ppbrown commited on
Commit
44d4e50
·
verified ·
1 Parent(s): 62ca5dc

Upload OneTrainer-XLsd32-phase1-LaionPlusWD.json with huggingface_hub

Browse files
OneTrainer-XLsd32-phase1-LaionPlusWD.json ADDED
@@ -0,0 +1,370 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version": 6,
3
+ "training_method": "FINE_TUNE",
4
+ "model_type": "STABLE_DIFFUSION_15",
5
+ "debug_mode": false,
6
+ "debug_dir": "debug",
7
+ "workspace_dir": "workspace/run",
8
+ "cache_dir": "workspace-cache/run",
9
+ "tensorboard": true,
10
+ "tensorboard_expose": true,
11
+ "tensorboard_port": 6006,
12
+ "validation": false,
13
+ "validate_after": 1,
14
+ "validate_after_unit": "EPOCH",
15
+ "continue_last_backup": false,
16
+ "include_train_config": "ALL",
17
+ "base_model_name": "/data/OneTrainer32/models/XLsd32-base-v0.safetensors",
18
+ "weight_dtype": "FLOAT_32",
19
+ "output_dtype": "FLOAT_32",
20
+ "output_model_format": "SAFETENSORS",
21
+ "output_model_destination": "models/XLsd32-dlionb16a8-phase1-LAION.sft",
22
+ "gradient_checkpointing": "ON",
23
+ "enable_async_offloading": true,
24
+ "enable_activation_offloading": true,
25
+ "layer_offload_fraction": 0.0,
26
+ "force_circular_padding": false,
27
+ "concept_file_name": "training_concepts/XLsd32.json",
28
+ "concepts": null,
29
+ "aspect_ratio_bucketing": true,
30
+ "latent_caching": true,
31
+ "clear_cache_before_training": false,
32
+ "learning_rate_scheduler": "CONSTANT",
33
+ "custom_learning_rate_scheduler": null,
34
+ "scheduler_params": [],
35
+ "learning_rate": 1.0,
36
+ "learning_rate_warmup_steps": 0.0,
37
+ "learning_rate_cycles": 1.0,
38
+ "learning_rate_min_factor": 0.0,
39
+ "epochs": 100,
40
+ "batch_size": 16,
41
+ "gradient_accumulation_steps": 8,
42
+ "ema": "OFF",
43
+ "ema_decay": 0.999,
44
+ "ema_update_step_interval": 5,
45
+ "dataloader_threads": 8,
46
+ "train_device": "cuda",
47
+ "temp_device": "cpu",
48
+ "train_dtype": "FLOAT_32",
49
+ "fallback_train_dtype": "FLOAT_32",
50
+ "enable_autocast_cache": true,
51
+ "only_cache": false,
52
+ "resolution": "512",
53
+ "frames": "25",
54
+ "attention_mechanism": "SDP",
55
+ "align_prop": false,
56
+ "align_prop_probability": 0.1,
57
+ "align_prop_loss": "AESTHETIC",
58
+ "align_prop_weight": 0.01,
59
+ "align_prop_steps": 20,
60
+ "align_prop_truncate_steps": 0.5,
61
+ "align_prop_cfg_scale": 7.0,
62
+ "mse_strength": 1.0,
63
+ "mae_strength": 0.0,
64
+ "log_cosh_strength": 0.0,
65
+ "vb_loss_strength": 1.0,
66
+ "loss_weight_fn": "DEBIASED_ESTIMATION",
67
+ "loss_weight_strength": 5.0,
68
+ "dropout_probability": 0.0,
69
+ "loss_scaler": "NONE",
70
+ "learning_rate_scaler": "NONE",
71
+ "clip_grad_norm": 1.0,
72
+ "offset_noise_weight": 0.0,
73
+ "perturbation_noise_weight": 0.0,
74
+ "rescale_noise_scheduler_to_zero_terminal_snr": false,
75
+ "force_v_prediction": false,
76
+ "force_epsilon_prediction": false,
77
+ "min_noising_strength": 0.0,
78
+ "max_noising_strength": 1.0,
79
+ "timestep_distribution": "UNIFORM",
80
+ "noising_weight": 0.0,
81
+ "noising_bias": 0.0,
82
+ "timestep_shift": 1.0,
83
+ "dynamic_timestep_shifting": false,
84
+ "unet": {
85
+ "__version": 0,
86
+ "model_name": "",
87
+ "include": true,
88
+ "train": true,
89
+ "stop_training_after": 0,
90
+ "stop_training_after_unit": "NEVER",
91
+ "learning_rate": null,
92
+ "weight_dtype": "FLOAT_32",
93
+ "dropout_probability": 0.0,
94
+ "train_embedding": true,
95
+ "attention_mask": false,
96
+ "guidance_scale": 1.0
97
+ },
98
+ "prior": {
99
+ "__version": 0,
100
+ "model_name": "",
101
+ "include": true,
102
+ "train": true,
103
+ "stop_training_after": 0,
104
+ "stop_training_after_unit": "NEVER",
105
+ "learning_rate": null,
106
+ "weight_dtype": "NONE",
107
+ "dropout_probability": 0.0,
108
+ "train_embedding": true,
