Update unet/unet_config.json
Browse files- unet/unet_config.json +7 -7
unet/unet_config.json
CHANGED
|
@@ -12,16 +12,16 @@
|
|
| 12 |
],
|
| 13 |
"attention_type": "default",
|
| 14 |
"block_out_channels": [
|
| 15 |
-
320,
|
| 16 |
640,
|
| 17 |
-
1280
|
|
|
|
| 18 |
],
|
| 19 |
"center_input_sample": false,
|
| 20 |
"class_embed_type": null,
|
| 21 |
"class_embeddings_concat": false,
|
| 22 |
"conv_in_kernel": 3,
|
| 23 |
"conv_out_kernel": 3,
|
| 24 |
-
"cross_attention_dim":
|
| 25 |
"cross_attention_norm": null,
|
| 26 |
"down_block_types": [
|
| 27 |
"DownBlock2D",
|
|
@@ -36,7 +36,7 @@
|
|
| 36 |
"flip_sin_to_cos": true,
|
| 37 |
"freq_shift": 0,
|
| 38 |
"in_channels": 4,
|
| 39 |
-
"layers_per_block":
|
| 40 |
"mid_block_only_cross_attention": null,
|
| 41 |
"mid_block_scale_factor": 1,
|
| 42 |
"mid_block_type": "UNetMidBlock2DCrossAttn",
|
|
@@ -47,7 +47,7 @@
|
|
| 47 |
"only_cross_attention": false,
|
| 48 |
"out_channels": 4,
|
| 49 |
"projection_class_embeddings_input_dim": 2816,
|
| 50 |
-
"resnet_out_scale_factor":
|
| 51 |
"resnet_skip_time_act": false,
|
| 52 |
"resnet_time_scale_shift": "default",
|
| 53 |
"sample_size": 128,
|
|
@@ -57,9 +57,9 @@
|
|
| 57 |
"time_embedding_type": "positional",
|
| 58 |
"timestep_post_act": null,
|
| 59 |
"transformer_layers_per_block": [
|
| 60 |
-
1,
|
| 61 |
2,
|
| 62 |
-
|
|
|
|
| 63 |
],
|
| 64 |
"up_block_types": [
|
| 65 |
"CrossAttnUpBlock2D",
|
|
|
|
| 12 |
],
|
| 13 |
"attention_type": "default",
|
| 14 |
"block_out_channels": [
|
|
|
|
| 15 |
640,
|
| 16 |
+
1280,
|
| 17 |
+
2560
|
| 18 |
],
|
| 19 |
"center_input_sample": false,
|
| 20 |
"class_embed_type": null,
|
| 21 |
"class_embeddings_concat": false,
|
| 22 |
"conv_in_kernel": 3,
|
| 23 |
"conv_out_kernel": 3,
|
| 24 |
+
"cross_attention_dim": 4096,
|
| 25 |
"cross_attention_norm": null,
|
| 26 |
"down_block_types": [
|
| 27 |
"DownBlock2D",
|
|
|
|
| 36 |
"flip_sin_to_cos": true,
|
| 37 |
"freq_shift": 0,
|
| 38 |
"in_channels": 4,
|
| 39 |
+
"layers_per_block": 3,
|
| 40 |
"mid_block_only_cross_attention": null,
|
| 41 |
"mid_block_scale_factor": 1,
|
| 42 |
"mid_block_type": "UNetMidBlock2DCrossAttn",
|
|
|
|
| 47 |
"only_cross_attention": false,
|
| 48 |
"out_channels": 4,
|
| 49 |
"projection_class_embeddings_input_dim": 2816,
|
| 50 |
+
"resnet_out_scale_factor": 2.0,
|
| 51 |
"resnet_skip_time_act": false,
|
| 52 |
"resnet_time_scale_shift": "default",
|
| 53 |
"sample_size": 128,
|
|
|
|
| 57 |
"time_embedding_type": "positional",
|
| 58 |
"timestep_post_act": null,
|
| 59 |
"transformer_layers_per_block": [
|
|
|
|
| 60 |
2,
|
| 61 |
+
4,
|
| 62 |
+
12
|
| 63 |
],
|
| 64 |
"up_block_types": [
|
| 65 |
"CrossAttnUpBlock2D",
|