A question I should know the answer to, so I'm sure there's a few folks who could help with this one. 🙂

If our data set has 10,000 different unique_ids, would that use up all our monthly API calls in one go, or would that only count as 1 API call?

I'm actually not sure at all, that's why I didn't answer 😅

I feel better for not knowing then! 😅

I think that’s from the python sdk, @azul (she/her) (nixtla) do you know if the sdk makes multiple calls to the api with this method?

thanks! here's a response: thanks for using timegpt. if you pass the entire dataset, that will count as 1 api call, without considering the number of unique_ids. in some cases, specially when you have a lot of unique_ids, the api can throw an error due to the size of the data. in that case, you can use

num_partitions=5

for example, and the sdk will call the API 5 times, each of one will have a subset of unique_ids.

So question for my own understanding: we don't partition automatically if the number of unique ids is very large? is it something worth adding?

no, we don't partition automtically 😞 yes, i thinkg it might be a good addition, could please you open an issue on nixtla/nixtla? so we can track that enhacement for the next sprints, thank you marco! i think it's a good idea

Thanks for the explanation @azul (she/her) (nixtla) Tbh, I also wasn't 100% sure 😅

Thanks @azul (she/her) (nixtla)! I'll think how we might format this as a question in our FAQ too.

nice @Tracy Teal! that would be very helpful