I have been trying for the past few days to train and deploy a custom Smart Reply model using a conversational dataset.
I have just over 30,000 conversations in cloud storage that I've successfully added to a conversational dataset.
My first attempt at training a model took ~8 hours and it was successful. However, after the training step was complete, the deployment step failed with no explanation other than "Internal Server Error."
I tried to redeploy the model multiple times, but always got the same error.
I also attempted all of these steps in both the console and via API - both lead to the same results.
The only other thing that I can think of is that it is related to location, but I have everything set to global.
Are there any specific issues with the contents of the conversational transcripts that are known to cause issues?
Any advice on what I can do to troubleshoot further?
EDIT: After re-running the 'human evaluation' option twice, the model successfully deployed.
LinkedIn
Twitter
