Commit Graph

  • 856df07f49 Merge branch 'dev' into stable-cascade Kohya S 2024-02-18 09:15:12 +09:00
  • d1fb480887 format by black Kohya S 2024-02-18 09:13:24 +09:00
  • 80ef59c115 support text encoder training in stable cascade Kohya S 2024-02-18 09:12:37 +09:00
  • 319bbf8057 add stage c tmp training code Kohya S 2024-02-17 23:59:20 +09:00
  • fa440208b7 add inference script Kohya S 2024-02-17 17:57:30 +09:00
  • 75e4a951d0 update readme Kohya S 2024-02-17 12:04:12 +09:00
  • 42f3318e17 Merge pull request #1116 from kohya-ss/dev_device_support Kohya S 2024-02-17 11:58:02 +09:00
  • baa0e97ced Merge branch 'dev' into dev_device_support Kohya S 2024-02-17 11:54:07 +09:00
  • 71ebcc5e25 update readme and gradual latent doc Kohya S 2024-02-12 14:52:19 +09:00
  • 93bed60762 fix to work --console_log_xxx options Kohya S 2024-02-12 14:49:29 +09:00
  • 41d32c0be4 Merge pull request #1117 from kohya-ss/gradual_latent_hires_fix Kohya S 2024-02-12 14:21:27 +09:00
  • cbe9c5dc06 supprt deep shink with regional lora, add prompter module Kohya S 2024-02-12 14:17:27 +09:00
  • d3745db764 add args for logging Kohya S 2024-02-12 13:15:21 +09:00
  • 358ca205a3 Merge branch 'dev' into dev_device_support Kohya S 2024-02-12 13:01:54 +09:00
  • c748719115 fix indent Kohya S 2024-02-12 12:59:45 +09:00
  • 98f42d3a0b Merge branch 'dev' into gradual_latent_hires_fix Kohya S 2024-02-12 12:59:25 +09:00
  • 35c6053de3 Merge pull request #1104 from kohya-ss/dev_improve_log Kohya S 2024-02-12 11:33:32 +09:00
  • 20ae603221 Merge branch 'dev' into gradual_latent_hires_fix Kohya S 2024-02-12 11:26:36 +09:00
  • 672851e805 Merge branch 'dev' into dev_improve_log Kohya S 2024-02-12 11:24:33 +09:00
  • e579648ce9 fix help for highvram arg Kohya S 2024-02-12 11:12:41 +09:00
  • e24d9606a2 add clean_memory_on_device and use it from training Kohya S 2024-02-12 11:10:52 +09:00
  • 75ecb047e2 Merge branch 'dev' into dev_device_support Kohya S 2024-02-11 19:51:28 +09:00
  • f897d55781 Merge pull request #1113 from kohya-ss/dev_multi_gpu_sample_gen Kohya S 2024-02-11 19:49:08 +09:00
  • 7202596393 log to print tag frequencies Kohya S 2024-02-10 09:59:12 +09:00
  • 03f0816f86 the reason not working grad accum steps found. it was becasue of my accelerate settings BootsofLagrangian 2024-02-09 17:47:49 +09:00
  • 5d9e2873f6 make rich to output to stderr instead of stdout Kohya S 2024-02-08 21:38:02 +09:00
  • 055f02e1e1 add logging args for training scripts Kohya S 2024-02-08 21:16:42 +09:00
  • 9b8ea12d34 update log initialization without rich Kohya S 2024-02-08 21:06:39 +09:00
  • 74fe0453b2 add comment for get_preferred_device Kohya S 2024-02-08 20:58:54 +09:00
  • a98fecaeb1 forgot setting mixed_precision for deepspeed. sorry BootsofLagrangian 2024-02-07 17:19:46 +09:00
  • 2445a5b74e remove test requirements BootsofLagrangian 2024-02-07 16:48:18 +09:00
  • 62556619bd fix full_fp16 compatible and train_step BootsofLagrangian 2024-02-07 16:42:05 +09:00
  • 7d2a9268b9 apply offloading method runable for all trainer BootsofLagrangian 2024-02-05 22:42:06 +09:00
