Add converted safetensors and original pth weights

FontDiffuser_training_phase_1_config.yaml

@@ -16,12 +16,15 @@ data_root: my_dataset
 drop_prob: 0.1
 enable_style_transform: false
 experience_name: FontDiffuser_training_phase_1
-export_onnx_path: net_arch.onnx
+export_onnx_path: null
 gradient_accumulation_steps: 2
 guidance_scale: 7.5
 guidance_type: classifier-free
+hf_repo_id: dzungpham/font-diffusion-generated-data
+hf_split: handwritten_original
+hf_token: null
 learning_rate: 0.0001
-local_rank: 0
+local_rank: -1
 log_interval: 50
 logging_dir: logs
 lr_scheduler: cosine
@@ -66,4 +69,5 @@ unet_channels: !!python/tuple
 - 128
 - 256
 - 512
+use_hf_dataset: true
 val_interval: 100

FontDiffuser_training_phase_2_config.yaml

@@ -5,7 +5,7 @@ adam_weight_decay: 0.01
 algorithm_type: dpmsolver++
 beta_scheduler: scaled_linear
 channel_attn: true
-ckpt_interval: 250
+ckpt_interval: 1
 content_encoder_downsample_size: 3
 content_image_size: !!python/tuple
 - 96
@@ -14,10 +14,15 @@ content_start_channel: 64
 correcting_x0_fn: null
 data_root: my_dataset
 drop_prob: 0.1
+enable_style_transform: false
 experience_name: FontDiffuser_training_phase_2
-gradient_accumulation_steps: 2
+export_onnx_path: null
+gradient_accumulation_steps: 1
 guidance_scale: 7.5
 guidance_type: classifier-free
+hf_repo_id: dzungpham/font-diffusion-generated-data
+hf_split: handwritten_original
+hf_token: null
 learning_rate: 1.0e-05
 local_rank: -1
 log_interval: 50
@@ -25,22 +30,26 @@ logging_dir: logs
 lr_scheduler: constant
 lr_warmup_steps: 100
 max_grad_norm: 1.0
-max_train_steps: 500
+max_train_steps: 2
 method: multistep
 mixed_precision: 'no'
 mode: refinement
 model_type: noise
 nce_layers: 0,1,2,3
 num_inference_steps: 20
-num_neg: 11
+num_neg: 12
+num_workers: 4
 offset_coefficient: 0.5
 order: 2
 output_dir: outputs/FontDiffuser
 perceptual_coefficient: 0.05
-phase_1_ckpt_dir: ckpt
+phase_1: false
+phase_1_ckpt_dir: outputs/FontDiffuser/checkpoint_step_1000
 phase_2: true
 report_to: wandb
 resolution: 96
+resume_from_checkpoint: null
+save_full_model: false
 sc_coefficient: 0.05
 scale_lr: false
 scr_ckpt_path: ckpt/scr_210000.pth
@@ -54,10 +63,11 @@ style_start_channel: 64
 t_end: null
 t_start: null
 temperature: 0.07
-train_batch_size: 8
+train_batch_size: 16
 unet_channels: !!python/tuple
 - 64
 - 128
 - 256
 - 512
+use_hf_dataset: true
 val_interval: 100

checkpoint_step_1/content_encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90d470624b5cbb4300ff2a54d716adb77d06f94f0530b31da30f422d67b99798
+size 4756580

checkpoint_step_1/scr.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7618bed9fb85450d0c95b57c7e874ddf725606cf9b90da6295a0c835c3332790
+size 284227968

checkpoint_step_1/style_encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeec15da854c814b48d5a5d85e3e667905e697ee5b7ec1dc43a4794b31615cb7
+size 82394556

checkpoint_step_1/training_state.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e5c434c81202c7b62c1b19bf34d041b8ba18d40156feed3431fbcda684fd297
+size 1206964193

checkpoint_step_1/unet.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41b460184de2f7239493374e09434005a0c90842f83cbda706743f2fc367aed5
+size 314927748

checkpoint_step_1000/content_encoder.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8302106359a2f47262f83d3793a3d5a84901b051fb9ca83733e5bf05596b3bb8
+oid sha256:8c443fd0f0f8a31406ed436ad47c5ad79052e6ff387b647b43807aaa9d1f7d35
 size 4756580

checkpoint_step_1000/style_encoder.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fd9a15d7c7463cf4c631ee05946562bb5d7451f944e69f019ba3611d1ca04e11
+oid sha256:722699298a1d4922e4fa2e5e86b713c328b2f37a181dad85516b00c557293331
 size 82394556

checkpoint_step_1000/training_state.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:291281b64ae387f05fc853c0ac4cd3ea30452aaf53db0bda6059cdd1c875059f
-size 1206964193
+oid sha256:b9d7e49e1038f80535cb9920fd36ed114f2cf1d93f1a53b9d894ae4899ce2400
+size 1206964257

checkpoint_step_1000/unet.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9001dfa303a2b68ec78e9495e5b75345ca0dcbd541007e7b240643e308dd1e40
+oid sha256:29804a197cfdce5ae634e49e7abe1e7db43da6f56d312cd2db6127f7188abfde
 size 314927748

checkpoint_step_500/content_encoder.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ce108e8b264e1be106a965a86805fd47b25db7f3fdf93b4ef2ac8c05513fa28d
+oid sha256:e3ef7bb5823c6614ce7983455a3e9b41d17032377ba9c0c18454f5abdc13d4ba
 size 4756580

checkpoint_step_500/style_encoder.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e4c53dd00df78c41e1977f5f2a17d4c9e0d92cafe87a88890ae0e4f906d468c7
+oid sha256:bd8574fdcd0cd3dbbc694c707cb3f6dd96bc6563e26bb108922a22a674405275
 size 82394556

checkpoint_step_500/training_state.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b12ad4fc475384fc57187c4ef34e2c32fb36ddeb26ce44004eda12d44a6edd8b
-size 1206964193
+oid sha256:b733b27e3d901d38344835618c38903398a2851100ea955c186ffd0ed08b69bf
+size 1206964257

checkpoint_step_500/unet.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:617b2e33aa034f2853b99c870afffefb88e7057a838e234746ff6a7a379a16aa
+oid sha256:000b933fe6ad05bfe85a7828f3ed3fe315c234216c3026e61b3e1ea9e22ed0eb
 size 314927748