Method sharding

python/tests/mlx_distributed_tests.py:281–287

(path, weight)

                return x

        def sharding(path, weight):
            parts = path.split(".")
            even = int(parts[1]) % 2 == 0
            if even:
                return 0
            else:
                return -1 if parts[-1] != "bias" else None

        mod = nn.Sequential(
            MyConv(3, 128, kernel_size=3),