transformer_additional_kwargs: patch_3d: false fake_3d: false casual_3d: true casual_3d_upsampler_index: [16, 20] time_patch_size: 4 basic_block_type: "motionmodule" time_position_encoding_before_transformer: false motion_module_type: "VanillaGrid" motion_module_kwargs: num_attention_heads: 8 num_transformer_block: 1 attention_block_types: [ "Temporal_Self", "Temporal_Self" ] temporal_position_encoding: true temporal_position_encoding_max_len: 4096 temporal_attention_dim_div: 1 block_size: 2 noise_scheduler_kwargs: beta_start: 0.0001 beta_end: 0.02 beta_schedule: "linear" steps_offset: 1 vae_kwargs: enable_magvit: false