  • 3970bf4080 maybe fix branch to run offloading BootsofLagrangian 2024-02-05 22:40:43 +09:00
  • 4295f91dcd fix all trainer about vae BootsofLagrangian 2024-02-05 20:19:56 +09:00
  • 2824312d5e fix vae type error during training sdxl BootsofLagrangian 2024-02-05 20:13:28 +09:00
  • 64873c1b43 fix offload_optimizer_device typo BootsofLagrangian 2024-02-05 17:11:50 +09:00
  • efd3b58973 Add logging arguments and update logging setup Kohya S 2024-02-04 20:44:10 +09:00
  • 6279b33736 fallback to basic logging if rich is not installed Kohya S 2024-02-04 18:28:54 +09:00
  • 5f6bf29e52 Replace print with logger if they are logs (#905) Yuta Hayashibe 2024-02-04 16:14:34 +07:00
  • e793d7780d reduce peak VRAM in sample gen Kohya S 2024-02-04 17:31:01 +09:00
  • ae0872ba3b search block-wise application weights nw_application_weights ykume 2024-02-04 14:56:31 +09:00
  • 1492bcbfa2 add --new_conv_rank option mgz 2024-02-03 23:18:55 -06:00
  • bf2de5620c fix formatting in resize_lora.py mgz 2024-02-03 20:09:37 -06:00
  • dfe08f395f support deepspeed BootsofLagrangian 2024-02-04 03:12:42 +09:00
  • 6269682c56 unificaition of gen scripts for SD and SDXL, work in progress Kohya S 2024-02-03 23:33:48 +09:00
  • 2f9a344297 fix typo Kohya S 2024-02-03 23:26:57 +09:00
  • 11aced3500 simplify multi-GPU sample generation Kohya S 2024-02-03 22:25:29 +09:00
  • 1567ce1e17 Enable distributed sample image generation on multi-GPU enviroment (#1061) DKnight54 2024-02-03 20:46:31 +08:00
  • 5cca1fdc40 add highvram option and do not clear cache in caching latents Kohya S 2024-02-01 21:55:55 +09:00
  • 9f0f0d573d Merge pull request #1092 from Disty0/dev_device_support Kohya S 2024-02-01 20:41:21 +09:00
  • 716a92cbed Bump crate-ci/typos from 1.16.26 to 1.17.2 dependabot[bot] 2024-02-01 01:57:52 +00:00
  • a6a2b5a867 Fix IPEX support and add XPU device to device_utils Disty0 2024-01-31 17:32:37 +03:00
  • 2ca4d0c831 Merge pull request #1054 from akx/mps Kohya S 2024-01-31 21:30:12 +09:00
  • 7f948db158 Merge pull request #1087 from mgz-dev/fix-imports-on-svd_merge_lora Kohya S 2024-01-31 21:08:40 +09:00
  • 9d7729c00d Merge pull request #1086 from Disty0/dev Kohya S 2024-01-31 21:06:34 +09:00
  • 988dee02b9 IPEX torch.tensor FP64 workaround Disty0 2024-01-30 01:52:32 +03:00
  • d4b9568269 fix broken import in svd_merge_lora script mgz 2024-01-28 11:59:07 -06:00
  • ccc3a481e7 Update IPEX Libs Disty0 2024-01-28 14:14:31 +03:00
  • 8f6f734a6f Merge branch 'dev' into gradual_latent_hires_fix Kohya S 2024-01-28 08:21:15 +09:00
  • cd19df49cd Merge pull request #1085 from kohya-ss/dev v0.8.3 Kohya S 2024-01-27 18:32:06 +09:00
  • 736365bdd5 update README.md Kohya S 2024-01-27 18:31:01 +09:00
  • 6ceedb9448 Merge branch 'main' into dev Kohya S 2024-01-27 18:23:52 +09:00
  • 930a3912a7 Merge pull request #1084 from fireicewolf/devel Kohya S 2024-01-27 18:22:00 +09:00
  • cf790d87c4 Merge pull request #1079 from feffy380/fix/fp8savestate Kohya S 2024-01-26 22:34:35 +09:00
  • 4e67fb8444 test DukeG 2024-01-26 20:22:49 +08:00
