|
{ |
|
"_class_name": "UNetVideoModel", |
|
"_diffusers_version": "0.9.0.dev0", |
|
"_name_or_path": "hf-models/stable-diffusion-x4-upscaler/unet", |
|
"act_fn": "silu", |
|
"attention_head_dim": 8, |
|
"block_out_channels": [ |
|
256, |
|
512, |
|
512, |
|
1024 |
|
], |
|
"center_input_sample": false, |
|
"cross_attention_dim": 1024, |
|
"down_block_types": [ |
|
"DownBlock3D", |
|
"CrossAttnDownBlock3D", |
|
"CrossAttnDownBlock3D", |
|
"CrossAttnDownBlock3D" |
|
], |
|
"downsample_padding": 1, |
|
"dual_cross_attention": false, |
|
"flip_sin_to_cos": true, |
|
"freq_shift": 0, |
|
"in_channels": 7, |
|
"layers_per_block": 2, |
|
"mid_block_scale_factor": 1, |
|
"norm_eps": 1e-05, |
|
"norm_num_groups": 32, |
|
"num_class_embeds": 1000, |
|
"only_cross_attention": [ |
|
true, |
|
true, |
|
true, |
|
false |
|
], |
|
"out_channels": 4, |
|
"sample_size": 128, |
|
"up_block_types": [ |
|
"CrossAttnUpBlock3D", |
|
"CrossAttnUpBlock3D", |
|
"CrossAttnUpBlock3D", |
|
"UpBlock3D" |
|
], |
|
"use_linear_projection": true, |
|
|
|
"down_temporal_idx": [0, 1, 2, 3], |
|
"mid_temporal": true, |
|
"up_temporal_idx": [0, 1, 2, 3], |
|
"temporal_module_config": { |
|
"attention_block_types": ["",""] |
|
} |
|
} |