ardha27 commited on
Commit
107d24a
1 Parent(s): 9fed519

Add files using upload-large-folder tool

Browse files
clement_niece.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28daab30d3e3a01547de1d001bcaaac090d248f805a997e74bf2cae5600d8320
3
+ size 23802909568
clement_niece_20241023-104334.json ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "adaptive_noise_scale": 0,
3
+ "additional_parameters": "",
4
+ "ae": "/home/Ubuntu/Downloads/ae.safetensors",
5
+ "apply_t5_attn_mask": false,
6
+ "async_upload": false,
7
+ "blocks_to_swap": 0,
8
+ "blockwise_fused_optimizers": false,
9
+ "bucket_no_upscale": true,
10
+ "bucket_reso_steps": 64,
11
+ "cache_latents": true,
12
+ "cache_latents_to_disk": true,
13
+ "caption_dropout_every_n_epochs": 0,
14
+ "caption_dropout_rate": 0,
15
+ "caption_extension": ".txt",
16
+ "clip_g": "",
17
+ "clip_l": "",
18
+ "clip_skip": 0,
19
+ "color_aug": false,
20
+ "cpu_offload_checkpointing": false,
21
+ "dataset_config": "",
22
+ "debiased_estimation_loss": false,
23
+ "disable_mmap_load_safetensors": false,
24
+ "discrete_flow_shift": 3.1582,
25
+ "double_blocks_to_swap": 0,
26
+ "dynamo_backend": "no",
27
+ "dynamo_mode": "default",
28
+ "dynamo_use_dynamic": false,
29
+ "dynamo_use_fullgraph": false,
30
+ "enable_bucket": false,
31
+ "epoch": 100,
32
+ "extra_accelerate_launch_args": "",
33
+ "flip_aug": false,
34
+ "flux1_cache_text_encoder_outputs": true,
35
+ "flux1_cache_text_encoder_outputs_to_disk": true,
36
+ "flux1_checkbox": true,
37
+ "flux1_clip_l": "/home/Ubuntu/Downloads/clip_l.safetensors",
38
+ "flux1_t5xxl": "/home/Ubuntu/Downloads/t5xxl_fp16.safetensors",
39
+ "flux_fused_backward_pass": true,
40
+ "fp8_base": false,
41
+ "full_bf16": true,
42
+ "full_fp16": false,
43
+ "fused_backward_pass": false,
44
+ "fused_optimizer_groups": 0,
45
+ "gpu_ids": "0",
46
+ "gradient_accumulation_steps": 1,
47
+ "gradient_checkpointing": true,
48
+ "guidance_scale": 1,
49
+ "huber_c": 0.1,
50
+ "huber_schedule": "snr",
51
+ "huggingface_path_in_repo": "",
52
+ "huggingface_repo_id": "",
53
+ "huggingface_repo_type": "",
54
+ "huggingface_repo_visibility": "",
55
+ "huggingface_token": "",
56
+ "ip_noise_gamma": 0,
57
+ "ip_noise_gamma_random_strength": false,
58
+ "keep_tokens": 0,
59
+ "learning_rate": 1e-05,
60
+ "learning_rate_te": 0,
61
+ "learning_rate_te1": 1e-05,
62
+ "learning_rate_te2": 1e-05,
63
+ "log_config": false,
64
+ "log_tracker_config": "",
65
+ "log_tracker_name": "",
66
+ "log_with": "",
67
+ "logging_dir": "/workspace/final_training_models/log",
68
+ "logit_mean": 0,
69
+ "logit_std": 1,
70
+ "loss_type": "l2",
71
+ "lr_scheduler": "constant",
72
+ "lr_scheduler_args": "",
73
+ "lr_scheduler_num_cycles": 1,
74
+ "lr_scheduler_power": 1,
75
+ "lr_scheduler_type": "",
76
+ "lr_warmup": 0,
77
+ "lr_warmup_steps": 0,
78
+ "main_process_port": 0,
79
+ "masked_loss": false,
80
+ "max_bucket_reso": 2048,
81
+ "max_data_loader_n_workers": 0,
82
+ "max_resolution": "1024,1024",
83
+ "max_timestep": 1000,
84
+ "max_token_length": 75,
85
+ "max_train_epochs": 0,
86
+ "max_train_steps": 0,
87
+ "mem_eff_attn": false,
88
+ "mem_eff_save": true,
89
+ "metadata_author": "",
90
+ "metadata_description": "",
91
+ "metadata_license": "",
92
+ "metadata_tags": "",
93
+ "metadata_title": "",
94
+ "min_bucket_reso": 256,
95
+ "min_snr_gamma": 0,
