Documentation for Custom LLM-as-a-judge Evaluations #32018
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mtullalizardi
force-pushed
the
miguel.tullalizardi/byop-documentation
branch
from
a0d0a2b to
5036d05
Compare
Preview links (active after the
|
|
Hi! Added a ticket for this one here. |
gsvigruha
reviewed
Oct 7, 2025
config/_default/menus/main.es.yaml
Outdated
| parent: llm_obs | ||
| identifier: llm_obs_evaluations | ||
| weight: 4 | ||
| - name: Custom LLM-as-a-Judge |
There was a problem hiding this comment.
you don't need to change these, just the english one
config/_default/menus/main.fr.yaml
Outdated
| parent: llm_obs | ||
| identifier: llm_obs_evaluations | ||
| weight: 4 | ||
| - name: Custom LLM-as-a-Judge |
config/_default/menus/main.ja.yaml
Outdated
| parent: llm_obs | ||
| identifier: llm_obs_evaluations | ||
| weight: 4 | ||
| - name: Custom LLM-as-a-Judge |
config/_default/menus/main.ko.yaml
Outdated
| parent: llm_obs | ||
| identifier: llm_obs_evaluations | ||
| weight: 4 | ||
| - name: Custom LLM-as-a-Judge |
|
|
||
| ### Custom LLM-as-a-Judge Evaluations | ||
|
|
||
| [Custom LLM-as-a-Judge Evaluations][1] allow you to define your own evaluation logic using natural language prompts. You can create custom evaluations to assess subjective or objective criteria—like tone, helpfulness, or factuality—and run them at scale across your traces and spans. |
There was a problem hiding this comment.
spaces around dashes criteria - like tone, helpfulness, or factuality - and
|
|
||
| ## Overview | ||
|
|
||
| Custom LLM-as-a-Judge Evaluations let you define your own evaluation logic to automatically assess your LLM applications. You can use natural language prompts to capture subjective or objective criteria—like tone, helpfulness, or factuality—and run them at scale across your traces and spans. |
There was a problem hiding this comment.
criteria - like tone, helpfulness, or factuality - and
| You define: | ||
| - The criteria (via prompt text) | ||
| - What is evaluated (e.g., a span's output) | ||
| - The model (e.g., GPT-4o) |
There was a problem hiding this comment.
nit: GPT-4o - format as code using backticks
| - The criteria (via prompt text) | ||
| - What is evaluated (e.g., a span's output) | ||
| - The model (e.g., GPT-4o) | ||
| - The output type (boolean, numeric score, or categorical label) |
There was a problem hiding this comment.
same format boolean, score and categorical as code since they refer to code concepts in our app
| - The model (e.g., GPT-4o) | ||
| - The output type (boolean, numeric score, or categorical label) | ||
|
|
||
| Datadog then runs this evaluation logic automatically against your spans, recording results as structured metrics that you can query, visualize, and monitor. |
There was a problem hiding this comment.
maybe structured evaluation metrics. "metrics" means this metric at datadog, we brand these things as evaluations on the UI or in the query language
There was a problem hiding this comment.
Changed the wording to just "...recording results for you to query, visualize, and monitor."
| - **Score** – Numeric rating (e.g., 1–5 scale for helpfulness) | ||
| - **Categorical** – Discrete labels (e.g., "Good", "Bad", "Neutral") | ||
|
|
||
| The schema ensures your results are structured for querying and dashboarding. For Anthropic and Bedrock models, only Boolean output types are allowed. |
There was a problem hiding this comment.
we should mention
- the fact that structured output is openAIs structured output and it needs to be edited by the user
- the keyword search logic for anthropic
gsvigruha
approved these changes
Oct 7, 2025
|
I pushed a few edits:
|
cswatt
approved these changes
Oct 7, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do? What is the motivation?
This PR adds documentation for the LLM Observability "Custom LLM-as-a-judge Evaluations" feature. This is a new feature we are GAing and is already deployed. We are waiting on getting the documentation up to turn on the feature flag.
Merge instructions
Merge readiness:
For Datadog employees:
Your branch name MUST follow the
<name>/<description>convention and include the forward slash (/). Without this format, your pull request will not pass CI, the GitLab pipeline will not run, and you won't get a branch preview. Getting a branch preview makes it easier for us to check any issues with your PR, such as broken links.If your branch doesn't follow this format, rename it or create a new branch and PR.
[6/5/2025] Merge queue has been disabled on the documentation repo. If you have write access to the repo, the PR has been reviewed by a Documentation team member, and all of the required checks have passed, you can use the Squash and Merge button to merge the PR. If you don't have write access, or you need help, reach out in the #documentation channel in Slack.
Additional notes