MLPMixerBlock.Config(
        channels_in=4,
        channels_out=4,
        seq_len=2,
        token_mixer=SwiGLU.Config(
            channels_in=2,
            channels_out=2,
            channels_hidden=256,
            expansion=2.6666666666666665,
            round_to=256,
            gate=True,
            bias=False,
            split_gate_projection=False,
            norm=None,
            act=priml.model.swiglu.silu,
            depth_index=(),
            init_weight=priml.model.init.unit_fan_in_uniform,
            init_weight_out=torch.nn.init.zeros_,
            up_proj=None,
            middle=None,
            down_proj=None,
            shard=None
        ),
        channel_mixer=SwiGLU.Config(
            channels_in=4,
            channels_out=4,
            channels_hidden=256,
            expansion=2.6666666666666665,
            round_to=256,
            gate=True,
            bias=False,
            split_gate_projection=False,
            norm=None,
            act=priml.model.swiglu.silu,
            depth_index=(),
            init_weight=priml.model.init.unit_fan_in_uniform,
            init_weight_out=torch.nn.init.zeros_,
            up_proj=None,
            middle=None,
            down_proj=None,
            shard=None
        ),
        norm_token=RMSNorm.Config(
            channels_in=2,
            channels_out=2,
            eps=1e-06,
            elementwise_affine=False,
            device=None,
            dtype=None
        ),
        norm_channel=RMSNorm.Config(
            channels_in=4,
            channels_out=4,
            eps=1e-06,
            elementwise_affine=False,
            device=None,
            dtype=None
        ),
        prenorm=True,
        depth_index=()
    )