96
+ "min_timestep": 0,
97
+ "mixed_precision": "bf16",
98
+ "mode_scale": 1.29,
99
+ "model_list": "custom",
100
+ "model_prediction_type": "raw",
101
+ "multi_gpu": false,
102
+ "multires_noise_discount": 0.3,
103
+ "multires_noise_iterations": 0,
104
+ "no_token_padding": false,
105
+ "noise_offset": 0,
106
+ "noise_offset_random_strength": false,
107
+ "noise_offset_type": "Original",
108
+ "num_cpu_threads_per_process": 2,
109
+ "num_machines": 1,
110
+ "num_processes": 1,
111
+ "optimizer": "Adafactor",
112
+ "optimizer_args": "scale_parameter=False relative_step=False warmup_init=False weight_decay=0.01",
113
+ "output_dir": "/workspace/final_training_models/model",
114
+ "output_name": "clement_niece",
115
+ "persistent_data_loader_workers": false,
116
+ "pretrained_model_name_or_path": "/workspace/flux1-dev.safetensors",
117
+ "prior_loss_weight": 1,
118
+ "random_crop": false,
119
+ "reg_data_dir": "",
120
+ "resume": "",
121
+ "resume_from_huggingface": "",
122
+ "sample_every_n_epochs": 0,
123
+ "sample_every_n_steps": 0,
124
+ "sample_prompts": "",
125
+ "sample_sampler": "euler_a",
126
+ "save_clip": false,
127
+ "save_every_n_epochs": 100,
128
+ "save_every_n_steps": 0,
129
+ "save_last_n_epochs_state": 0,
130
+ "save_last_n_steps": 0,
131
+ "save_last_n_steps_state": 0,
132
+ "save_model_as": "safetensors",
133
+ "save_precision": "fp16",
134
+ "save_state": false,
135
+ "save_state_on_train_end": false,
136
+ "save_state_to_huggingface": false,
137
+ "save_t5xxl": false,
138
+ "scale_v_pred_loss_like_noise_pred": false,
139
+ "sd3_cache_text_encoder_outputs": false,
140
+ "sd3_cache_text_encoder_outputs_to_disk": false,
141
+ "sd3_checkbox": false,
142
+ "sd3_text_encoder_batch_size": 1,
143
+ "sdxl": false,
144
+ "sdxl_cache_text_encoder_outputs": false,
145
+ "sdxl_no_half_vae": false,
146
+ "seed": 1,
147
+ "shuffle_caption": false,
148
+ "single_blocks_to_swap": 0,
149
+ "skip_cache_check": false,
150
+ "split_mode": false,
151
+ "stop_text_encoder_training": 0,
152
+ "t5xxl": "",
153
+ "t5xxl_device": "",
154
+ "t5xxl_dtype": "bf16",
155
+ "t5xxl_max_token_length": 512,
156
+ "timestep_sampling": "sigmoid",
157
+ "train_batch_size": 7,
158
+ "train_blocks": "all",
159
+ "train_data_dir": "/workspace/final_training_models/img",
160
+ "v2": false,
161
+ "v_parameterization": false,
162
+ "v_pred_like_loss": 0,
163
+ "vae": "",
164
+ "vae_batch_size": 4,
165
+ "wandb_api_key": "",
166
+ "wandb_run_name": "",
167
+ "weighted_captions": false,
168
+ "weighting_scheme": "logit_normal",
169
+ "xformers": "sdpa"
170
+ }
clement_niece_20241023-104644.json ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "adaptive_noise_scale": 0,
3
+ "additional_parameters": "",
4
+ "ae": "/workspace/ae.safetensors",
5
+ "apply_t5_attn_mask": false,
6
+ "async_upload": false,
7
+ "blocks_to_swap": 0,
8
+ "blockwise_fused_optimizers": false,
9
+ "bucket_no_upscale": true,
10
+ "bucket_reso_steps": 64,
11
+ "cache_latents": true,
12
+ "cache_latents_to_disk": true,
13
+ "caption_dropout_every_n_epochs": 0,
14
+ "caption_dropout_rate": 0,
15
+ "caption_extension": ".txt",
16
+ "clip_g": "",
17
+ "clip_l": "",
18
+ "clip_skip": 0,
19
+ "color_aug": false,
20
+ "cpu_offload_checkpointing": false,
21
+ "dataset_config": "",
22
+ "debiased_estimation_loss": false,
23
