Thanks for your MTurk setup! It's definitely helped me get Turk model evaluations running quickly.
Just wanted to point out an issue I ran into yesterday (3/22) with the nice qa_data_collection MTurk demo. On a fresh ParlAI install, this demo successfully launches the MTurk webpage, but loops infinitely within this while loop, even when I start the task. I think the issue is that it never finds that a valid_agent has started the task. Thus, I just see a gray screen with no text or task.
I have reproduced this issue on 3 separate machines (1 Mac, 2 Linux) and 2 browsers (Chrome and Safari).
Thanks for your MTurk setup! It's definitely helped me get Turk model evaluations running quickly.
Just wanted to point out an issue I ran into yesterday (3/22) with the nice qa_data_collection MTurk demo. On a fresh ParlAI install, this demo successfully launches the MTurk webpage, but loops infinitely within this while loop, even when I start the task. I think the issue is that it never finds that a valid_agent has started the task. Thus, I just see a gray screen with no text or task.
I have reproduced this issue on 3 separate machines (1 Mac, 2 Linux) and 2 browsers (Chrome and Safari).
Specifically, I run:
And here's an example output: