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=<function priml.model.init.kaiming_uniform at 0xdefacedeface>,
                        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=<function priml.model.swiglu.silu at 0xdefacedeface>,
                        depth_index=(),
                        init_weight=<function priml.model.init.unit_fan_in_uniform at 0xdefacedeface>,
                        init_weight_out=<function torch.nn.init.zeros_ at 0xdefacedeface>,
                        shard=None
                ),
                depth_index=()
        )