+ "disable_mmap_load_safetensors": false,
24
+ "discrete_flow_shift": 3.1582,
25
+ "double_blocks_to_swap": 0,
26
+ "dynamo_backend": "no",
27
+ "dynamo_mode": "default",
28
+ "dynamo_use_dynamic": false,
29
+ "dynamo_use_fullgraph": false,
30
+ "enable_bucket": false,
31
+ "epoch": 100,
32
+ "extra_accelerate_launch_args": "",
33
+ "flip_aug": false,
34
+ "flux1_cache_text_encoder_outputs": true,
35
+ "flux1_cache_text_encoder_outputs_to_disk": true,
36
+ "flux1_checkbox": true,
37
+ "flux1_clip_l": "/workspace/clip_l.safetensors",
38
+ "flux1_t5xxl": "/workspace/t5xxl_fp16.safetensors",
39
+ "flux_fused_backward_pass": true,
40
+ "fp8_base": false,
41
+ "full_bf16": true,
42
+ "full_fp16": false,
43
+ "fused_backward_pass": false,
44
+ "fused_optimizer_groups": 0,
45
+ "gpu_ids": "0",
46
+ "gradient_accumulation_steps": 1,
47
+ "gradient_checkpointing": true,
48
+ "guidance_scale": 1,
49
+ "huber_c": 0.1,
50
+ "huber_schedule": "snr",
51
+ "huggingface_path_in_repo": "",
52
+ "huggingface_repo_id": "",
53
+ "huggingface_repo_type": "",
54
+ "huggingface_repo_visibility": "",
55
+ "huggingface_token": "",
56
+ "ip_noise_gamma": 0,
57
+ "ip_noise_gamma_random_strength": false,
58
+ "keep_tokens": 0,
59
+ "learning_rate": 1e-05,
60
+ "learning_rate_te": 0,
61
+ "learning_rate_te1": 1e-05,
62
+ "learning_rate_te2": 1e-05,
63
+ "log_config": false,
64
+ "log_tracker_config": "",
65
+ "log_tracker_name": "",
66
+ "log_with": "",
67
+ "logging_dir": "/workspace/final_training_models/log",
68
+ "logit_mean": 0,
69
+ "logit_std": 1,
70
+ "loss_type": "l2",
71
+ "lr_scheduler": "constant",
72
+ "lr_scheduler_args": "",
73
+ "lr_scheduler_num_cycles": 1,
74
+ "lr_scheduler_power": 1,
75
+ "lr_scheduler_type": "",
76
+ "lr_warmup": 0,
77
+ "lr_warmup_steps": 0,
78
+ "main_process_port": 0,
79
+ "masked_loss": false,
80
+ "max_bucket_reso": 2048,
81
+ "max_data_loader_n_workers": 0,
82
+ "max_resolution": "1024,1024",
83
+ "max_timestep": 1000,
84
+ "max_token_length": 75,
85
+ "max_train_epochs": 0,
86
+ "max_train_steps": 0,
87
+ "mem_eff_attn": false,
88
+ "mem_eff_save": true,
89
+ "metadata_author": "",
90
+ "metadata_description": "",
91
+ "metadata_license": "",
92
+ "metadata_tags": "",
93
+ "metadata_title": "",
94
+ "min_bucket_reso": 256,
95
+ "min_snr_gamma": 0,
96
+ "min_timestep": 0,
97
+ "mixed_precision": "bf16",
98
+ "mode_scale": 1.29,
99
+ "model_list": "custom",
100
+ "model_prediction_type": "raw",
101
+ "multi_gpu": false,
102
+ "multires_noise_discount": 0.3,
103
+ "multires_noise_iterations": 0,
104
+ "no_token_padding": false,
105
+ "noise_offset": 0,
106
+ "noise_offset_random_strength": false,
107
+ "noise_offset_type": "Original",
108
+ "num_cpu_threads_per_process": 2,
109
+ "num_machines": 1,
110
+ "num_processes": 1,
111
+ "optimizer": "Adafactor",
112
+ "optimizer_args": "scale_parameter=False relative_step=False warmup_init=False weight_decay=0.01",
113
+ "output_dir": "/workspace/final_training_models/model",
114
+ "output_name": "clement_niece",
115
+ "persistent_data_loader_workers": false,
116
+ "pretrained_model_name_or_path": "/workspace/flux1-dev.safetensors",
117
+ "prior_loss_weight": 1,
118
+ "random_crop": false,
119
+ "reg_data_dir": "",
120
+ "resume": "",
121
+ "resume_from_huggingface": "",
122
+ "sample_every_n_epochs": 0,
123
+ "sample_every_n_steps": 0,
