Steps 10k

logs/text2image-fine-tune/1713653217.150866/events.out.tfevents.1713653217.r-nroggendorff-jupyter-hz4x4swz-870f3-uz007.653.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48a4539245f8e51c7b65ea37d1b5c11dc5e2138e3dd3e3e59c2d005411766e34
+size 2257

logs/text2image-fine-tune/1713653217.1672676/hparams.yml

@@ -0,0 +1,51 @@
+adam_beta1: 0.9
+adam_beta2: 0.999
+adam_epsilon: 1.0e-08
+adam_weight_decay: 0.01
+allow_tf32: false
+cache_dir: null
+caption_column: text
+center_crop: false
+checkpointing_steps: 100000
+checkpoints_total_limit: null
+dataloader_num_workers: 0
+dataset_config_name: null
+dataset_name: nroggendorff/zelda
+enable_xformers_memory_efficient_attention: false
+gradient_accumulation_steps: 1
+gradient_checkpointing: false
+hub_model_id: null
+hub_token: null
+image_column: image
+input_perturbation: 0
+learning_rate: 0.0001
+local_rank: -1
+logging_dir: logs
+lr_scheduler: constant
+lr_warmup_steps: 500
+max_grad_norm: 1.0
+max_train_samples: null
+max_train_steps: 1000
+mixed_precision: fp16
+noise_offset: 0
+non_ema_revision: null
+num_train_epochs: 10
+output_dir: sd-zelda-model
+prediction_type: null
+pretrained_model_name_or_path: stabilityai/stable-diffusion-xl-base-1.0
+push_to_hub: false
+random_flip: false
+report_to: tensorboard
+resolution: 384
+resume_from_checkpoint: null
+revision: null
+scale_lr: false
+seed: null
+snr_gamma: null
+tracker_project_name: text2image-fine-tune
+train_batch_size: 1
+train_data_dir: null
+use_8bit_adam: false
+use_ema: false
+validation_epochs: 5
+variant: null

logs/text2image-fine-tune/1713653355.345343/events.out.tfevents.1713653355.r-nroggendorff-jupyter-hz4x4swz-870f3-uz007.726.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cf0a33c2372715b6ad344e864c74bd3be0106e7b5ac48ff5a983f196d19dba0
+size 2249

logs/text2image-fine-tune/1713653355.3618417/hparams.yml

@@ -0,0 +1,51 @@
+adam_beta1: 0.9
+adam_beta2: 0.999
+adam_epsilon: 1.0e-08
+adam_weight_decay: 0.01
+allow_tf32: false
+cache_dir: null
+caption_column: text
+center_crop: false
+checkpointing_steps: 100000
+checkpoints_total_limit: null
+dataloader_num_workers: 0
+dataset_config_name: null
+dataset_name: nroggendorff/zelda
+enable_xformers_memory_efficient_attention: false
+gradient_accumulation_steps: 1
+gradient_checkpointing: false
+hub_model_id: null
+hub_token: null
+image_column: image
+input_perturbation: 0
+learning_rate: 0.0001
+local_rank: -1
+logging_dir: logs
+lr_scheduler: constant
+lr_warmup_steps: 500
+max_grad_norm: 1.0
+max_train_samples: null
+max_train_steps: 1000
+mixed_precision: fp16
+noise_offset: 0
+non_ema_revision: null
+num_train_epochs: 10
+output_dir: sd-zelda-model
+prediction_type: null
+pretrained_model_name_or_path: stabilityai/stable-diffusion-2-1
+push_to_hub: false
+random_flip: false
+report_to: tensorboard
+resolution: 384
+resume_from_checkpoint: null
+revision: null
+scale_lr: false
+seed: null
+snr_gamma: null
+tracker_project_name: text2image-fine-tune
+train_batch_size: 1
+train_data_dir: null
+use_8bit_adam: false
+use_ema: false
+validation_epochs: 5
+variant: null

