Segmentation Fault on M1 Mac.

Running the sample,
_python dlrm_s_pytorch.py --mini-batch-size=2 --data-size=6 --debug-mode_

**Outputs:**

>
> Using CPU...
> model arch:
> mlp top arch 3 layers, with input to output dimensions:
> [8 4 2 1]
> # of interactions
> 8
> mlp bot arch 2 layers, with input to output dimensions:
> [4 3 2]
> # of features (sparse and dense)
> 4
> dense feature size
> 4
> sparse feature size
> 2
> # of embeddings (= # of sparse features) 3, with dimensions 2x:
> [4 3 2]
> data (inputs and targets):
> mini-batch: 0
> tensor([[0.6965, 0.2861, 0.2269, 0.5513],
> [0.7195, 0.4231, 0.9808, 0.6848]])
> tensor([[1, 2],
> [1, 1],
> [1, 1]], dtype=torch.int32)
> [tensor([1, 0, 1]), tensor([0, 1]), tensor([1, 0])]
> tensor([[0.3618],
> [0.2283]])
> mini-batch: 1
> tensor([[0.2937, 0.6310, 0.0921, 0.4337],
> [0.4309, 0.4937, 0.4258, 0.3123]])
> tensor([[1, 2],
> [1, 2],
> [1, 1]], dtype=torch.int32)
> [tensor([3, 0, 2]), tensor([1, 1, 2]), tensor([1, 1])]
> tensor([[0.6031],
> [0.5451]])
> mini-batch: 2
> tensor([[0.3428, 0.3041, 0.4170, 0.6813],
> [0.8755, 0.5104, 0.6693, 0.5859]])
> tensor([[2, 1],
> [1, 2],
> [1, 1]], dtype=torch.int32)
> [tensor([2, 3, 2]), tensor([0, 0, 2]), tensor([1, 1])]
> tensor([[0.5568],
> [0.1590]])
> initial parameters (weights and bias):
> [[-0.34693 0.19553]
> [-0.18123 0.19197]
> [ 0.05438 -0.11105]
> [ 0.42513 0.34167]]
> [[-0.16466 -0.52702]
> [-0.22543 -0.11757]
> [ 0.23667 0.57199]]
> [[-0.20377 0.3713 ]
> [ 0.13177 0.27111]]
> [[-0.16825 -0.58044 -0.39152 -0.64812]
> [ 1.11561 0.0879 0.61481 -0.67743]
> [ 0.09677 0.62959 -0.17907 0.55115]]
> [-0.62618 -0.7872 0.21905]
> [[-0.23981 0.40607 -1.25093]
> [ 0.45048 1.64331 -0.01557]]
> [0.02414 0.12696]
> [[-0.76015 0.17397 -0.65541 -0.1746 0.5074 -0.30015 0.20463 0.41345]
> [ 0.1138 -0.55969 -0.13573 0.79993 -0.82672 -0.11259 -0.2254 0.04929]
> [ 0.30546 0.65675 -0.11032 0.33164 0.20402 0.19365 -0.23022 -0.40715]
> [-0.44909 -0.30881 0.13133 0.31066 0.13206 -0.22411 0.73728 0.62007]]
> [-0.177 -0.41172 0.06511 0.63365]
> [[ 0.19212 0.32132 -0.12244 0.26343]
> [ 0.89174 -0.13837 0.08274 0.14654]]
> [ 0.20062 -0.99836]
> [[-1.53246 -0.83254]]
> [0.16794]
> time/loss/accuracy (if enabled):
> zsh: segmentation fault python dlrm_s_pytorch.py --mini-batch-size=2 --data-size=6 --debug-mode

Start by running the reported command, `python dlrm_s_pytorch.py --mini-batch-size=2 --data-size=6 --debug-mode`, on an M1 Mac and inspect `dlrm_s_pytorch.py` around the point where execution reaches `time/loss/accuracy`. Trace the segmentation fault and verify that the sample completes without crashing on the reported platform.

Written by the indexing model from the issue text.