Trained for 9 epochs and 29000 steps.

Trained with datasets ['text-embeds', 'shutterstock', 'nijijourney', 'bg20k-1024', 'photo-aesthetics', 'text-1mp', 'cinemamix-1mp']
Learning rate 1e-06, batch size 16, and 1 gradient accumulation steps.
Used DDPM noise scheduler for training with epsilon prediction type and rescaled_betas_zero_snr=False
Using 'linspace' timestep spacing.
Base model: terminusresearch/pixart-900m-1024-ft-v0.6
VAE: madebyollin/sdxl-vae-fp16-fix

README.md

@@ -552,7 +552,7 @@ You may reuse the base model text encoder for inference.
 ## Training settings
 
 - Training epochs: 9
-- Training steps: 28000
+- Training steps: 29000
 - Learning rate: 1e-06
 - Effective batch size: 16
  - Micro-batch size: 16

optimizer.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:76d7bb467da553a835821f0275cf25229a3c48b1ad3d973e6716b78ae83227d6
+oid sha256:f0f1a87ce7027c9df4bb157da743de90cbcf81226b08cc9391b95280f8951d01
 size 5451415117

random_states_0.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c5aeb18faf79676f2ee545f39f140432a36ea35bee2122fcb285157a0dc3af6
-size 14408
+oid sha256:c71c8545cf72a871d41ab8f17385d91812b6b71f7d635371065d84b43b2dc15a
+size 14344

scheduler.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:60aea79d269a53ee5e7c794d7fe99811cd524960de92007320fabeeaa901d0e8
+oid sha256:3efc7413a4d89d5eeb49b1360238abd012dcbf3d569d6276ee9ed92cb768a65d
 size 1000

training_state-bg20k-1024.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:520f19e531e79fa90f5d0f7864d97c40800c51bb4a15ce0b433e52d0abfeaacc
-size 22245114
+oid sha256:d127af0c937cc235f459e7b8aa5b2d83bb12415da25829c2e868e659292ed975
+size 23589519

training_state.json

@@ -1 +1 @@
-{"global_step": 28000, "epoch_step": 28000, "epoch": 10, "exhausted_backends": ["celebrities", "gay", "moviecollection", "ethnic", "cinemamix-1mp", "shutterstock", "nijijourney", "photo-aesthetics"], "repeats": {"movieposters": 20, "celebrities": 0, "normalnudes": 3, "moviecollection": 0, "gay": 0, "ethnic": 0, "nsfw-1024": 0, "experimental": 0, "cinemamix-1mp": 0, "text-1mp": 3, "shutterstock": 0, "nijijourney": 0, "photo-aesthetics": 0}}
+{"global_step": 29000, "epoch_step": 29000, "epoch": 10, "exhausted_backends": ["celebrities", "gay", "moviecollection", "ethnic", "cinemamix-1mp", "shutterstock", "nijijourney", "photo-aesthetics"], "repeats": {"movieposters": 20, "celebrities": 0, "normalnudes": 3, "moviecollection": 0, "gay": 0, "ethnic": 0, "nsfw-1024": 0, "experimental": 0, "cinemamix-1mp": 0, "text-1mp": 3, "shutterstock": 0, "nijijourney": 0, "photo-aesthetics": 0}}

transformer/config.json

@@ -1,7 +1,7 @@
 {
  "_class_name": "PixArtTransformer2DModel",
  "_diffusers_version": "0.30.0.dev0",
- "_name_or_path": "/notebooks/datasets/lite-models/checkpoint-27000",
+ "_name_or_path": "/notebooks/datasets/lite-models/checkpoint-28000",
  "activation_fn": "gelu-approximate",
  "attention_bias": true,
  "attention_head_dim": 72,

transformer/diffusion_pytorch_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:27ffe91d2138c69750efbab1372b36f6616d42cee5f368ad7bc7a1defec305c5
+oid sha256:d8da14668cbc58f7c040223971538d5b1d31d4fb228080714fbed2e0adf2d238
 size 1816969728