logs/text2image-fine-tune/events.out.tfevents.1713653217.r-nroggendorff-jupyter-hz4x4swz-870f3-uz007.653.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d0b84ded0cd65f8c9df0b944ad1b7f690940391c42681b1d4f811b7949d0754
+size 88

logs/text2image-fine-tune/events.out.tfevents.1713653355.r-nroggendorff-jupyter-hz4x4swz-870f3-uz007.726.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea357e20abfac01ae3abccc188d039c1d24ba007811839491897159930201a82
+size 48961

model_index.json

@@ -1,7 +1,7 @@
 {
   "_class_name": "StableDiffusionPipeline",
   "_diffusers_version": "0.28.0.dev0",
-  "_name_or_path": "runwayml/stable-diffusion-v1-5",
+  "_name_or_path": "stabilityai/stable-diffusion-2-1",
   "feature_extractor": [
     "transformers",
     "CLIPImageProcessor"
@@ -10,14 +10,14 @@
     null,
     null
   ],
-  "requires_safety_checker": true,
+  "requires_safety_checker": false,
   "safety_checker": [
-    "stable_diffusion",
-    "StableDiffusionSafetyChecker"
+    null,
+    null
   ],
   "scheduler": [
     "diffusers",
-    "PNDMScheduler"
+    "DDIMScheduler"
   ],
   "text_encoder": [
     "transformers",

scheduler/scheduler_config.json

@@ -1,15 +1,20 @@
 {
-  "_class_name": "PNDMScheduler",
+  "_class_name": "DDIMScheduler",
   "_diffusers_version": "0.28.0.dev0",
   "beta_end": 0.012,
   "beta_schedule": "scaled_linear",
   "beta_start": 0.00085,
   "clip_sample": false,
+  "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,
   "skip_prk_steps": true,
   "steps_offset": 1,
+  "thresholding": false,
   "timestep_spacing": "leading",
   "trained_betas": null
 }

text_encoder/config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "runwayml/stable-diffusion-v1-5",
+  "_name_or_path": "stabilityai/stable-diffusion-2-1",
   "architectures": [
     "CLIPTextModel"
   ],
@@ -7,18 +7,18 @@
   "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": 768,
+  "projection_dim": 512,
   "torch_dtype": "float16",
   "transformers_version": "4.40.0",
   "vocab_size": 49408

text_encoder/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:660c6f5b1abae9dc498ac2d21e1347d2abdb0cf6c0c0c8576cd796491d9a6cdd
-size 246144152
+oid sha256:bc1827c465450322616f06dea41596eac7d493f4e95904dcb51f0fc745c4e13f
+size 680820392

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": "0.28.0.dev0",
-  "_name_or_path": "runwayml/stable-diffusion-v1-5",
+  "_name_or_path": "stabilityai/stable-diffusion-2-1",
   "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,
@@ -63,6 +68,6 @@
     "CrossAttnUpBlock2D",
     "CrossAttnUpBlock2D"
   ],
-  "upcast_attention": false,
-  "use_linear_projection": false
+  "upcast_attention": true,
+  "use_linear_projection": true
 }

unet/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad86ff4163d5e684d0fbb27917cab7a6bdde4629bf66b6fa35da0d917bc663ce
-size 3438167536
+oid sha256:2977a69e5c59eb9155ae7976dedd25cb05e382e0fea3f4fe75fb7b50e699205b
+size 3463726504

vae/config.json

@@ -1,7 +1,7 @@
 {
   "_class_name": "AutoencoderKL",
   "_diffusers_version": "0.28.0.dev0",
-  "_name_or_path": "runwayml/stable-diffusion-v1-5",
+  "_name_or_path": "stabilityai/stable-diffusion-2-1",
   "act_fn": "silu",
   "block_out_channels": [
     128,
@@ -23,7 +23,7 @@
   "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 sha256:4fbcf0ebe55a0984f5a5e00d8c4521d52359af7229bb4d81890039d2aa16dd7c
+oid sha256:3e4c08995484ee61270175e9e7a072b66a6e4eeb5f0c266667fe1f45b90daf9a
 size 167335342