Summary: 2 instances, 1 unique

Text	Count
results = neuronperf.torch.benchmark(neuron_model_file, example_inputs, [batch_size], n_models=[4]) # FIXME: n_models should not be hard-coded	2