124
+ "sample_prompts": "",
125
+ "sample_sampler": "euler_a",
126
+ "save_clip": false,
127
+ "save_every_n_epochs": 100,
128
+ "save_every_n_steps": 0,
129
+ "save_last_n_epochs_state": 0,
130
+ "save_last_n_steps": 0,
131
+ "save_last_n_steps_state": 0,
132
+ "save_model_as": "safetensors",
133
+ "save_precision": "fp16",
134
+ "save_state": false,
135
+ "save_state_on_train_end": false,
136
+ "save_state_to_huggingface": false,
137
+ "save_t5xxl": false,
138
+ "scale_v_pred_loss_like_noise_pred": false,
139
+ "sd3_cache_text_encoder_outputs": false,
140
+ "sd3_cache_text_encoder_outputs_to_disk": false,
141
+ "sd3_checkbox": false,
142
+ "sd3_text_encoder_batch_size": 1,
143
+ "sdxl": false,
144
+ "sdxl_cache_text_encoder_outputs": false,
145
+ "sdxl_no_half_vae": false,
146
+ "seed": 1,
147
+ "shuffle_caption": false,
148
+ "single_blocks_to_swap": 0,
149
+ "skip_cache_check": false,
150
+ "split_mode": false,
151
+ "stop_text_encoder_training": 0,
152
+ "t5xxl": "",
153
+ "t5xxl_device": "",
154
+ "t5xxl_dtype": "bf16",
155
+ "t5xxl_max_token_length": 512,
156
+ "timestep_sampling": "sigmoid",
157
+ "train_batch_size": 7,
158
+ "train_blocks": "all",
159
+ "train_data_dir": "/workspace/final_training_models/img",
160
+ "v2": false,
161
+ "v_parameterization": false,
162
+ "v_pred_like_loss": 0,
163
+ "vae": "",
164
+ "vae_batch_size": 4,
165
+ "wandb_api_key": "",
166
+ "wandb_run_name": "",
167
+ "weighted_captions": false,
168
+ "weighting_scheme": "logit_normal",
169
+ "xformers": "sdpa"
170
+ }
config_dreambooth-20241023-104334.toml ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ adaptive_noise_scale = 0
2
+ ae = "/home/Ubuntu/Downloads/ae.safetensors"
3
+ blocks_to_swap = 0
4
+ bucket_no_upscale = true
5
+ bucket_reso_steps = 64
6
+ cache_latents = true
7
+ cache_latents_to_disk = true
8
+ cache_text_encoder_outputs = true
9
+ cache_text_encoder_outputs_to_disk = true
10
+ caption_dropout_every_n_epochs = 0
11
+ caption_dropout_rate = 0
12
+ caption_extension = ".txt"
13
+ clip_l = "/home/Ubuntu/Downloads/clip_l.safetensors"
14
+ discrete_flow_shift = 3.1582
15
+ double_blocks_to_swap = 0
16
+ dynamo_backend = "no"
17
+ epoch = 100
18
+ full_bf16 = true
19
+ fused_backward_pass = true
20
+ gradient_accumulation_steps = 1
21
+ gradient_checkpointing = true
22
+ guidance_scale = 1
23
+ huber_c = 0.1
24
+ huber_schedule = "snr"
25
+ keep_tokens = 0
26
+ learning_rate = 1e-5
27
+ learning_rate_te = 0
28
+ logging_dir = "/workspace/final_training_models/log"
29
+ loss_type = "l2"
30
+ lr_scheduler = "constant"
31
+ lr_scheduler_args = []
32
+ lr_scheduler_num_cycles = 1
33
+ lr_scheduler_power = 1
34
+ lr_warmup_steps = 0
35
+ max_bucket_reso = 2048
36
+ max_data_loader_n_workers = 0
37
+ max_timestep = 1000
38
+ max_token_length = 75
39
+ max_train_steps = 72
40
+ mem_eff_save = true
41
+ min_bucket_reso = 256
42
+ mixed_precision = "bf16"
43
+ model_prediction_type = "raw"
44
+ multires_noise_discount = 0.3
45
+ multires_noise_iterations = 0
46
+ noise_offset = 0
47
+ noise_offset_type = "Original"
48
+ optimizer_args = [ "scale_parameter=False", "relative_step=False", "warmup_init=False", "weight_decay=0.01",]
49
+ optimizer_type = "Adafactor"
50
+ output_dir = "/workspace/final_training_models/model"
51
+ output_name = "clement_niece"
52