109
+ "attention_mask": false,
110
+ "guidance_scale": 1.0
111
+ },
112
+ "text_encoder": {
113
+ "__version": 0,
114
+ "model_name": "",
115
+ "include": true,
116
+ "train": false,
117
+ "stop_training_after": 30,
118
+ "stop_training_after_unit": "EPOCH",
119
+ "learning_rate": null,
120
+ "weight_dtype": "FLOAT_32",
121
+ "dropout_probability": 0.0,
122
+ "train_embedding": true,
123
+ "attention_mask": false,
124
+ "guidance_scale": 1.0
125
+ },
126
+ "text_encoder_layer_skip": 0,
127
+ "text_encoder_2": {
128
+ "__version": 0,
129
+ "model_name": "",
130
+ "include": true,
131
+ "train": true,
132
+ "stop_training_after": 30,
133
+ "stop_training_after_unit": "EPOCH",
134
+ "learning_rate": null,
135
+ "weight_dtype": "NONE",
136
+ "dropout_probability": 0.0,
137
+ "train_embedding": true,
138
+ "attention_mask": false,
139
+ "guidance_scale": 1.0
140
+ },
141
+ "text_encoder_2_layer_skip": 0,
142
+ "text_encoder_3": {
143
+ "__version": 0,
144
+ "model_name": "",
145
+ "include": true,
146
+ "train": true,
147
+ "stop_training_after": 30,
148
+ "stop_training_after_unit": "EPOCH",
149
+ "learning_rate": null,
150
+ "weight_dtype": "NONE",
151
+ "dropout_probability": 0.0,
152
+ "train_embedding": true,
153
+ "attention_mask": false,
154
+ "guidance_scale": 1.0
155
+ },
156
+ "text_encoder_3_layer_skip": 0,
157
+ "vae": {
158
+ "__version": 0,
159
+ "model_name": "",
160
+ "include": true,
161
+ "train": true,
162
+ "stop_training_after": null,
163
+ "stop_training_after_unit": "NEVER",
164
+ "learning_rate": null,
165
+ "weight_dtype": "FLOAT_32",
166
+ "dropout_probability": 0.0,
167
+ "train_embedding": true,
168
+ "attention_mask": false,
169
+ "guidance_scale": 1.0
170
+ },
171
+ "effnet_encoder": {
172
+ "__version": 0,
173
+ "model_name": "",
174
+ "include": true,
175
+ "train": true,
176
+ "stop_training_after": null,
177
+ "stop_training_after_unit": "NEVER",
178
+ "learning_rate": null,
179
+ "weight_dtype": "NONE",
180
+ "dropout_probability": 0.0,
181
+ "train_embedding": true,
182
+ "attention_mask": false,
183
+ "guidance_scale": 1.0
184
+ },
185
+ "decoder": {
186
+ "__version": 0,
187
+ "model_name": "",
188
+ "include": true,
189
+ "train": true,
190
+ "stop_training_after": null,
191
+ "stop_training_after_unit": "NEVER",
192
+ "learning_rate": null,
193
+ "weight_dtype": "NONE",
194
+ "dropout_probability": 0.0,
195
+ "train_embedding": true,
196
+ "attention_mask": false,
197
+ "guidance_scale": 1.0
198
+ },
199
+ "decoder_text_encoder": {
200
+ "__version": 0,
201
+ "model_name": "",
202
+ "include": true,
203
+ "train": true,
204
+ "stop_training_after": null,
205
+ "stop_training_after_unit": "NEVER",
206
+ "learning_rate": null,
207
+ "weight_dtype": "NONE",
208
+ "dropout_probability": 0.0,
209
+ "train_embedding": true,
210
+ "attention_mask": false,
211
+ "guidance_scale": 1.0
212
+ },
213
+ "decoder_vqgan": {
214
+ "__version": 0,
215
+ "model_name": "",
216
+ "include": true,
217
+ "train": true,
218
+ "stop_training_after": null,
219
+ "stop_training_after_unit": "NEVER",
220
+ "learning_rate": null,
221
+ "weight_dtype": "NONE",
222
+ "dropout_probability": 0.0,
223
+ "train_embedding": true,
224
+ "attention_mask": false,
225
+ "guidance_scale": 1.0
226
+ },
227
+ "masked_training": false,
228
+ "unmasked_probability": 0.1,
229
+ "unmasked_weight": 0.1,
230
+ "normalize_masked_area_loss": false,
231
+ "embedding_learning_rate": null,
232
+ "preserve_embedding_norm": false,
233
+ "embedding": {
234
+ "__version": 0,
235
+ "uuid": "8ea9e884-0a1d-4aa5-9f2f-1ab6ba24ee12",
236
+ "model_name": "",
237
+ "placeholder": "<embedding>",
238
+ "train": true,
239
+ "stop_training_after": null,
240
+ "stop_training_after_unit": "NEVER",
241
+ "token_count": 1,
242
