Training is extremely slow on Gluonts [Torch] #3082
Replies: 6 comments 1 reply
-
@kashif @goriccardo @jaheba @tomzx anyone please help me |
Beta Was this translation helpful? Give feedback.
-
@khawar-islam so it could be that your dataset is a single huge time series? is that the intention? |
Beta Was this translation helpful? Give feedback.
-
so I suspect you have a single huge time series and thus the dataset operations are taking too long, namely the data-time feature creation etc. My suggestion is to split the training data up into smaller (say 1 week or 1 month) time series with the appropriate |
Beta Was this translation helpful? Give feedback.
-
also I noticed that you have one thing to keep in mind with these covariates is that they need to be know for all the time-points you wish to predict into the future |
Beta Was this translation helpful? Give feedback.
-
Also, I should remind you not to tag random folks into your issue... this is an open-source project where folks volunteer their time for free and I do not appreciate this approach as it wastes our collective time. |
Beta Was this translation helpful? Give feedback.
-
Dear @kashif thank you for your suggestion, I removed
Any comment on results? |
Beta Was this translation helpful? Give feedback.
-
Description
I am quite frustrating because I am training a model and training is very very slow on RTX 3080. I am training on 500 CSV files. If anyone can help me for this,
To Reproduce
Define the DeepAR estimator
Epoch 0: | | 3/? [08:49<00:00, 0.01it/s, v_num=22]
Beta Was this translation helpful? Give feedback.
All reactions