update model

scheduler/scheduler_config.json

@@ -4,14 +4,15 @@
   "beta_end": 0.012,
   "beta_schedule": "scaled_linear",
   "beta_start": 0.00085,
-  "clip_sample": true,
+  "clip_sample": false,
   "clip_sample_range": 1.0,
   "dynamic_thresholding_ratio": 0.995,
   "num_train_timesteps": 1000,
+  "original_inference_steps": 50,
   "prediction_type": "epsilon",
   "rescale_betas_zero_snr": false,
   "sample_max_value": 1.0,
-  "set_alpha_to_one": true,
+  "set_alpha_to_one": false,
   "steps_offset": 0,
   "thresholding": false,
   "timestep_spacing": "leading",

text_encoder/config.json

@@ -1,22 +1,20 @@
 {
-  "_name_or_path": "/tmp/test/tiny-stable-diffusion-torch/text_encoder",
   "architectures": [
     "CLIPTextModel"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 0,
-  "dropout": 0.0,
   "eos_token_id": 2,
   "hidden_act": "quick_gelu",
   "hidden_size": 32,
   "initializer_factor": 1.0,
   "initializer_range": 0.02,
-  "intermediate_size": 37,
+  "intermediate_size": 64,
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 77,
   "model_type": "clip_text_model",
-  "num_attention_heads": 4,
-  "num_hidden_layers": 5,
+  "num_attention_heads": 8,
+  "num_hidden_layers": 3,
   "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 sha256:103860291b96610a23d9dda96f6a3c4e6c6dd67e5984a0d5e7c8e62769ac6412
-size 283212
+oid sha256:5edffee1c2ae276b4ecf6e6fc290c2da8224cda3a5a57d1d86a468e8d7e70b62
+size 246416

tokenizer/tokenizer_config.json

@@ -7,15 +7,15 @@
       "normalized": true,
       "rstrip": false,
       "single_word": false,
-      "special": true
+      "special": false
     },
     "1": {
       "content": "<|endoftext|>",
       "lstrip": false,
-      "normalized": true,
+      "normalized": false,
       "rstrip": false,
       "single_word": false,
-      "special": true
+      "special": false
     }
   },
   "additional_special_tokens": [],

unet/config.json

@@ -1,7 +1,6 @@
 {
   "_class_name": "UNet2DConditionModel",
   "_diffusers_version": "0.22.0.dev0",
-  "_name_or_path": "/tmp/test/tiny-stable-diffusion-torch/unet",
   "act_fn": "silu",
   "addition_embed_type": null,
   "addition_embed_type_num_heads": 64,
@@ -9,8 +8,8 @@
   "attention_head_dim": 8,
   "attention_type": "default",
   "block_out_channels": [
-    32,
-    64
+    4,
+    8
   ],
   "center_input_sample": false,
   "class_embed_type": null,
@@ -31,12 +30,12 @@
   "flip_sin_to_cos": true,
   "freq_shift": 0,
   "in_channels": 4,
-  "layers_per_block": 2,
+  "layers_per_block": 1,
   "mid_block_only_cross_attention": null,
   "mid_block_scale_factor": 1,
   "mid_block_type": "UNetMidBlock2DCrossAttn",
   "norm_eps": 1e-05,
-  "norm_num_groups": 32,
+  "norm_num_groups": 2,
   "num_attention_heads": null,
   "num_class_embeds": null,
   "only_cross_attention": false,
@@ -45,8 +44,9 @@
   "resnet_out_scale_factor": 1.0,
   "resnet_skip_time_act": false,
   "resnet_time_scale_shift": "default",
-  "sample_size": 64,
-  "time_cond_proj_dim": null,
+  "reverse_transformer_layers_per_block": null,
+  "sample_size": 32,
+  "time_cond_proj_dim": 32,
   "time_embedding_act_fn": null,
   "time_embedding_dim": null,
   "time_embedding_type": "positional",

unet/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c69e435ef37a4e0fe3471d2c9488924bdf534cb51c08e9450420784d53336d73
-size 5742160
+oid sha256:673e65bc368c3e781f88c2464f0f34ea3da79a3ccc1640a9d71c569997cf5926
+size 96608

vae/config.json

@@ -1,11 +1,10 @@
 {
   "_class_name": "AutoencoderKL",
   "_diffusers_version": "0.22.0.dev0",
-  "_name_or_path": "/tmp/test/tiny-stable-diffusion-torch/vae",
   "act_fn": "silu",
   "block_out_channels": [
-    32,
-    64
+    4,
+    8
   ],
   "down_block_types": [
     "DownEncoderBlock2D",
@@ -15,9 +14,9 @@
   "in_channels": 3,
   "latent_channels": 4,
   "layers_per_block": 1,
-  "norm_num_groups": 32,
+  "norm_num_groups": 2,
   "out_channels": 3,
-  "sample_size": 128,
+  "sample_size": 32,
   "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 sha256:7999b516094ac4239121d9f62a37e6be99620e7016220011bb99cf8b1a391aad
-size 2646532
+oid sha256:b023ca2edc02c318f3c46b7c8c05d21c188724d40b51855065fec0f91bf950ef
+size 59772