+ "initial_embedding_text": "*"
243
+ },
244
+ "additional_embeddings": [],
245
+ "embedding_weight_dtype": "FLOAT_32",
246
+ "cloud": {
247
+ "__version": 0,
248
+ "enabled": false,
249
+ "type": "RUNPOD",
250
+ "file_sync": "NATIVE_SCP",
251
+ "create": true,
252
+ "name": "OneTrainer",
253
+ "tensorboard_tunnel": true,
254
+ "sub_type": "",
255
+ "gpu_type": "",
256
+ "volume_size": 100,
257
+ "min_download": 0,
258
+ "remote_dir": "/workspace",
259
+ "huggingface_cache_dir": "/workspace/huggingface_cache",
260
+ "onetrainer_dir": "/workspace/OneTrainer",
261
+ "install_cmd": "git clone https://github.com/Nerogar/OneTrainer",
262
+ "install_onetrainer": true,
263
+ "update_onetrainer": true,
264
+ "detach_trainer": false,
265
+ "run_id": "job1",
266
+ "download_samples": true,
267
+ "download_output_model": true,
268
+ "download_saves": true,
269
+ "download_backups": false,
270
+ "download_tensorboard": false,
271
+ "delete_workspace": false,
272
+ "on_finish": "NONE",
273
+ "on_error": "NONE",
274
+ "on_detached_finish": "NONE",
275
+ "on_detached_error": "NONE"
276
+ },
277
+ "peft_type": "LORA",
278
+ "lora_model_name": "",
279
+ "lora_rank": 16,
280
+ "lora_alpha": 1.0,
281
+ "lora_decompose": false,
282
+ "lora_decompose_norm_epsilon": true,
283
+ "lora_weight_dtype": "FLOAT_32",
284
+ "lora_layers": "",
285
+ "lora_layer_preset": null,
286
+ "bundle_additional_embeddings": true,
287
+ "optimizer": {
288
+ "__version": 0,
289
+ "optimizer": "DADAPT_LION",
290
+ "adam_w_mode": false,
291
+ "alpha": null,
292
+ "amsgrad": false,
293
+ "beta1": 0.9,
294
+ "beta2": 0.999,
295
+ "beta3": null,
296
+ "bias_correction": false,
297
+ "block_wise": false,
298
+ "capturable": false,
299
+ "centered": false,
300
+ "clip_threshold": null,
301
+ "d0": 1e-06,
302
+ "d_coef": null,
303
+ "dampening": null,
304
+ "decay_rate": null,
305
+ "decouple": false,
306
+ "differentiable": false,
307
+ "eps": null,
308
+ "eps2": null,
309
+ "foreach": false,
310
+ "fsdp_in_use": false,
311
+ "fused": false,
312
+ "fused_back_pass": false,
313
+ "growth_rate": null,
314
+ "initial_accumulator_value": null,
315
+ "initial_accumulator": null,
316
+ "is_paged": false,
317
+ "log_every": 0,
318
+ "lr_decay": null,
319
+ "max_unorm": null,
320
+ "maximize": false,
321
+ "min_8bit_size": null,
322
+ "momentum": null,
323
+ "nesterov": false,
324
+ "no_prox": false,
325
+ "optim_bits": null,
326
+ "percentile_clipping": null,
327
+ "r": null,
328
+ "relative_step": false,
329
+ "safeguard_warmup": false,
330
+ "scale_parameter": false,
331
+ "stochastic_rounding": true,
332
+ "use_bias_correction": false,
333
+ "use_triton": false,
334
+ "warmup_init": false,
335
+ "weight_decay": 0.0,
336
+ "weight_lr_power": null,
337
+ "decoupled_decay": false,
338
+ "fixed_decay": false,
339
+ "rectify": false,
340
+ "degenerated_to_sgd": false,
341
+ "k": null,
342
+ "xi": null,
343
+ "n_sma_threshold": null,
344
+ "ams_bound": false,
345
+ "adanorm": false,
346
+ "adam_debias": false,
347
+ "slice_p": null,
348
+ "cautious": false
349
+ },
350
+ "optimizer_defaults": {},
351
+ "sample_definition_file_name": "training_samples/samples.json",
352
+ "samples": null,
353
+ "sample_after": 2000,
354
+ "sample_after_unit": "STEP",
355
+ "sample_skip_first": 0,
356
+ "sample_image_format": "PNG",
357
+ "sample_video_format": "MP4",
358
+ "sample_audio_format": "MP3",
359
+ "samples_to_tensorboard": false,
360
+ "non_ema_sampling": false,
361
+ "backup_after": 30,
362
+ "backup_after_unit": "NEVER",
363
+ "rolling_backup": false,
364
+ "rolling_backup_count": 3,
365
+ "backup_before_save": false,
366
+ "save_every": 1,
367
+ "save_every_unit": "EPOCH",
368
+ "save_skip_first": 0,
369
+ "save_filename_prefix": "XLsd32-dlionb16a8-phase1-LAION-"
370
+ }