Commit
•
81e3081
1
Parent(s):
7691734
Create causal_video_vae/config.json
Browse files- causal_video_vae/config.json +93 -0
causal_video_vae/config.json
ADDED
|
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "CausalVideoVAE",
|
| 3 |
+
"_diffusers_version": "0.30.3",
|
| 4 |
+
"_name_or_path": "../pyramid-flow-sd3/causal_video_vae",
|
| 5 |
+
"add_post_quant_conv": true,
|
| 6 |
+
"decoder_act_fn": "silu",
|
| 7 |
+
"decoder_block_dropout": [
|
| 8 |
+
0.0,
|
| 9 |
+
0.0,
|
| 10 |
+
0.0,
|
| 11 |
+
0.0
|
| 12 |
+
],
|
| 13 |
+
"decoder_block_out_channels": [
|
| 14 |
+
128,
|
| 15 |
+
256,
|
| 16 |
+
512,
|
| 17 |
+
512
|
| 18 |
+
],
|
| 19 |
+
"decoder_in_channels": 16,
|
| 20 |
+
"decoder_layers_per_block": [
|
| 21 |
+
3,
|
| 22 |
+
3,
|
| 23 |
+
3,
|
| 24 |
+
3
|
| 25 |
+
],
|
| 26 |
+
"decoder_norm_num_groups": 32,
|
| 27 |
+
"decoder_out_channels": 3,
|
| 28 |
+
"decoder_spatial_up_sample": [
|
| 29 |
+
true,
|
| 30 |
+
true,
|
| 31 |
+
true,
|
| 32 |
+
false
|
| 33 |
+
],
|
| 34 |
+
"decoder_temporal_up_sample": [
|
| 35 |
+
true,
|
| 36 |
+
true,
|
| 37 |
+
true,
|
| 38 |
+
false
|
| 39 |
+
],
|
| 40 |
+
"decoder_type": "causal_vae_conv",
|
| 41 |
+
"decoder_up_block_types": [
|
| 42 |
+
"UpDecoderBlockCausal3D",
|
| 43 |
+
"UpDecoderBlockCausal3D",
|
| 44 |
+
"UpDecoderBlockCausal3D",
|
| 45 |
+
"UpDecoderBlockCausal3D"
|
| 46 |
+
],
|
| 47 |
+
"downsample_scale": 8,
|
| 48 |
+
"encoder_act_fn": "silu",
|
| 49 |
+
"encoder_block_dropout": [
|
| 50 |
+
0.0,
|
| 51 |
+
0.0,
|
| 52 |
+
0.0,
|
| 53 |
+
0.0
|
| 54 |
+
],
|
| 55 |
+
"encoder_block_out_channels": [
|
| 56 |
+
128,
|
| 57 |
+
256,
|
| 58 |
+
512,
|
| 59 |
+
512
|
| 60 |
+
],
|
| 61 |
+
"encoder_double_z": true,
|
| 62 |
+
"encoder_down_block_types": [
|
| 63 |
+
"DownEncoderBlockCausal3D",
|
| 64 |
+
"DownEncoderBlockCausal3D",
|
| 65 |
+
"DownEncoderBlockCausal3D",
|
| 66 |
+
"DownEncoderBlockCausal3D"
|
| 67 |
+
],
|
| 68 |
+
"encoder_in_channels": 3,
|
| 69 |
+
"encoder_layers_per_block": [
|
| 70 |
+
2,
|
| 71 |
+
2,
|
| 72 |
+
2,
|
| 73 |
+
2
|
| 74 |
+
],
|
| 75 |
+
"encoder_norm_num_groups": 32,
|
| 76 |
+
"encoder_out_channels": 16,
|
| 77 |
+
"encoder_spatial_down_sample": [
|
| 78 |
+
true,
|
| 79 |
+
true,
|
| 80 |
+
true,
|
| 81 |
+
false
|
| 82 |
+
],
|
| 83 |
+
"encoder_temporal_down_sample": [
|
| 84 |
+
true,
|
| 85 |
+
true,
|
| 86 |
+
true,
|
| 87 |
+
false
|
| 88 |
+
],
|
| 89 |
+
"encoder_type": "causal_vae_conv",
|
| 90 |
+
"interpolate": false,
|
| 91 |
+
"sample_size": 256,
|
| 92 |
+
"scaling_factor": 0.13025
|
| 93 |
+
}
|