Upload v1 weights
Browse files
v1/causalvae_ucf101_rec01_model.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45f105b9f6c74defcb27b2fab0242aa314c819418e1d9da00d9466f1533ca9af
|
| 3 |
+
size 986738235
|
v1/causalvae_ucf101_rec01_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e1b9ec1b3a9cae0d9bc41e7d1dfc72118f2a674e9cdb5b71c6289b069cf4b28
|
| 3 |
+
size 986646668
|
v1/config.json
ADDED
|
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "CausalVAEModel",
|
| 3 |
+
"_diffusers_version": "0.27.2",
|
| 4 |
+
"attn_resolutions": [],
|
| 5 |
+
"decoder_attention": "AttnBlock3D",
|
| 6 |
+
"decoder_conv_in": "CausalConv3d",
|
| 7 |
+
"decoder_conv_out": "CausalConv3d",
|
| 8 |
+
"decoder_mid_resnet": "ResnetBlock3D",
|
| 9 |
+
"decoder_resnet_blocks": [
|
| 10 |
+
"ResnetBlock3D",
|
| 11 |
+
"ResnetBlock3D",
|
| 12 |
+
"ResnetBlock3D",
|
| 13 |
+
"ResnetBlock3D"
|
| 14 |
+
],
|
| 15 |
+
"decoder_spatial_upsample": [
|
| 16 |
+
"",
|
| 17 |
+
"SpatialUpsample2x",
|
| 18 |
+
"SpatialUpsample2x",
|
| 19 |
+
"SpatialUpsample2x"
|
| 20 |
+
],
|
| 21 |
+
"decoder_temporal_upsample": [
|
| 22 |
+
"",
|
| 23 |
+
"",
|
| 24 |
+
"TimeUpsample2x",
|
| 25 |
+
"TimeUpsample2x"
|
| 26 |
+
],
|
| 27 |
+
"double_z": true,
|
| 28 |
+
"dropout": 0.0,
|
| 29 |
+
"embed_dim": 4,
|
| 30 |
+
"encoder_attention": "AttnBlock3D",
|
| 31 |
+
"encoder_conv_in": "CausalConv3d",
|
| 32 |
+
"encoder_conv_out": "CausalConv3d",
|
| 33 |
+
"encoder_mid_resnet": "ResnetBlock3D",
|
| 34 |
+
"encoder_resnet_blocks": [
|
| 35 |
+
"ResnetBlock3D",
|
| 36 |
+
"ResnetBlock3D",
|
| 37 |
+
"ResnetBlock3D",
|
| 38 |
+
"ResnetBlock3D"
|
| 39 |
+
],
|
| 40 |
+
"encoder_spatial_downsample": [
|
| 41 |
+
"SpatialDownsample2x",
|
| 42 |
+
"SpatialDownsample2x",
|
| 43 |
+
"SpatialDownsample2x",
|
| 44 |
+
""
|
| 45 |
+
],
|
| 46 |
+
"encoder_temporal_downsample": [
|
| 47 |
+
"TimeDownsample2x",
|
| 48 |
+
"TimeDownsample2x",
|
| 49 |
+
"",
|
| 50 |
+
""
|
| 51 |
+
],
|
| 52 |
+
"hidden_size": 128,
|
| 53 |
+
"hidden_size_mult": [
|
| 54 |
+
1,
|
| 55 |
+
2,
|
| 56 |
+
4,
|
| 57 |
+
4
|
| 58 |
+
],
|
| 59 |
+
"loss_params": {
|
| 60 |
+
"disc_start": 2001,
|
| 61 |
+
"disc_weight": 0.5,
|
| 62 |
+
"kl_weight": 1e-06,
|
| 63 |
+
"logvar_init": 0.0
|
| 64 |
+
},
|
| 65 |
+
"loss_type": "models.videovae.losses.LPIPSWithDiscriminator",
|
| 66 |
+
"lr": 1e-05,
|
| 67 |
+
"num_res_blocks": 2,
|
| 68 |
+
"q_conv": "CausalConv3d",
|
| 69 |
+
"resolution": 256,
|
| 70 |
+
"z_channels": 4
|
| 71 |
+
}
|