We would like to add a summary table breaking down the benchmarks into the number we have from each model and categories from the metadata. For example, here is what we had in the Q3 slides: The idea ...
batch_sizes: List[int] = field(default_factory=lambda: [1, 4, 8, 16, 32]) input_sizes: List[Tuple[int, int]] = field(default_factory=lambda: [(64, 64), (128, 128 ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results