What is going on with fine tuned model usage? It seems like since about 2 weeks ago, any calls (once an HOUR) attempting to do a single query against a custom model (in this case a Curie trained model), error or times out at least 50% of the time.
It’s usually some variation of error 429 (too many requests, model is currently being loaded, server is over loaded at the moment).
Is this normal? If it is I’m going to have to completely rethink using custom models at all.
EDIT: I tried deleting/retraining the model as well to see if that was the issue, but it didn’t help. This app also has a grand total of our internal dev team, and no one else, so I highly doubt its a throttling thing.