Update unet/config.json

unet/config.json

@@ -1,16 +1,19 @@
 {
   "_class_name": "UNet2DConditionModel",
-  "_diffusers_version": "0.19.0",
+  "_diffusers_version": "0.22.0.dev0",
   "act_fn": "silu",
-  "addition_embed_type": null,
+  "addition_embed_type": "text_time",
   "addition_embed_type_num_heads": 64,
-  "addition_time_embed_dim": null,
-  "attention_head_dim": 8,
+  "addition_time_embed_dim": 256,
+  "attention_head_dim": [
+    5,
+    10,
+    20
+  ],
   "attention_type": "default",
   "block_out_channels": [
     320,
     640,
-    1280,
     1280
   ],
   "center_input_sample": false,
@@ -18,13 +21,12 @@
   "class_embeddings_concat": false,
   "conv_in_kernel": 3,
   "conv_out_kernel": 3,
-  "cross_attention_dim": 768,
+  "cross_attention_dim": 2048,
   "cross_attention_norm": null,
   "down_block_types": [
+    "DownBlock2D",
     "CrossAttnDownBlock2D",
-    "CrossAttnDownBlock2D",
-    "CrossAttnDownBlock2D",
-    "DownBlock2D"
+    "CrossAttnDownBlock2D"
   ],
   "downsample_padding": 1,
   "dropout": 0.0,
@@ -44,24 +46,26 @@
   "num_class_embeds": null,
   "only_cross_attention": false,
   "out_channels": 4,
-  "projection_class_embeddings_input_dim": null,
+  "projection_class_embeddings_input_dim": 2816,
   "resnet_out_scale_factor": 1.0,
   "resnet_skip_time_act": false,
   "resnet_time_scale_shift": "default",
-  "reverse_transformer_layers_per_block": null,
-  "sample_size": 64,
+  "sample_size": 128,
   "time_cond_proj_dim": null,
   "time_embedding_act_fn": null,
   "time_embedding_dim": null,
   "time_embedding_type": "positional",
   "timestep_post_act": null,
-  "transformer_layers_per_block": 1,
+  "transformer_layers_per_block": [
+    1,
+    2,
+    10
+  ],
   "up_block_types": [
-    "UpBlock2D",
     "CrossAttnUpBlock2D",
     "CrossAttnUpBlock2D",
-    "CrossAttnUpBlock2D"
+    "UpBlock2D"
   ],
-  "upcast_attention": false,
-  "use_linear_projection": false
+  "upcast_attention": null,
+  "use_linear_projection": true
 }