search

Nixtla / nixtla

TimeGPT-1: production ready pre-trained Time Series Foundation Model for forecasting and anomaly detection. Generative pretrained transformer for time series trained on over 100B data points. It's capable of accurately predicting various domains such as retail, electricity, finance, and IoT with just a few lines of code 🚀.
https://docs.nixtla.io
Other
2.31k stars 187 forks source link

Connection to TimeGEN model deployed in Azure AI studio fails #421

Closed andreasgeorgklauser closed 2 months ago

andreasgeorgklauser commented 4 months ago

Hi,

I have successfully deployed the TimeGEN-1 model in Azure AI studio. Then I'm trying to connect to his model following the documentation: https://docs.nixtla.io/docs/deployment-azureai It states to use the model "azureai" in case when connecting to Azure AI. However, this model is not supported, the forecast method throws an error.

So I'm trying the code below without specifying a model in the forecast method. Then I'm getting a different exception:

Exception has occurred: ConnectError [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self-signed certificate in certificate chain (_ssl.c:1000) ssl.SSLCertVerificationError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self-signed certificate in certificate chain (_ssl.c:1000) During handling of the above exception, another exception occurred: httpcore.ConnectError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self-signed certificate in certificate chain (_ssl.c:1000)

The code I'm using is:

import pandas as pd from nixtla import NixtlaClient

nixtla_client = NixtlaClient(api_key = '(I pasted the api-key from Azure AI studio here)', base_url= '(I pasted the endpoint from Azure AI studio here)')

df = pd.read_csv('https://raw.githubusercontent.com/Nixtla/transfer-learning-time-series/main/datasets/electricity-short.csv')

fcst_df = nixtla_client.forecast(df, h=24, level=[80, 90])

nixtla_client.plot(df, fcst_df, time_col='timestamp', target_col='value', level=[80, 90])

Best regards,

Andreas

elephaint commented 3 months ago

At this moment you shouldn't specify the model in Azureai (our readme is incorrect with regards to that at the moment, unfortunately).

Not sure why it then gives the SSL error, though. Are you sure the endpoint has the correct format and the certificate is correct?

elephaint commented 3 months ago

Hi Andreas,

Just wanted to follow up on this - did you take the verification steps I suggested? Are you still experiencing the issue?

mcasalaina commented 3 months ago

Similar issue here, although I don't get an SSL error, I get "Key not found, verify in https://dashboard.nixtla.io/" even though I specified an Azure AI endpoint and key. What's the deal with this? The docs are clear as mud.

mcasalaina commented 3 months ago

(I also tried specifying a model as "azureai" but that didn't work either, saying the only supported models were TimeGEN-1 and TimeGEN-1-long-horizon)

mcasalaina commented 3 months ago

Update: seems like it worked when I defined the environment variables literally as in the doc - base_url = os.environ["AZURE_AI_NIXTLA_BASE_URL"] api_key = os.environ["AZURE_AI_NIXTLA_API_KEY"].

mergenthaler commented 3 months ago

@marcopeix, it would be great if we made this clearer in the docs and also integrate some test in our CI/CD for Azure.

github-actions[bot] commented 2 months ago

This issue has been automatically closed because it has been awaiting a response for too long. When you have time to to work with the maintainers to resolve this issue, please post a new comment and it will be re-opened. If the issue has been locked for editing by the time you return to it, please open a new issue and reference this one.