Skip to content

[GSK-1283] Additional LLM issue detectors #1190

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 96 commits into from
Jul 25, 2023
Merged

[GSK-1283] Additional LLM issue detectors #1190

merged 96 commits into from
Jul 25, 2023

Conversation

@mathieuroques
Copy link
Contributor

Adding 3 issues to llm scan:

  • Gender bias
  • Unethical bias
  • Minority bias

@mathieuroques mathieuroques requested a review from mattbit June 21, 2023 15:13
@linear
Copy link

linear bot commented Jun 21, 2023

GSK-1283 LLM Issues 2.1

Add these issues to LLM issues

  • Gender neutrality test
  • Non-ethical behaviour
  • Security
  • Racism

mathieuroques and others added 26 commits June 21, 2023 17:17
@mathieuroques
fixed regex issue
f474fb9
@mathieuroques
Added detoxify lib to the project
318a36b
@mathieuroques
Added minority bias scan
0c60601
@mathieuroques
bug fix in minority bias detector
75a6247
@mathieuroques
Merge branch 'main' into feature/gsk-1283-llm-issues-21
44bc1f1
@mathieuroques
Formatted imports
f487fbf
@mathieuroques
Merge branch 'feature/gsk-1283-llm-issues-21' of https://github.com/G…
0e2837b 
…iskard-AI/giskard into feature/gsk-1283-llm-issues-21
@mathieuroques
Fixed bug in if statement
5d4f2bf
@mathieuroques
debug all the bias detector
7aac16c
@mathieuroques
bug fix
501e615
@mathieuroques
bug fix
e2d3ef0
@mathieuroques
Display update
700c801
@mathieuroques
Stereotype generation title
1e76185
@mathieuroques
Replaced Dataset by DataFrame
4fd5c0a
@mathieuroques
Fixed metric bug
fce8aae
@mathieuroques
Display fix
6345fb4
@mathieuroques
Add prompt to minority
de012ed
@mattbit
Merge branch 'main' into feature/gsk-1283-llm-issues-21
e674ada
@mathieuroques
Removed prompt from minority
5576bc2
@mathieuroques
Merge branch 'feature/gsk-1283-llm-issues-21' of https://github.com/G…
fb8963b 
…iskard-AI/giskard into feature/gsk-1283-llm-issues-21
@mathieuroques
Added metric to gender bias display
fd62170
@mathieuroques
Added prompt to minority
08511f9
@mathieuroques
display metric for gender
7c25e9f
@mathieuroques
Bug fix
7f9a56b
@mathieuroques
display percentage
362386c
@mathieuroques
Changed description gender bias
88fb490
@mattbit mattbit assigned mattbit and unassigned mathieuroques Jul 21, 2023
@mattbit
Copy link
Member

mattbit commented Jul 21, 2023

I will take this over, there are still things to fix.

@mattbit
Copy link
Member

mattbit commented Jul 21, 2023

Ok @mathieuroques, we’re almost there, can you:

  1. Run pdm lock and commit the updated lock file (I fixed the dependencies problems)
  2. Update the tutorial notebook
  3. Do some manual testing to ensure everything is working as expected.

@mattbit
Copy link
Member

mattbit commented Jul 21, 2023

@mathieuroques ?

mattbit
mattbit approved these changes Jul 24, 2023
View reviewed changes
Copy link
Member

@mattbit mattbit left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok for me

@mattbit mattbit requested a review from rabah-khalek July 24, 2023 16:02
@sonarqubecloud
Copy link

Kudos, SonarCloud Quality Gate passed!    Quality Gate passed

Bug A 0 Bugs
Vulnerability A 0 Vulnerabilities
Security Hotspot A 0 Security Hotspots
Code Smell A 0 Code Smells

87.8% 87.8% Coverage
0.0% 0.0% Duplication

rabah-khalek
rabah-khalek approved these changes Jul 25, 2023
View reviewed changes
Copy link
Contributor

@rabah-khalek rabah-khalek left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cool stuff.

python-client/giskard/scanner/llm/gender_stereotype_detector.py
Comment on lines +88 to +89
has_masculine_pronouns = bool(re.search(r"\b(he|him|his)\b", sentence.lower()))
has_feminine_pronouns = bool(re.search(r"\b(she|her)\b", sentence.lower()))
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are we only supporting English? I thought french as well.

Copy link
Member

@mattbit mattbit Jul 25, 2023 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yep we basically support English only for now…

@mattbit mattbit merged commit 5e2a45d into main Jul 25, 2023
@mattbit mattbit deleted the feature/gsk-1283-llm-issues-21 branch July 25, 2023 11:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mattbit mattbit mattbit approved these changes

+1 more reviewer

@rabah-khalek rabah-khalek rabah-khalek approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

@mattbit mattbit

Labels

safe for build

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@mathieuroques @mattbit @rabah-khalek @luca-martial