Upload StableDiffusionPipeline

model_index.json

@@ -1,7 +1,7 @@
 {
   "_class_name": "StableDiffusionPipeline",
   "_diffusers_version": "0.24.0",
-  "_name_or_path": "CompVis/stable-diffusion-v1-4",
+  "_name_or_path": "stabilityai/stable-diffusion-2",
   "feature_extractor": [
     "transformers",
     "CLIPImageProcessor"
@@ -10,7 +10,7 @@
     null,
     null
   ],
-  "requires_safety_checker": true,
+  "requires_safety_checker": false,
   "safety_checker": [
     null,
     null

scheduler/scheduler_config.json

@@ -8,7 +8,7 @@
   "clip_sample_range": 1.0,
   "dynamic_thresholding_ratio": 0.995,
   "num_train_timesteps": 1000,
-  "prediction_type": "epsilon",
+  "prediction_type": "v_prediction",
   "rescale_betas_zero_snr": false,
   "sample_max_value": 1.0,
   "set_alpha_to_one": false,

text_encoder/config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "CompVis/stable-diffusion-v1-4",
+  "_name_or_path": "stabilityai/stable-diffusion-2",
   "architectures": [
     "CLIPTextModel"
   ],
@@ -7,16 +7,16 @@
   "bos_token_id": 0,
   "dropout": 0.0,
   "eos_token_id": 2,
-  "hidden_act": "quick_gelu",
-  "hidden_size": 768,
+  "hidden_act": "gelu",
+  "hidden_size": 1024,
   "initializer_factor": 1.0,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 4096,
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 77,
   "model_type": "clip_text_model",
-  "num_attention_heads": 12,
-  "num_hidden_layers": 12,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 23,
   "pad_token_id": 1,
   "projection_dim": 512,
   "torch_dtype": "float32",

text_encoder/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:778d02eb9e707c3fbaae0b67b79ea0d1399b52e624fb634f2f19375ae7c047c3
-size 492265168
+oid sha256:67e013543d4fac905c882e2993d86a2d454ee69dc9e8f37c0c23d33a48959d15
+size 1361596304

tokenizer/special_tokens_map.json

@@ -13,7 +13,7 @@
     "rstrip": false,
     "single_word": false
   },
-  "pad_token": "<|endoftext|>",
+  "pad_token": "!",
   "unk_token": {
     "content": "<|endoftext|>",
     "lstrip": false,

tokenizer/tokenizer_config.json

@@ -1,6 +1,14 @@
 {
   "add_prefix_space": false,
   "added_tokens_decoder": {
+    "0": {
+      "content": "!",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
     "49406": {
       "content": "<|startoftext|>",
       "lstrip": false,
@@ -24,7 +32,7 @@
   "eos_token": "<|endoftext|>",
   "errors": "replace",
   "model_max_length": 77,
-  "pad_token": "<|endoftext|>",
+  "pad_token": "!",
   "tokenizer_class": "CLIPTokenizer",
   "unk_token": "<|endoftext|>"
 }

unet/config.json

@@ -1,12 +1,17 @@
 {
   "_class_name": "UNet2DConditionModel",
   "_diffusers_version": "0.24.0",
-  "_name_or_path": "CompVis/stable-diffusion-v1-4",
+  "_name_or_path": "stabilityai/stable-diffusion-2",
   "act_fn": "silu",
   "addition_embed_type": null,
   "addition_embed_type_num_heads": 64,
   "addition_time_embed_dim": null,
-  "attention_head_dim": 8,
+  "attention_head_dim": [
+    5,
+    10,
+    20,
+    20
+  ],
   "attention_type": "default",
   "block_out_channels": [
     320,
@@ -19,7 +24,7 @@
   "class_embeddings_concat": false,
   "conv_in_kernel": 3,
   "conv_out_kernel": 3,
-  "cross_attention_dim": 768,
+  "cross_attention_dim": 1024,
   "cross_attention_norm": null,
   "down_block_types": [
     "CrossAttnDownBlock2D",
@@ -50,7 +55,7 @@
   "resnet_skip_time_act": false,
   "resnet_time_scale_shift": "default",
   "reverse_transformer_layers_per_block": null,
-  "sample_size": 64,
+  "sample_size": 96,
   "time_cond_proj_dim": null,
   "time_embedding_act_fn": null,
   "time_embedding_dim": null,
@@ -64,5 +69,5 @@
     "CrossAttnUpBlock2D"
   ],
   "upcast_attention": false,
-  "use_linear_projection": false
+  "use_linear_projection": true
 }

unet/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:426a69fdd5fc9552c879707127964b65625fc7a61dfd0789aa6b7bbae990de83
-size 3438167536
+oid sha256:2bacfcc040106637ec334f46bc5514428c1080c4304dda8f1f0cce7937c4f553
+size 3463726504

vae/config.json

@@ -1,7 +1,7 @@
 {
   "_class_name": "AutoencoderKL",
   "_diffusers_version": "0.24.0",
-  "_name_or_path": "CompVis/stable-diffusion-v1-4",
+  "_name_or_path": "stabilityai/stable-diffusion-2",
   "act_fn": "silu",
   "block_out_channels": [
     128,
@@ -21,7 +21,7 @@
   "layers_per_block": 2,
   "norm_num_groups": 32,
   "out_channels": 3,
-  "sample_size": 512,
+  "sample_size": 768,
   "scaling_factor": 0.18215,
   "up_block_types": [
     "UpDecoderBlock2D",

vae/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b4d2b5932bb4151e54e694fd31ccf51fca908223c9485bd56cd0e1d83ad94c49
+oid sha256:2aa1f43011b553a4cba7f37456465cdbd48aab7b54b9348b890e8058ea7683ec
 size 334643268