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

Where to Find EvaluationResult Explanation in Phoenix UI | Arize AI Community

Phoenix Support

Where to Find EvaluationResult Explanation in Phoenix UI

·Aug 13, 2025 07:53 PM

can anyone point to me where is the explanation in the EvaluationResult surfaced on phoenix UI, please? https://github.com/Arize-ai/phoenix/blob/3b320d0d2fff4718c7565719b2656db23b440c7a/[…]hoenix-client/src/phoenix/client/resources/experiments/types.py

17 comments

· Sorted by Oldest

RunLLM
APP
·
The explanation field in EvaluationResult is surfaced in the Phoenix UI as part of the evaluation results for experiments and annotations. When you run experiments or log evaluations, the UI displays the explanation alongside the label and score for each evaluated example, allowing users to drill down into individual cases and understand why a particular label was assigned by the evaluator. This is visible in the experiment results table and in the detailed view for each example or span in the UI, as shown in the Phoenix documentation and confirmed by the user guide.Would you like more detail on how to access or visualize these explanations in the UI?Sources: - user guide
💡 Hint: Mention RunLLM in the thread for followups.
Zhuyi X.
·
yeah, where can I find the explanation on UI?
Zhuyi X.
·
I'm using run_experiment should it be somewhere in this screenshot?
Zhuyi X.
·
execution_accuracy is one of the evaluator output
Zhuyi X.
·
Or do you have an example screenshot where it should show up?
Zhuyi X.
·
what's application traces in the context of experiments?
Zhuyi X.
·
Could you show me an example where you can actually see the explanation?
Zhuyi X.
·
is that the same as Experiment Run Trace page? my annotations is empty
Zhuyi X.
·
I'm just using run_experiments, how to get the traces into app traces?
Zhuyi X.
·
I feel I do need to see the explanation to continue using phoenix for experiment, otherwise, it's not useful if I can't debug with it. or is there any way to get the explanation?
Zhuyi X.
·
ah, sounds good, I'll try that. thanks!
Zhuyi X.
·
but still much more preferrably, it could show up in UI already.
Zhuyi X.
·
I'm switching to use phoenix.client.experiments.run_experiment now
Zhuyi X.
·
just supporting phoenix.client.experiments.run_experiment is good for my case
Zhuyi X.
·
now I see the eval is due to some perm error