  • 50f631c768 test DukeG 2024-01-26 20:02:48 +08:00
  • 85bc371ebc test DukeG 2024-01-26 18:58:47 +08:00
  • 322ee52c77 Update requirements.txt feffy380 2024-01-25 19:15:53 +01:00
  • c576f80639 Fix ControlNetLLLite training issue #1069 Kohya S 2024-01-25 18:43:07 +09:00
  • 478156b4f7 Refactor device determination to function; add MPS fallback Aarni Koskela 2024-01-16 15:01:59 +02:00
  • afc38707d5 Refactor memory cleaning into a single function Aarni Koskela 2024-01-16 14:47:44 +02:00
  • 2e4bee6f24 Log accelerator device Aarni Koskela 2024-01-16 15:10:24 +02:00
  • d5ab97b69b Merge pull request #1067 from kohya-ss/dev v0.8.2 Kohya S 2024-01-23 21:04:16 +09:00
  • 7cb44e4502 update readme Kohya S 2024-01-23 21:02:40 +09:00
  • 7a20df5ad5 Merge pull request #1064 from KohakuBlueleaf/fix-grad-sync Kohya S 2024-01-23 20:33:55 +09:00
  • bea4362e21 Merge pull request #1060 from akx/refactor-xpu-init Kohya S 2024-01-23 20:25:37 +09:00
  • 6805cafa9b fix TI training crashes in multigpu #1019 Kohya S 2024-01-23 20:17:19 +09:00
  • 711b40ccda Avoid always sync Kohaku-Blueleaf 2024-01-23 11:49:03 +08:00
  • 696dd7f668 Fix dtype issue in PyTorch 2.0 for generating samples in training sdxl network Kohya S 2024-01-22 12:43:37 +09:00
  • e0a3c69223 update readme Kohya S 2024-01-20 18:47:10 +09:00
  • c59249a664 Add options to reduce memory usage in extract_lora_from_models.py closes #1059 Kohya S 2024-01-20 18:45:54 +09:00
  • fef172966f Add network_multiplier for dataset and train LoRA Kohya S 2024-01-20 16:24:43 +09:00
  • 5a1ebc4c7c format by black Kohya S 2024-01-20 13:10:45 +09:00
  • 2a0f45aea9 update readme Kohya S 2024-01-20 11:08:20 +09:00
  • 1f77bb6e73 fix to work sample generation in fp8 ref #1057 Kohya S 2024-01-20 10:57:42 +09:00
  • a7ef6422b6 fix to work with torch 2.0 Kohya S 2024-01-20 10:00:30 +09:00
  • 9cfa68c92f [Experimental Feature] FP8 weight dtype for base model when running train_network (or sdxl_train_network) (#1057) Kohaku-Blueleaf 2024-01-20 08:46:53 +08:00
  • 6f3f701d3d Deduplicate ipex initialization code Aarni Koskela 2024-01-18 18:23:21 +02:00
  • d2a99a19d4 Merge pull request #1056 from kohya-ss/dev v0.8.1 Kohya S 2024-01-17 21:41:36 +09:00
  • 0395a35543 Merge branch 'main' into dev Kohya S 2024-01-17 21:39:13 +09:00
  • 987d4a969d update readme Kohya S 2024-01-17 21:38:49 +09:00
  • 976d092c68 fix text encodes are on gpu even when not trained Kohya S 2024-01-17 21:31:50 +09:00
  • e6b15c7e4a Merge pull request #1053 from akx/sdpa Kohya S 2024-01-16 21:50:45 +09:00
  • ef50436464 Fix typo --spda (it's --sdpa) Aarni Koskela 2024-01-16 14:32:48 +02:00
  • 26d35794e3 Merge pull request #1052 from kohya-ss/dev v0.8.0 Kohya S 2024-01-15 21:39:02 +09:00
  • dcf0eeb5b6 update readme Kohya S 2024-01-15 21:35:26 +09:00
  • 32b759a328 Add wandb_run_name parameter to init_kwargs #1032 Kohya S 2024-01-14 22:02:03 +09:00
  • 09ef3ffa8b Merge branch 'main' into dev Kohya S 2024-01-14 21:49:25 +09:00
  • aab265e431 Fix an issue with saving as diffusers sd1/2 model close #1033 Kohya S 2024-01-04 21:43:50 +09:00