Skip to content

Enhance documentation for advanced experiment runs #35320

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

Open
CFLJacquet wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Enhance documentation for advanced experiment runs #35320

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions content/en/llm_observability/experiments/advanced_runs.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,22 @@
---
title: Advanced Experiment Runs
description: Run experiments multiple times to account for model variability and automate experiment execution in CI/CD pipelines.
description: Run experiments multiple times to account for model variability on a subset of the dataset, and automate experiment execution in CI/CD pipelines.
---

This page discusses advanced topics in running experiments, including [multiple experiment runs](#multiple-runs) and [setting up experiments in CI/CD](#setting-up-your-experiment-in-cicd).

## Run an Experiment on a subset of the Dataset

Check warning on line 8 in content/en/llm_observability/experiments/advanced_runs.md

View workflow job for this annotation

GitHub Actions / vale

Datadog.headings 

'Run an Experiment on a subset of the Dataset' should use sentence-style capitalization.

First, add tags to your dataset records. They can be a unique identifier (e.g `name:test_use_case_1`) or represent a property of the scenario (e.g `difficulty:hard`).

Then, use the `tags` argument of `LLMObs.pull_dataset()` to filter down the dataset to the records you want to run an Experiment on.

Example
```
LLMObs.pull_dataset(dataset_name="my-dataset", tags=["env:prod", "version:1.0"])
Copy link

@chatgpt-codex-connector chatgpt-codex-connector bot Mar 16, 2026

Choose a reason for hiding this comment

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

P2 Badge Assign filtered dataset before running experiment

The example calls LLMObs.pull_dataset(...) without storing its return value, so following these steps can leave users running LLMObs.experiment(..., dataset=dataset) with an unfiltered dataset variable from earlier code. In the subset-run workflow this defeats the intended tag filter and can unexpectedly execute (and bill for) the full dataset.

Useful? React with 👍 / 👎.

```
Finally, run the Experiment as usual.

## Multiple runs

You can run the same experiment multiple times to account for model non-determinism. You can use the [LLM Observability Python SDK][1] or [Experiments API][2] to specify how many iterations to run; subsequently, each dataset record is executed that many times using the same tasks and evaluators.
Expand Down Expand Up @@ -215,4 +227,4 @@

[1]: /llm_observability/instrumentation/sdk?tab=python
[2]: /llm_observability/experiments/api
[3]: https://app.datadoghq.com/llm/experiments
[3]: https://app.datadoghq.com/llm/experiments
Loading