Out of memory

Out of memory error while calling to_batch here. I tried to load the data on my laptop and it is working. I don't know why, but I could not train on the conv pc.

• check again if submodules are up to date
• check if the pipeline works properly with minimal training effort (1 bag, 5 epochs)
  • seems to work after some modifications to have consistent variables 😆
• consider following error that occurs after training a model d1, d2, _ = tree_util.tree_leaves(data)[0].shape ^^^^^^^^^ ValueError: too many values to unpack (expected 3)
  • solved with the modifications on consistency in variables
• Modifications towards validation with the latest changes (adapt test_loader = jax.tree_util.tree_map(lambda l: l.reshape((l.shape[0] * batch_size, window_size, l.shape[-1])),
test_loader))