In documents, add a new hallucination detection model
Add the ability to highlight identified hallucinations using a gradient color schema based on the score
Alternatives Considered
No response
Additional Context
No response
Originally created by @thistleknot on GitHub (Jul 15, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/15742
### Check Existing Issues
- [x] I have searched the existing issues and discussions.
### Problem Description
Add the ability to highlight using a gradient hallucinations
default proposed model
https://huggingface.co/KRLabsOrg/lettucedect-base-modernbert-en-v1/tree/main
### Desired Solution you'd like
In documents, add a new hallucination detection model
Add the ability to highlight identified hallucinations using a gradient color schema based on the score
### Alternatives Considered
_No response_
### Additional Context
_No response_
Use an LLM (that will hallucinate itself) to detect hallucinations of another LLM?
<!-- gh-comment-id:3074599719 -->
@Classic298 commented on GitHub (Jul 15, 2025):
what?
Use an LLM (that will hallucinate itself) to detect hallucinations of another LLM?
<!-- gh-comment-id:3074877293 -->
@thistleknot commented on GitHub (Jul 15, 2025):
Check out lettucedetect
Its a model when given context and response will identify text in the
response that wasn't directly sourced from the context
Similar to ragas metrics (in terms of providing response and context to
evaluate)
On Tue, Jul 15, 2025, 10:38 AM Classic298 ***@***.***> wrote:
> *Classic298* left a comment (open-webui/open-webui#15742)
> <https://github.com/open-webui/open-webui/issues/15742#issuecomment-3074599719>
>
> what?
>
> Use an LLM (that will hallucinate itself) to detect hallucinations of
> another LLM?
>
> —
> Reply to this email directly, view it on GitHub
> <https://github.com/open-webui/open-webui/issues/15742#issuecomment-3074599719>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ABHKKOQQXTIWXSVD4ODLTJ33IU37XAVCNFSM6AAAAACBST62TCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTANZUGU4TSNZRHE>
> .
> You are receiving this because you authored the thread.Message ID:
> ***@***.***>
>
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @thistleknot on GitHub (Jul 15, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/15742
Check Existing Issues
Problem Description
Add the ability to highlight using a gradient hallucinations
default proposed model
https://huggingface.co/KRLabsOrg/lettucedect-base-modernbert-en-v1/tree/main
Desired Solution you'd like
In documents, add a new hallucination detection model
Add the ability to highlight identified hallucinations using a gradient color schema based on the score
Alternatives Considered
No response
Additional Context
No response
@Classic298 commented on GitHub (Jul 15, 2025):
what?
Use an LLM (that will hallucinate itself) to detect hallucinations of another LLM?
@thistleknot commented on GitHub (Jul 15, 2025):
Check out lettucedetect
Its a model when given context and response will identify text in the
response that wasn't directly sourced from the context
Similar to ragas metrics (in terms of providing response and context to
evaluate)
On Tue, Jul 15, 2025, 10:38 AM Classic298 @.***> wrote: