DeepVAR Implementation

[PriyanshiGupta1701]

How do I implement DeepVAR using DeepAR and MultivariateNormalDistributionLoss, when i provide the target as a list of strings i.e multiple columns, and call the trainer it displays following error:

TypeError Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/pytorch_forecasting/metrics/base_metrics.py in forward(self, y_pred, y_actual, **kwargs)
377 try:
--> 378 res = metric(
379 y_pred[idx],

30 frames
TypeError: MultiHorizonMetric.update() got an unexpected keyword argument 'encoder_target'

During handling of the above exception, another exception occurred:

AttributeError Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/torch/distributions/utils.py in _sum_rightmost(value, dim)
76 if dim == 0:
77 return value
---> 78 required_shape = value.shape[:-dim] + (-1,)
79 return value.reshape(required_shape).sum(-1)
80

AttributeError: 'float' object has no attribute 'shape'