Arize AI Community Icon

Home
Events

Arize AX Releases
Arize AX Support
Arize News
Arize Observe
Discussions
Introductions
Phoenix Support

Powered by Tightknit

Phoenix Support

Phoenix Support

Issue with Repetitions Parameter in run_experiment from phoenix.experiments

·Oct 23, 2025 07:41 PM·

Hello! Why can't I use the repetitions parameter when using the run_experiment from phoenix.experiments ? It works if I import the function from phoenix.client

3 comments

· Sorted by Oldest

RunLLM
APP
·
·
The repetitions parameter is only supported in the new experiments API via phoenix.client (e.g., client.experiments.run_experiment or from phoenix.client.experiments import run_experiment). The legacy phoenix.experiments module does not support the repetitions parameter—this feature was added in Phoenix 11.37.0+ and is not available in the old API [docs, changelog, source].To use repetitions, switch to the new client API as shown below: LBKSfrom phoenix.client import Client client = Client() dataset = client.datasets.get_dataset(dataset="my-dataset") experiment = client.experiments.run_experiment( dataset=dataset, task=my_task, experiment_name="greeting-experiment", repetitions=3 )LBKSWould you like more detail on migrating to the new API or on how repetitions work?Sources: - How to use repetitions (docs) - Phoenix client experiments source - changelog
💡 Hint: Mention RunLLM in the thread for followups.
José M.
·
·
RunLLM so is recommended to not use the legacy phoenix.experiments in favor of the new api, right?
RunLLM
APP
·
·
Yes, it is recommended to use the new phoenix.client API for experiments, as it supports new features like repetitions and ongoing enhancements. The legacy phoenix.experiments module is being phased out and lacks support for recent improvements [docs, release notes].Would you like guidance on migrating your code to the new API?Sources: - How to use repetitions (docs) - Release notes: Experiments module in phoenix-client
💡 Hint: Mention RunLLM in the thread for followups.