update config

model_index.json

@@ -15,7 +15,7 @@
   ],
   "unet": [
     "diffusers",
-    "UNetConditionalModel"
+    "UNet2DConditionModel"
   ],
   "vqvae": [
     "diffusers",

unet/config.json

@@ -1,44 +1,49 @@
 {
-  "_class_name": "UNetConditionalModel",
+  "_class_name": "UNet2DConditionModel",
   "_diffusers_version": "0.0.4",
-  "attention_resolutions": [
-    4,
-    2,
-    1
-  ],
-  "attn_resolutions": null,
+  "act_fn": "silu",
+  "attention_head_dim": 8,
   "block_channels": [
     320,
     640,
     1280,
     1280
   ],
+  "block_out_channels": [
+    320,
+    640,
+    1280,
+    1280
+  ],
   "center_input_sample": false,
   "conv_resample": true,
   "down_blocks": [
-    "UNetResCrossAttnDownBlock2D",
-    "UNetResCrossAttnDownBlock2D",
-    "UNetResCrossAttnDownBlock2D",
-    "UNetResDownBlock2D"
+    "CrossAttnDownBlock2D",
+    "CrossAttnDownBlock2D",
+    "CrossAttnDownBlock2D",
+    "DownBlock2D"
   ],
   "downsample_padding": 1,
   "downscale_freq_shift": 0,
-  "dropout": 0,
   "flip_sin_to_cos": true,
-  "image_size": 32,
+  "freq_shift": 0,
+  "image_size": null,
   "in_channels": 4,
+  "layers_per_block": 2,
   "mid_block_scale_factor": 1,
-  "name_or_path": "fusing/latent-diffusion-text2im-large",
+  "norm_eps": 1e-05,
+  "norm_num_groups": 32,
   "num_head_channels": 8,
-	"num_attention_heads": 8,
   "num_res_blocks": 2,
   "out_channels": 4,
+  "resnet_act_fn": "silu",
   "resnet_eps": 1e-05,
-  "resnet_num_groups": 32,
+  "resnet_num_groups": 30,
+  "sample_size": 32,
   "up_blocks": [
-    "UNetResUpBlock2D",
-    "UNetResCrossAttnUpBlock2D",
-    "UNetResCrossAttnUpBlock2D",
-    "UNetResCrossAttnUpBlock2D"
+    "UpBlock2D",
+    "CrossAttnUpBlock2D",
+    "CrossAttnUpBlock2D",
+    "CrossAttnUpBlock2D"
   ]
 }