Hugging Face's logo

dzungpham
/
font-architect

Image-to-Image
Diffusers
Safetensors
English
font-diffusion
contrastive-learning
font-generation
character-synthesis
style-transfer
dpm-solver
Model card Files
xet
Community
1
dzungpham commited on
Commit
6843c5e
·
verified ·
1 Parent(s): 5f582a3

Finish finetuning 5000 steps for phase 1 start from pretrained weights

Browse files
Files changed (10) hide show
  1. finetuned-5P1/FontDiffuser_training_phase_1_config.yaml +70 -0
  2. finetuned-5P1/checkpoint_step_1500/content_encoder.safetensors +3 -0
  3. finetuned-5P1/checkpoint_step_1500/style_encoder.safetensors +3 -0
  4. finetuned-5P1/checkpoint_step_1500/unet.safetensors +3 -0
  5. finetuned-5P1/checkpoint_step_3000/content_encoder.safetensors +3 -0
  6. finetuned-5P1/checkpoint_step_3000/style_encoder.safetensors +3 -0
  7. finetuned-5P1/checkpoint_step_3000/unet.safetensors +3 -0
  8. finetuned-5P1/final/content_encoder.safetensors +3 -0
  9. finetuned-5P1/final/style_encoder.safetensors +3 -0
  10. finetuned-5P1/final/unet.safetensors +3 -0
finetuned-5P1/FontDiffuser_training_phase_1_config.yaml ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ adam_beta1: 0.9
2
+ adam_beta2: 0.999
3
+ adam_epsilon: 1.0e-08
4
+ adam_weight_decay: 0.01
5
+ algorithm_type: dpmsolver++
6
+ beta_scheduler: scaled_linear
7
+ channel_attn: true
8
+ ckpt_interval: 1500
9
+ content_encoder_downsample_size: 3
10
+ content_image_size: !!python/tuple
11
+ - 96
12
+ - 96
13
+ content_start_channel: 64
14
+ controlnet: false
15
+ correcting_x0_fn: null
16
+ data_root: my_dataset
17
+ drop_prob: 0.1
18
+ enable_style_transform: false
19
+ experience_name: FontDiffuser_training_phase_1
20
+ gradient_accumulation_steps: 2
21
+ guidance_scale: 7.5
22
+ guidance_type: classifier-free
23
+ instructpix2pix: false
24
+ learning_rate: 0.0001
25
+ local_rank: 0
26
+ log_interval: 50
27
+ logging_dir: logs
28
+ lr_scheduler: cosine
29
+ lr_warmup_steps: 200
30
+ max_grad_norm: 1.0
31
+ max_train_steps: 3000
32
+ method: multistep
33
+ mixed_precision: 'no'
34
+ mode: refinement
35
+ model_type: noise
36
+ nce_layers: 0,1,2,3
37
+ num_inference_steps: 20
38
+ num_neg: 16
39
+ num_workers: 4
40
+ offset_coefficient: 0.5
41
+ order: 2
42
+ output_dir: outputs/FontDiffuser/finetuned-5P1
43
+ perceptual_coefficient: 0.05
44
+ phase_1: true
45
+ phase_1_ckpt_dir: ckpt/finetuned-2P1/final
46
+ phase_2: false
47
+ report_to: wandb
48
+ resolution: 96
49
+ resume_from_checkpoint: null
50
+ save_full_model: false
51
+ sc_coefficient: 0.01
52
+ scale_lr: false
53
+ scr_ckpt_path: null
54
+ scr_image_size: 96
55
+ seed: 123
56
+ skip_type: time_uniform
57
+ style_image_size: !!python/tuple
58
+ - 96
59
+ - 96
60
+ style_start_channel: 64
61
+ t_end: null
62
+ t_start: null
63
+ temperature: 0.07
64
+ train_batch_size: 8
65
+ unet_channels: !!python/tuple
66
+ - 64
67
+ - 128
68
+ - 256
69
+ - 512
70
+ val_interval: 100
finetuned-5P1/checkpoint_step_1500/content_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b276a658e636288c591172f8e17e99f8f174fd221fd885c913755255a0ccc1ff
3
+ size 4756580
finetuned-5P1/checkpoint_step_1500/style_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77ff7365d50bf65cd68fc84466f7b029c38413226bea882dfb0040e4ef2e8dfb
3
+ size 82394556
finetuned-5P1/checkpoint_step_1500/unet.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39eb7ffc980c848ce6b160c39e2588e1ac3623a1d9c303e2380bce413fba0220
3
+ size 314927748
finetuned-5P1/checkpoint_step_3000/content_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08ddb4123c0321e975573ba1227abfa703dacc446e6f2903388931da98da236c
3
+ size 4756580
finetuned-5P1/checkpoint_step_3000/style_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b266e2fd93c33b4fe9f09ad10a66227e84716fb3b419a7ef8c4b9c32ed33733a
3
+ size 82394556
finetuned-5P1/checkpoint_step_3000/unet.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dceb1675f6dcded7459710a5f1391b3a15b48bcef1cfe752ed5e8894762140cb
3
+ size 314927748
finetuned-5P1/final/content_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08ddb4123c0321e975573ba1227abfa703dacc446e6f2903388931da98da236c
3
+ size 4756580
finetuned-5P1/final/style_encoder.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b266e2fd93c33b4fe9f09ad10a66227e84716fb3b419a7ef8c4b9c32ed33733a
3
+ size 82394556
finetuned-5P1/final/unet.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dceb1675f6dcded7459710a5f1391b3a15b48bcef1cfe752ed5e8894762140cb
3
+ size 314927748