+ persistent_data_loader_workers = 0
53
+ pretrained_model_name_or_path = "/workspace/flux1-dev.safetensors"
54
+ prior_loss_weight = 1
55
+ resolution = "1024,1024"
56
+ sample_prompts = "/workspace/final_training_models/model/sample/prompt.txt"
57
+ sample_sampler = "euler_a"
58
+ save_every_n_epochs = 100
59
+ save_model_as = "safetensors"
60
+ save_precision = "fp16"
61
+ sdpa = true
62
+ seed = 1
63
+ single_blocks_to_swap = 0
64
+ t5xxl = "/home/Ubuntu/Downloads/t5xxl_fp16.safetensors"
65
+ t5xxl_max_token_length = 512
66
+ timestep_sampling = "sigmoid"
67
+ train_batch_size = 7
68
+ train_blocks = "all"
69
+ train_data_dir = "/workspace/final_training_models/img"
70
+ vae_batch_size = 4
71
+ wandb_run_name = "clement_niece"
config_dreambooth-20241023-104644.toml ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ adaptive_noise_scale = 0
2
+ ae = "/workspace/ae.safetensors"
3
+ blocks_to_swap = 0
4
+ bucket_no_upscale = true
5
+ bucket_reso_steps = 64
6
+ cache_latents = true
7
+ cache_latents_to_disk = true
8
+ cache_text_encoder_outputs = true
9
+ cache_text_encoder_outputs_to_disk = true
10
+ caption_dropout_every_n_epochs = 0
11
+ caption_dropout_rate = 0
12
+ caption_extension = ".txt"
13
+ clip_l = "/workspace/clip_l.safetensors"
14
+ discrete_flow_shift = 3.1582
15
+ double_blocks_to_swap = 0
16
+ dynamo_backend = "no"
17
+ epoch = 100
18
+ full_bf16 = true
19
+ fused_backward_pass = true
20
+ gradient_accumulation_steps = 1
21
+ gradient_checkpointing = true
22
+ guidance_scale = 1
23
+ huber_c = 0.1
24
+ huber_schedule = "snr"
25
+ keep_tokens = 0
26
+ learning_rate = 1e-5
27
+ learning_rate_te = 0
28
+ logging_dir = "/workspace/final_training_models/log"
29
+ loss_type = "l2"
30
+ lr_scheduler = "constant"
31
+ lr_scheduler_args = []
32
+ lr_scheduler_num_cycles = 1
33
+ lr_scheduler_power = 1
34
+ lr_warmup_steps = 0
35
+ max_bucket_reso = 2048
36
+ max_data_loader_n_workers = 0
37
+ max_timestep = 1000
38
+ max_token_length = 75
39
+ max_train_steps = 72
40
+ mem_eff_save = true
41
+ min_bucket_reso = 256
42
+ mixed_precision = "bf16"
43
+ model_prediction_type = "raw"
44
+ multires_noise_discount = 0.3
45
+ multires_noise_iterations = 0
46
+ noise_offset = 0
47
+ noise_offset_type = "Original"
48
+ optimizer_args = [ "scale_parameter=False", "relative_step=False", "warmup_init=False", "weight_decay=0.01",]
49
+ optimizer_type = "Adafactor"
50
+ output_dir = "/workspace/final_training_models/model"
51
+ output_name = "clement_niece"
52
+ persistent_data_loader_workers = 0
53
+ pretrained_model_name_or_path = "/workspace/flux1-dev.safetensors"
54
+ prior_loss_weight = 1
55
+ resolution = "1024,1024"
56
+ sample_prompts = "/workspace/final_training_models/model/sample/prompt.txt"
57
+ sample_sampler = "euler_a"
58
+ save_every_n_epochs = 100
59
+ save_model_as = "safetensors"
60
+ save_precision = "fp16"
61
+ sdpa = true
62
+ seed = 1
63
+ single_blocks_to_swap = 0
64
+ t5xxl = "/workspace/t5xxl_fp16.safetensors"
65
+ t5xxl_max_token_length = 512
66
+ timestep_sampling = "sigmoid"
67
+ train_batch_size = 7
68
+ train_blocks = "all"
69
+ train_data_dir = "/workspace/final_training_models/img"
70
+ vae_batch_size = 4
71
+ wandb_run_name = "clement_niece"
sample/.ipynb_checkpoints/prompt-checkpoint.txt ADDED
File without changes
sample/prompt.txt ADDED
File without changes