|
{ |
|
"_class_name": "ControlNetSVDModel", |
|
"_diffusers_version": "0.24.0", |
|
"_name_or_path": "/input/hanlin/keypoints_control/outputs/outputs_no_depth_2nodes_frames16_tune_all_side10_withdepth_naug1.0_bs8x1/checkpoint-228000", |
|
"addition_time_embed_dim": 256, |
|
"block_out_channels": [ |
|
320, |
|
640, |
|
1280, |
|
1280 |
|
], |
|
"conditioning_channels": 3, |
|
"conditioning_embedding_out_channels": [ |
|
16, |
|
32, |
|
96, |
|
256 |
|
], |
|
"cross_attention_dim": 1024, |
|
"down_block_types": [ |
|
"CrossAttnDownBlockSpatioTemporal", |
|
"CrossAttnDownBlockSpatioTemporal", |
|
"CrossAttnDownBlockSpatioTemporal", |
|
"DownBlockSpatioTemporal" |
|
], |
|
"feature_channels": 160, |
|
"feature_out_channels": [ |
|
160, |
|
160, |
|
256, |
|
256 |
|
], |
|
"in_channels": 8, |
|
"layers_per_block": 2, |
|
"num_attention_heads": [ |
|
5, |
|
10, |
|
20, |
|
20 |
|
], |
|
"num_frames": 25, |
|
"out_channels": 4, |
|
"projection_class_embeddings_input_dim": 768, |
|
"sample_size": 96, |
|
"transformer_layers_per_block": 1, |
|
"up_block_types": [ |
|
"UpBlockSpatioTemporal", |
|
"CrossAttnUpBlockSpatioTemporal", |
|
"CrossAttnUpBlockSpatioTemporal", |
|
"CrossAttnUpBlockSpatioTemporal" |
|
], |
|
"with_id_feature": false |
|
} |
|
|