API tools faq
paste
Advertisement
Guest User

Stable Cascade Lora Prodigy - Preset

a guest
Mar 2nd, 2024
355
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 9.82 KB | None | 0 0
raw download clone embed print report
  1. {
  2. "__version": 2,
  3. "training_method": "LORA",
  4. "model_type": "STABLE_CASCADE_1",
  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": false,
  11. "continue_last_backup": false,
  12. "include_train_config": "NONE",
  13. "base_model_name": "stabilityai/stable-cascade-prior",
  14. "weight_dtype": "BFLOAT_16",
  15. "output_dtype": "BFLOAT_16",
  16. "output_model_format": "SAFETENSORS",
  17. "output_model_destination": "./workspace/run/save/ak31-1716-50.safetensors",
  18. "gradient_checkpointing": true,
  19. "concept_file_name": "training_concepts/concepts.json",
  20. "concepts": null,
  21. "circular_mask_generation": false,
  22. "random_rotate_and_crop": false,
  23. "aspect_ratio_bucketing": true,
  24. "latent_caching": true,
  25. "clear_cache_before_training": true,
  26. "learning_rate_scheduler": "CONSTANT",
  27. "learning_rate": 1.0,
  28. "learning_rate_warmup_steps": 200,
  29. "learning_rate_cycles": 1,
  30. "epochs": 50,
  31. "batch_size": 1,
  32. "gradient_accumulation_steps": 1,
  33. "ema": "OFF",
  34. "ema_decay": 0.999,
  35. "ema_update_step_interval": 5,
  36. "train_device": "cuda",
  37. "temp_device": "cpu",
  38. "train_dtype": "FLOAT_16",
  39. "fallback_train_dtype": "BFLOAT_16",
  40. "only_cache": false,
  41. "resolution": "1024",
  42. "attention_mechanism": "XFORMERS",
  43. "align_prop": false,
  44. "align_prop_probability": 0.1,
  45. "align_prop_loss": "AESTHETIC",
  46. "align_prop_weight": 0.01,
  47. "align_prop_steps": 20,
  48. "align_prop_truncate_steps": 0.5,
  49. "align_prop_cfg_scale": 7.0,
  50. "mse_strength": 1.0,
  51. "mae_strength": 0.0,
  52. "vb_loss_strength": 1.0,
  53. "min_snr_gamma": 3.0,
  54. "loss_scaler": "NONE",
  55. "learning_rate_scaler": "NONE",
  56. "offset_noise_weight": 0.06,
  57. "perturbation_noise_weight": 0.0,
  58. "rescale_noise_scheduler_to_zero_terminal_snr": false,
  59. "force_v_prediction": false,
  60. "force_epsilon_prediction": false,
  61. "min_noising_strength": 0.0,
  62. "max_noising_strength": 1.0,
  63. "noising_weight": 0.0,
  64. "noising_bias": 0.5,
  65. "unet": {
  66. "__version": 0,
  67. "model_name": "",
  68. "train": true,
  69. "stop_training_after": 0,
  70. "stop_training_after_unit": "NEVER",
  71. "learning_rate": null,
  72. "weight_dtype": "NONE"
  73. },
  74. "prior": {
  75. "__version": 0,
  76. "model_name": "",
  77. "train": true,
  78. "stop_training_after": 0,
  79. "stop_training_after_unit": "NEVER",
  80. "learning_rate": null,
  81. "weight_dtype": "NONE"
  82. },
  83. "text_encoder": {
  84. "__version": 0,
  85. "model_name": "",
  86. "train": true,
  87. "stop_training_after": 50,
  88. "stop_training_after_unit": "EPOCH",
  89. "learning_rate": null,
  90. "weight_dtype": "NONE"
  91. },
  92. "text_encoder_layer_skip": 0,
  93. "text_encoder_2": {
  94. "__version": 0,
  95. "model_name": "",
  96. "train": true,
  97. "stop_training_after": 30,
  98. "stop_training_after_unit": "EPOCH",
  99. "learning_rate": null,
  100. "weight_dtype": "NONE"
  101. },
  102. "text_encoder_2_layer_skip": 0,
  103. "vae": {
  104. "__version": 0,
  105. "model_name": "",
  106. "train": true,
  107. "stop_training_after": null,
  108. "stop_training_after_unit": "NEVER",
  109. "learning_rate": null,
  110. "weight_dtype": "FLOAT_32"
  111. },
  112. "effnet_encoder": {
  113. "__version": 0,
  114. "model_name": "./models/effnet_encoder.safetensors",
  115. "train": true,
  116. "stop_training_after": null,
  117. "stop_training_after_unit": "NEVER",
  118. "learning_rate": null,
  119. "weight_dtype": "FLOAT_16"
  120. },
  121. "decoder": {
  122. "__version": 0,
  123. "model_name": "stabilityai/stable-cascade",
  124. "train": true,
  125. "stop_training_after": null,
  126. "stop_training_after_unit": "NEVER",
  127. "learning_rate": null,
  128. "weight_dtype": "FLOAT_16"
  129. },
  130. "decoder_text_encoder": {
  131. "__version": 0,
  132. "model_name": "",
  133. "train": true,
  134. "stop_training_after": null,
  135. "stop_training_after_unit": "NEVER",
  136. "learning_rate": null,
  137. "weight_dtype": "NONE"
  138. },
  139. "decoder_vqgan": {
  140. "__version": 0,
  141. "model_name": "",
  142. "train": true,
  143. "stop_training_after": null,
  144. "stop_training_after_unit": "NEVER",
  145. "learning_rate": null,
  146. "weight_dtype": "NONE"
  147. },
  148. "masked_training": false,
  149. "unmasked_probability": 0.1,
  150. "unmasked_weight": 0.1,
  151. "normalize_masked_area_loss": false,
