MMDiTBlock.Config(
        channels_in=4,
        channels_out=4,
        num_streams=2,
        attn=MultiStreamAttention.Config(
            channels_in=4,
            channels_out=4,
            num_streams=2,
            streams=[],
            num_heads=2,
            channels_head=2,
            num_heads_kv=2,
            bias=False,
            dropout=0.0,
            causal=False,
            rope=[],
            norm_qk=None,
            share_qk_norm=True,
            norm_out=None,
            attn_kernel=SdpaFused.Config(),
            init_weight=priml.model.init.kaiming_uniform,
            depth_index=()
        ),
        streams=[],
        cond_dim=0,
        ffn=SwiGLU.Config(
            channels_in=4,
            channels_out=4,
            channels_hidden=4,
            expansion=2.6666666666666665,
            round_to=1,
            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
        ),
        depth_index=()
    )
