Kernel Dies
-
Hi @support,
I tried to train a deep learning model, which previously was trained up until an error unrelated to the memory occurred (around 20 epochs), but the kernel because of memory issues dies now.
The same model was also trained successfully locally on 8gb ram.
Do you know what could be the reason for memory runout and are there solutions to this problem besides reducing model complexity?
Regards -
@magenta-kabuto Hello.
Here is an example of how to run the strategy in single-pass mode. Rewrite your strategy for single-pass mode and submit it to the contest. It should calculate much faster.https://github.com/quantiacs/strategy-ml_lstm_multiple_features/blob/master/strategy.ipynb
-
@vyacheslav_b, thx a lot again
-
@vyacheslav_b Your suggestion worked, so thx for that. Is there a way to retrain the models like its done in the machine learning backtester, for submission purposes?
-
@magenta-kabuto
Hello. If you are using single-pass mode, then the strategy will retrein every day. This is mentioned in the template itself in the Model Retraining Frequency section, and you can see it in the strategy's source code. -
@vyacheslav_b perfect. It wasnt obvious to me that single pass was meant by that. Thank you