  152. "embeddings": [
  153. {
  154. "__version": 0,
  155. "model_name": "",
  156. "train": true,
  157. "stop_training_after": null,
  158. "stop_training_after_unit": "NEVER",
  159. "token_count": 1,
  160. "initial_embedding_text": "*",
  161. "weight_dtype": "FLOAT_32"
  162. }
  163. ],
  164. "embedding_weight_dtype": "FLOAT_32",
  165. "lora_model_name": "",
  166. "lora_rank": 64,
  167. "lora_alpha": 1.0,
  168. "lora_weight_dtype": "FLOAT_32",
  169. "optimizer": {
  170. "__version": 0,
  171. "optimizer": "PRODIGY",
  172. "adam_w_mode": false,
  173. "alpha": null,
  174. "amsgrad": false,
  175. "beta1": 0.9,
  176. "beta2": 0.999,
  177. "beta3": null,
  178. "bias_correction": false,
  179. "block_wise": false,
  180. "capturable": false,
  181. "centered": false,
  182. "clip_threshold": null,
  183. "d0": 1e-06,
  184. "d_coef": 1.0,
  185. "dampening": null,
  186. "decay_rate": null,
  187. "decouple": true,
  188. "differentiable": false,
  189. "eps": 1e-08,
  190. "eps2": null,
  191. "foreach": false,
  192. "fsdp_in_use": false,
  193. "fused": false,
  194. "growth_rate": "inf",
  195. "initial_accumulator_value": null,
  196. "is_paged": false,
  197. "log_every": null,
  198. "lr_decay": null,
  199. "max_unorm": null,
  200. "maximize": false,
  201. "min_8bit_size": null,
  202. "momentum": null,
  203. "nesterov": false,
  204. "no_prox": false,
  205. "optim_bits": null,
  206. "percentile_clipping": null,
  207. "relative_step": false,
  208. "safeguard_warmup": true,
  209. "scale_parameter": false,
  210. "use_bias_correction": true,
  211. "use_triton": false,
  212. "warmup_init": false,
  213. "weight_decay": 0.0
  214. },
  215. "optimizer_defaults": {
  216. "ADAFACTOR": {
  217. "__version": 0,
  218. "optimizer": "ADAFACTOR",
  219. "adam_w_mode": false,
  220. "alpha": null,
  221. "amsgrad": false,
  222. "beta1": null,
  223. "beta2": null,
  224. "beta3": null,
  225. "bias_correction": false,
  226. "block_wise": false,
  227. "capturable": false,
  228. "centered": false,
  229. "clip_threshold": 1.0,
  230. "d0": null,
  231. "d_coef": null,
  232. "dampening": null,
  233. "decay_rate": -0.8,
  234. "decouple": false,
  235. "differentiable": false,
  236. "eps": 1e-30,
  237. "eps2": 0.001,
  238. "foreach": false,
  239. "fsdp_in_use": false,
  240. "fused": false,
  241. "growth_rate": null,
  242. "initial_accumulator_value": null,
  243. "is_paged": false,
  244. "log_every": null,
  245. "lr_decay": null,
  246. "max_unorm": null,
  247. "maximize": false,
  248. "min_8bit_size": null,
  249. "momentum": null,
  250. "nesterov": false,
  251. "no_prox": false,
  252. "optim_bits": null,
  253. "percentile_clipping": null,
  254. "relative_step": false,
  255. "safeguard_warmup": false,
  256. "scale_parameter": false,
  257. "use_bias_correction": false,
  258. "use_triton": false,
  259. "warmup_init": false,
  260. "weight_decay": 0.0
  261. },
  262. "PRODIGY": {
  263. "__version": 0,
  264. "optimizer": "PRODIGY",
  265. "adam_w_mode": false,
  266. "alpha": null,
  267. "amsgrad": false,
  268. "beta1": 0.9,
  269. "beta2": 0.999,
  270. "beta3": null,
  271. "bias_correction": false,
  272. "block_wise": false,
  273. "capturable": false,
  274. "centered": false,
  275. "clip_threshold": null,
  276. "d0": 1e-06,
  277. "d_coef": 1.0,
  278. "dampening": null,
  279. "decay_rate": null,
  280. "decouple": true,
  281. "differentiable": false,
  282. "eps": 1e-08,
  283. "eps2": null,
  284. "foreach": false,
  285. "fsdp_in_use": false,
  286. "fused": false,
  287. "growth_rate": "inf",
  288. "initial_accumulator_value": null,
  289. "is_paged": false,
  290. "log_every": null,
  291. "lr_decay": null,
  292. "max_unorm": null,
  293. "maximize": false,
  294. "min_8bit_size": null,
  295. "momentum": null,
  296. "nesterov": false,
  297. "no_prox": false,
  298. "optim_bits": null,
  299. "percentile_clipping": null,
  300. "relative_step": false,
  301. "safeguard_warmup": true,
  302. "scale_parameter": false,
  303. "use_bias_correction": true,
  304. "use_triton": false,
  305. "warmup_init": false,
  306. "weight_decay": 0.0
  307. }
  308. },
  309. "sample_definition_file_name": "training_samples/samples.json",
  310. "samples": null,
  311. "sample_after": 10,
  312. "sample_after_unit": "MINUTE",
  313. "sample_image_format": "JPG",
  314. "samples_to_tensorboard": true,
  315. "non_ema_sampling": true,
  316. "backup_after": 30,
  317. "backup_after_unit": "MINUTE",
  318. "rolling_backup": false,
  319. "rolling_backup_count": 3,
  320. "backup_before_save": true,
  321. "save_after": 4,
  322. "save_after_unit": "EPOCH"
  323. }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!