nodes/edenartlab-sd-lora-trainer
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
361 Commits 4 Branches 0 Tags
b166f7c2745ede1d9d6294a08951eb6661ffa287
Commit Graph
87 Commits
Author SHA1 Message Date
xander 6a4fa1ef8d add progressbar to node 2024-07-18 21:04:43 +02:00
xander 3cd17086f3 trainer comfyui v1 2024-07-18 05:50:04 +02:00
xander 01523350e1 cleanup save dir 2024-07-18 04:00:21 +02:00
xander 85599dc725 minor changes 2024-07-18 03:43:43 +02:00
xander 4f084d251c updates for comfyui node 2024-07-18 03:12:11 +02:00
aiXander fb521e9dbd add print flush 2024-07-16 04:39:29 -07:00
aiXander 340336b53d print config pre training start 2024-07-11 06:11:33 -07:00
xander b5cf857b1f push small changes 2024-07-09 18:33:52 +02:00
xander e1813c75d1 updates to avoid OOM when plotting hist 2024-07-06 22:18:47 +02:00
xander 892cf61c52 add disable_ti flag 2024-07-06 18:43:03 +02:00
xander fdd531d3c5 fix full finetuning and add 8bitadam 2024-07-06 17:42:49 +02:00
xander faf864ce15 tiny tweak to learning rates for SDXL 2024-07-05 17:07:33 +02:00
xander b02221b23c cleanup before sd3 integration 2024-06-13 14:24:32 +02:00
xander d8b3175b57 cleanup before sd3 integration 2024-06-13 13:57:20 +02:00
xander 41ddfbcca6 add resource monitoring 2024-05-29 23:33:42 +02:00
aiXander eb144d5212 increase starting lr for sd15 2024-05-16 16:46:08 -07:00
aiXander 3f7ba82a25 tiny fixes 2024-05-16 16:02:46 -07:00
aiXander 6e32be4dfd bugfix 2024-05-16 14:31:57 -07:00
aiXander 9102e8e5e5 render with active pipe when not in debug mode 2024-05-16 09:28:22 -07:00
aiXander 3f7298633d update cog run settings 2024-05-15 14:59:48 -07:00
xander 4389ae75c8 force quit training after max steps 2024-05-14 15:21:48 +02:00
xander d66863b4cb more cleanup 2024-05-14 15:12:49 +02:00
xander b51727219e cleanup and try to fix last inference step 2024-05-14 11:42:57 +02:00
xander 7d78a3d76d updates 2024-04-29 19:37:45 +02:00
xander 45e6f50c24 small updates 2024-04-29 12:10:17 +02:00
xander 39ee9c9630 set defaults 2024-04-27 16:47:35 +02:00
xander 8da2f58c20 everything is working 2024-04-26 14:52:32 +02:00
xander 7d469d843c cleanup and make all this work also for sd15 2024-04-26 14:38:25 +02:00
xander 2e5912f629 improve token regularization 2024-04-26 13:32:01 +02:00
xander af6c4676d5 bugfixes 2024-04-25 21:41:57 +02:00
xander 9ae9bc19d2 mask improvements 2024-04-25 21:21:14 +02:00
xander 7218c7e8c2 more tok reg updates 2024-04-24 17:51:42 +02:00
xander e1b05b1b99 update loss plots, reduce tok_reg value 2024-04-24 14:42:05 +02:00
aiXander d4923ab6d2 tiny typo fix 2024-04-22 19:39:16 -07:00
aiXander b48419247a refactor token regularization 2024-04-22 19:08:36 -07:00
aiXander 4d9f8c2e67 more tweaks 2024-04-22 12:22:20 -07:00
aiXander 784bfbf14f prep for cog push 2024-04-22 11:27:04 -07:00
xander 41b7895bff add unet_prodigy_growth_factor 2024-04-22 14:36:04 +02:00
mayukhdeb 5b4020d448 save sd15 models as they're supposed to be saved 2024-04-20 06:21:24 -07:00
mayukhdeb a33a33295e never load from disk during training 2024-04-19 05:36:24 -07:00
aiXander 3e44dff1da remove pipe before loading from disk 2024-04-19 05:19:49 -07:00
aiXander 13733a4f19 Merge branch 'main' of https://github.com/edenartlab/trainer into main 2024-04-19 04:58:09 -07:00
aiXander ccaeab8e71 freeze_ti param, save txt-encoder loras into savetensors file 2024-04-19 04:57:59 -07:00
mayukhdeb b568c0c283 load ;checkpoint from disk once training is complete 2024-04-19 04:55:02 -07:00
aiXander d1d8eb125e fix 2024-04-19 01:41:28 -07:00
aiXander ea67904a9a simplify 2024-04-19 01:39:43 -07:00
aiXander 6f5253c1d4 all seems to work 2024-04-18 08:36:14 -07:00
aiXander f5a1de5ffe move token warmup to before lora init 2024-04-18 07:54:38 -07:00
aiXander 2af80788ca fix txt gradients 2024-04-18 07:50:39 -07:00
aiXander 284b934567 fix typo 2024-04-18 07:30:35 -07:00