Kashif Rasul

so i do not know what you are trying to do... but typically you set the prediction length to be as large as you have test data for... so that...

right I believe in that error message should be the size that is the input_size you need to pass... can you paste the full error?

ah damn i remember in the new version of pytorch it doesn't print the actual dims... let me check my notebook

can you kindly try: ``` estimator = TempFlowEstimator( target_dim=int(dataset.metadata.feat_static_cat[0].cardinality), prediction_length=dataset.metadata.prediction_length, cell_type='GRU', input_size=3856, num_cells=128, freq=dataset.metadata.freq, scaling=True, n_blocks=4, dropout_rate=0.3, trainer=Trainer(device=device, epochs=21, learning_rate=1e-3, num_batches_per_epoch=100, batch_size=32, num_workers=8) ) ```

the ideal solution is to calculate the data cov. feature sizes as well as the multivariate dim and the amount of lag features etc. and then calculate it... I believe...

opps i forgot to fix the readme... yes just use the `input_size=19`

good question @vfdev-5 so the `input_size` is the size of the features which are then passed to the RNN for example. The feature size depends on the freq of the...

@siqil can you kindly email me [email protected] and we can figure out the issue together in a call...

thank you! I will fix it! In the meantime can you set the input_size to 63 and try?

thanks for the detailed report.... since the model trains for a while I suppose the issue is with the data being considered. Can you kindly see if it works with...