🙌

❤️

hey y'all. @Max (Nixtla) sent me an invite to the TimeGPT beta. Gonna report problems (small and big) as I spot them. 🧵

welcome @Daniel Lee! 🎉

can anyone give me brief about what is timegpt??

is there any way we can vie the pre-print? I’m quite interested!

@Bartosz Bohaterewicz, I will move your request to this channel.

Hello.. I have read the TimeGPT docs and it says this:

TimeGPT, requires a certain number of initial observations to generate reliable forecasts. Therefore, while interpreting the output, it's important to be aware that the first few observations serve as the basis for the model's predictions and are not themselves predicted values.

My question is, what is the minimal number of observations? how many? Thank you

<!here> having trouble getting into https://dashboard.nixtla.io/ - anyone else expecting this. Attempting to login via Google Auth

Hey guys, hope you are doing well! I have submitted an interest ticket via the website https://www.nixtla.io/ in order to use TimeGPT, but there is no response to me. Sorry to bother you, I already applied for an open token serval days before, but i did work properly(got 429 status code). So I want to know how I can use it normally. Thanks in advance.

Hey team, I followed this instruction to set up my account and get API key, but I get the following error

ApiError: status_code: 429, body: {'data': None, 'message': 'Too many requests', 'details': 'You have reached your request limit, email <mailto:ops@nixtla.io|ops@nixtla.io> to continue using the API', 'code': 'A20', 'requestID': 'UA4KMLDWE2', 'support': 'If you have questions or need support, please email <mailto:ops@nixtla.io|ops@nixtla.io>'}

when running the sample code from https://docs.nixtla.io/docs/quickstart . Can anyone help with this?

Hi team, great job on the TimeGPT beta release presentation! I was wondering if more information was available on how exogenous data series are used in the model? I saw on the presentation and on https://nixtla.github.io/nixtla/timegpt.html that data series besides the target can be fed into TimeGPT, and importances are computed, but how do they actually help shape the output? To my limited knowledge secondary data streams are not a feature of other GPT systems like ChatGPT, are there any published papers on the techniques you based the design on you could direct me to? Does it help to identify the exogenous columns (e.g. Price, Weather, Promotion etc.), so the system can associate it with training examples? What sort of exogenous columns were used in training? If I'm feeding the system a large number of series and fine-tuning, will it learn (for example) categorical columns and how they relate different series to one another? Should categorical variables be one-hot encoded or can it handle them like LightGBM?

Hi i was given acces to timeGPT, but i dont understand what are my credentials

Hello! I would like to try timeGPT. I have been using and contributing to Prophet for a long time. I would like to contribute if necessary. I would appreciate it if you could give me the access key.

Hi Team, congrats on developing such a great library 🙂 . I have recently started playing with it and I got an error when trying to use 'Country Holidays':

Exception: You have to install additional libraries to use holidays, please install them using

pip install "nixtlats[date_extras]"

I installed the requirements as stated in the error.

pip install "nixtlats[date_extras]"

But I still get the same error.

Hi Team, I keep getting:

Copy code

WARNING:nixtlats.timegpt:The specified horizon "h" exceeds the model horizon. This may lead to less accurate forecasts. Please consider using a smaller horizon.

Are there any rules on the max horizon? Does it depend on each dataset? Is there any way to know what is the mentioned model horizon? Thanks.

Hey team, I followed this instruction to set up my account and get API key, but I get the following error

{
  "data": null,
  "message": "Too many requests",
  "details": "You have reached your request limit, email ops@nixtla.io to continue using the API",
  "code": "A20",
  "requestID": "QN2M78TJA9",
  "support": "If you have questions or need support, please email ops@nixtla.io"
}

when running the sample code from https://docs.nixtla.io/docs/quickstart . Can anyone help with this?

Hello! I have submitted an application to the waiting list of TimeGPT. I wonder how long the review will take. Thanks!

@Max (Nixtla) Hello Max, I submitted my application just now, could you review it and give me the access to TimeGPT

Thanks!