Lorenzo Stella
Lorenzo Stella
> However, this operation is relatively slow (~1 ms on my local machine, ~100x slower than checking the invariant) so we might want to only do this when required. But...
> * always require input `freq` instead of trying to guess it > * always resample if needed I'd say that resampling can be done by default if `freq` is...
> I think either df.index.freq must be defined or it needs to be passed, but what we should not do is to guess what the frequency of the data is....
@Serendipity31 it seems to me like the warning is not related to Optuna: you can check this by just training a standalone `DeepAREstimator` on the same data. The warning appears...
Interesting! Using again feather / arrow will not be as fast?
Partially addressed in #2756, is this waiting to be enabled by default in models?
@StatMixedML have you also set the numpy seed? Numpy is used to sample instances from training data to construct batches, so it plays a role as well
@StatMixedML you can also do that via pytorch lightning: ```python import pytorch_lightning as pl pl.seed_everything(0) ``` Docstring: ``` Signature: pl.seed_everything( seed: Union[int, NoneType] = None, workers: bool = False, )...
I'll keep it open since it doesn't hurt to track it as something we should document
I think this makes sense. Doesn't really need to be backwards incompatible, right?