| { | |
| "_class_name": "AutoencoderKLLTXVideo", | |
| "_diffusers_version": "0.37.0", | |
| "block_out_channels": [ | |
| 32, | |
| 32, | |
| 32, | |
| 32, | |
| 32 | |
| ], | |
| "decoder_block_out_channels": [ | |
| 8, | |
| 16, | |
| 32 | |
| ], | |
| "decoder_causal": false, | |
| "decoder_inject_noise": [ | |
| false, | |
| false, | |
| false, | |
| false | |
| ], | |
| "decoder_layers_per_block": [ | |
| 1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "decoder_spatio_temporal_scaling": [ | |
| true, | |
| true, | |
| true | |
| ], | |
| "down_block_types": [ | |
| "LTXVideo095DownBlock3D", | |
| "LTXVideo095DownBlock3D", | |
| "LTXVideo095DownBlock3D", | |
| "LTXVideo095DownBlock3D" | |
| ], | |
| "downsample_type": [ | |
| "spatial", | |
| "temporal", | |
| "spatiotemporal", | |
| "spatiotemporal" | |
| ], | |
| "encoder_causal": true, | |
| "in_channels": 3, | |
| "latent_channels": 128, | |
| "layers_per_block": [ | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "out_channels": 3, | |
| "patch_size": 4, | |
| "patch_size_t": 1, | |
| "resnet_norm_eps": 1e-06, | |
| "scaling_factor": 1.0, | |
| "spatial_compression_ratio": 32, | |
| "spatio_temporal_scaling": [ | |
| true, | |
| true, | |
| true, | |
| true | |
| ], | |
| "temporal_compression_ratio": 8, | |
| "timestep_conditioning": true, | |
| "upsample_factor": [ | |
| 2, | |
| 2, | |
| 2 | |
| ], | |
| "upsample_residual": [ | |
| true, | |
| true, | |
| true | |
| ] | |
| } | |