Model trains but I'm getting a RuntimeError during Inference

Deep Learning
1

I am using Learner on a custom model. The model training works with learner.fit_one_cycle() but when I try to predict an output from the model for inference, I get the following error.

RuntimeError: bool value of Tensor with more than one value is ambiguous

This is my model architecture

import torch as th
class model(th.nn.Module):
“”“model structure.”""

def __init__(self, n_hiddens=20, kernel_size=3):
    """Init the model structure with the number of hidden units.
    :param n_hiddens: Number of hidden units
    :type n_hiddens: int
    """
    super().__init__()
    
    self.conv = th.nn.Sequential(th.nn.Conv1d(2, n_hiddens, kernel_size),
                                 th.nn.ReLU(),
                                 th.nn.Conv1d(n_hiddens, n_hiddens,
                                              kernel_size),
                                 th.nn.ReLU())
    # self.batch_norm = th.nn.BatchNorm1d(n_hiddens, affine=False)
    self.dense = th.nn.Sequential(th.nn.Linear(n_hiddens, n_hiddens),
                                  th.nn.ReLU(),
                                  th.nn.Linear(n_hiddens, 1)
                                  )

def forward(self, x):
    """Passing data through the network.
    :param x: 2d tensor containing both (x,y) Variables
    :return: output of the net
    """

    features = self.conv(x).mean(dim=2)
    return self.dense(features)

My training data is of the form x_train.shape = (n,2, 500) and y_train.shape = (n,1)
Can anyone help me out with where I might be going wrong?

2

I did not instantiate the model before running inference.

Solved by doing :-
model = learner.model()
y_pred = model(x)

I was earlier running model(x) before instantiating it!