NOTE differences between Torch and TensorFlow

ResNet implem:
- stride > 2 (can be corrected) is done a bit different
- init of weights is different (for conv and dense)
- Empirically: SGD works much better for PyTorch (reaching 95.5%).
- Batch norm seems to be implemented differently for the 2 frameworks: it causes a difference in loss at training time. But we can't verify that when comparing the batch norm implem in isolation.

Review the ResNet implementations and the PyTorch/TensorFlow behavior described in this issue, focusing on stride handling, weight initialization, SGD results, and batch normalization. Determine where benchmark differences are documented; done means the relevant differences and any verification limits are recorded clearly.

Written by the indexing model from the issue text.