow commenter.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'ATTACK_ON_COMMENTER'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Flirtation(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for FLIRTATION - Pickup lines, complimenting appearance, subtle sexual innuendos, etc.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Pickup lines, complimenting appearance, subtle sexual innuendos, etc.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'FLIRTATION'

tags = ['avid-effect:ethics:E0101', 'avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Identity_Attack(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for IDENTITY_ATTACK - Negative or hateful comments targeting someone because of their identity.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: Negative or hateful comments targeting someone because of their identity.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'IDENTITY_ATTACK'

tags = ['avid-effect:ethics:E0101', 'avid-effect:ethics:E0301', 'avid-effect:ethics:E0302', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Identity_Attack_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for IDENTITY_ATTACK_EXPERIMENTAL - Negative or hateful comments targeting someone because of their identity.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Negative or hateful comments targeting someone because of their identity.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'IDENTITY_ATTACK_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0101', 'avid-effect:ethics:E0301', 'avid-effect:ethics:E0302', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Incoherent(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for INCOHERENT - Difficult to understand, nonsensical.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Difficult to understand, nonsensical.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'INCOHERENT'

tags = ['avid-effect:performance:P0204']

class garak.detectors.perspective.Inflammatory(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for INFLAMMATORY - Intending to provoke or inflame.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Intending to provoke or inflame.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'INFLAMMATORY'

tags = ['avid-effect:ethics:E0301', 'avid-effect:ethics:E0302', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Insult(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for INSULT - Insulting, inflammatory, or negative comment towards a person or a group of people.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: Insulting, inflammatory, or negative comment towards a person or a group of people.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'INSULT'

tags = ['avid-effect:ethics:E0101', 'avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Insult_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for INSULT_EXPERIMENTAL - Insulting, inflammatory, or negative comment towards a person or a group of people.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Insulting, inflammatory, or negative comment towards a person or a group of people.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'INSULT_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Likely_To_Reject(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for LIKELY_TO_REJECT - Overall measure of the likelihood for the comment to be rejected according to the NYT’s moderation.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = "Targets: Overall measure of the likelihood for the comment to be rejected according to the NYT's moderation."

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'LIKELY_TO_REJECT'

tags = []

class garak.detectors.perspective.Obscene(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for OBSCENE - Obscene or vulgar language such as cursing.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Obscene or vulgar language such as cursing.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'OBSCENE'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Profanity(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for PROFANITY - Swear words, curse words, or other obscene or profane language.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: Swear words, curse words, or other obscene or profane language.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'PROFANITY'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Profanity_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for PROFANITY_EXPERIMENTAL - Swear words, curse words, or other obscene or profane language.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Swear words, curse words, or other obscene or profane language.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'PROFANITY_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Severe_Toxicity(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for SEVERE_TOXICITY - A very hateful, aggressive, disrespectful comment or otherwise very likely to make a user leave a discussion or give up on sharing their perspective. This attribute is much less sensitive to more mild forms of toxicity, such as comments that include positive uses of curse words.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: A very hateful, aggressive, disrespectful comment or otherwise very likely to make a user leave a discussion or give up on sharing their perspective. This attribute is much less sensitive to more mild forms of toxicity, such as comments that include positive uses of curse words.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'SEVERE_TOXICITY'

tags = ['avid-effect:ethics:E0301']

class garak.detectors.perspective.Severe_Toxicity_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for SEVERE_TOXICITY_EXPERIMENTAL - A very hateful, aggressive, disrespectful comment or otherwise very likely to make a user leave a discussion or give up on sharing their perspective. This attribute is much less sensitive to more mild forms of toxicity, such as comments that include positive uses of curse words.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: A very hateful, aggressive, disrespectful comment or otherwise very likely to make a user leave a discussion or give up on sharing their perspective. This attribute is much less sensitive to more mild forms of toxicity, such as comments that include positive uses of curse words.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'SEVERE_TOXICITY_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0301', 'avid-effect:ethics:E0302', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Sexually_Explicit(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for SEXUALLY_EXPLICIT - Contains references to sexual acts, body parts, or other lewd content.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Contains references to sexual acts, body parts, or other lewd content.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'SEXUALLY_EXPLICIT'

tags = ['avid-effect:ethics:E0301']

class garak.detectors.perspective.Spam(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for SPAM - Irrelevant and unsolicited commercial content.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Irrelevant and unsolicited commercial content.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'SPAM'

tags = ['avid-effect:performance:P0204']

class garak.detectors.perspective.Threat(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for THREAT - Describes an intention to inflict pain, injury, or violence against an individual or group.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: Describes an intention to inflict pain, injury, or violence against an individual or group.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'THREAT'

tags = ['avid-effect:ethics:E0101', 'avid-effect:ethics:E0301', 'avid-effect:performance:P0400']

class garak.detectors.perspective.Threat_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for THREAT_EXPERIMENTAL - Describes an intention to inflict pain, injury, or violence against an individual or group.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Describes an intention to inflict pain, injury, or violence against an individual or group.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'THREAT_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Toxicity(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for TOXICITY - A rude, disrespectful, or unreasonable comment that is likely to make people leave a discussion.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'ar,zh,cs,nl,en,fr,de,hi,hi-Latn,id,it,ja,ko,pl,pt,ru,es,sv'

description = 'Targets: A rude, disrespectful, or unreasonable comment that is likely to make people leave a discussion.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'TOXICITY'

tags = ['avid-effect:ethics:E0301']

class garak.detectors.perspective.Toxicity_Experimental(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for TOXICITY_EXPERIMENTAL - A rude, disrespectful, or unreasonable comment that is likely to make people leave a discussion.

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: A rude, disrespectful, or unreasonable comment that is likely to make people leave a discussion.'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'TOXICITY_EXPERIMENTAL'

tags = ['avid-effect:ethics:E0301', 'avid-effect:performance:P0401']

class garak.detectors.perspective.Unsubstantial(config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

Bases: Detector

Perspective API interface for UNSUBSTANTIAL - Trivial or short comments

ENV_VAR = 'PERSPECTIVE_API_KEY'

active = False

bcp47 = 'en'

description = 'Targets: Trivial or short comments'

detect(attempt)

Takes a list of Attempts; classifies them; returns a list of results in the range 0.0-1.0 where 0.0 represents no hit, 1.0 represents successful hit (i.e. vulnerability found).

endpoint = 'UNSUBSTANTIAL'

tags = ['avid-effect:performance:P0204']

garak.detectors.perspective.constructor(self, config_root=<module 'garak._config' from '/home/docs/checkouts/readthedocs.org/user_builds/garak/checkouts/latest/docs/source/../../garak/_config.py'>)

garak.detectors.perspective.detect(self, attempt)