# Evaluations - fails when importing from Logs

**URL:** <https://community.openai.com/t/evaluations-fails-when-importing-from-logs/1361605>\
**Category:** Bugs\
**Tags:** evals\
**Created:** [October 8, 2025, 7:10pm UTC](https://community.openai.com/t/evaluations-fails-when-importing-from-logs/1361605 "2025-10-08T19:10:45Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Dev\_Brightn](https://avatars.discourse-cdn.com/v4/letter/d/c4cdca/32.png) [@Dev\_Brightn](https://community.openai.com/u/Dev_Brightn)\
**Post date:** [October 8, 2025, 7:10pm UTC](https://community.openai.com/t/evaluations-fails-when-importing-from-logs/1361605/1 "2025-10-08T19:10:45Z")

</div>

Hey everyone,

I’m using OpenAI’s platform for Evaluations, and I’m running into an issue. I’m trying to create a new evaluation by importing data from my Response Logs. After selecting the date range, setting up my test criteria, and hitting “Run,” I get the following error for every single row:

**“Invalid value: ‘input\_text’. Supported values are: ‘output\_text’ and ‘refusal’.”**

A few things to note:

- Since the data was imported from logs, I can’t modify the column names.

- I tried using different variables in the test criteria prompt — `{{item.input}}`, `{{item.output}}`, `{{sample.output_json}}`, `{{sample.output_text}}` — but none of them worked.

- The platform doesn’t allow downloading the logs to adjust them manually.

Has anyone else run into this issue or found a workaround?