39870411d8
More guidance work. Improved LoRA module resolver for unet. Added vega mappings and LoRA training for it. Various other bigfixes and changes
Jaret Burkett2023-12-15 06:02:10 -07:00
bd2bce9b92
Switched to trailing timestep spacing to make timesteps for consistant across schedulers. Honed in on targeted guidance. It is finally perfect. (I think)
Jaret Burkett2023-11-29 14:32:48 -07:00
537af79b0d
Merge branch 'main' of github.com:ostris/ai-toolkit
Jaret Burkett2023-11-29 10:13:55 -07:00
7624241032
More fixes for noise schedules and fixed targeted guidance inverted masked prior
Jaret Burkett2023-11-29 10:13:31 -07:00
93ea955d7c
Added refiner fine tuning. Works, but needs some polish.
Jaret Burkett2023-11-05 17:15:03 -07:00
8a9e8f708f
Added base for using guidance during training. Still not working right.
Jaret Burkett2023-11-05 04:03:32 -07:00
d35733ac06
Added support for training ssd-1B. Added support for saving models into diffusers format. We can currently save in safetensors format for ssd-1b, but diffusers cannot load it yet.
Jaret Burkett2023-11-03 05:01:16 -06:00
7d707b2fe6
Added masking to slider training. Something is still weird though
Jaret Burkett2023-11-01 14:51:29 -06:00
a899ec91c8
Added some split prompting started code, adamw8bit, replacements improving, learnable snr gos. A lot of good stuff.
Jaret Burkett2023-11-01 06:52:21 -06:00
436a09430e
Added flat snr gamma vs min. Fixes timestep timing
Jaret Burkett2023-10-29 15:41:55 -06:00
d742792ee4
Fixed issue with not loading short prompt
Jaret Burkett2023-10-24 16:19:32 -06:00
002279cec3
Allow short and long caption combinations like form the new captioning system. Merge the network into the model before inference and reextract when done. Doubles inference speed on locon models during inference. allow splitting a batch into individual components and run them through alone. Basicallt gradient accumulation with single batch size.
Jaret Burkett2023-10-24 16:02:07 -06:00
73c8b50975
Added ability to use adagrad from transformers
Jaret Burkett2023-10-24 11:16:01 -06:00
34eb563d55
Added dataset tagging and management tools using llava
Jaret Burkett2023-10-24 10:11:47 -06:00
07bf7bd7de
Allow augmentations and targeting different loss types fron the config file
Jaret Burkett2023-10-18 03:04:57 -06:00
da6302ada8
added a method to apply multipliers to noise and latents prior to combining
Jaret Burkett2023-10-17 06:09:16 -06:00
a05459afaf
Fixed issue with adapters that only had 1 input channel. Added ability to set the percentage chance of adapter matching
Jaret Burkett2023-10-15 15:13:35 -06:00
38e441a29c
allow flipping for point of interesting autocropping. allow num repeats. Fixed some bugs with new free u
Jaret Burkett2023-10-12 21:02:47 -06:00
f4c90bb589
Added config to set the min value of a mask
Jaret Burkett2023-10-09 15:47:54 -06:00
bb1d3793e3
Added ability to add masks to dataloader and sd trainer to adjust weight of image
Jaret Burkett2023-10-09 11:21:00 -06:00
1d3de678aa
fixed bug with trigger word embedding. Allow control images to load from the dataloader or legacy way
Jaret Burkett2023-10-09 06:21:49 -06:00
b1cfafa0c6
always prompt both even when training only one encoder
Jaret Burkett2023-10-07 11:23:09 -06:00
cac8754399
Allow for training loras on onle one text encoder for sdxl
Jaret Burkett2023-10-06 08:11:56 -06:00
f73402473b
Bug fixes. Added some functionality to help with private extensions
Jaret Burkett2023-10-05 07:09:34 -06:00
579650eaf8
Fixed big issue with bucketing dataloader and added random cripping to a point of interest
Jaret Burkett2023-10-02 18:31:08 -06:00
320e109c5f
Allow loading from a json detail file for captions
Jaret Burkett2023-10-02 13:25:09 -06:00
560251a24f
fixed issue with down block residuals when doing slider cfg on sdxl with t2i adapter assisted training
Jaret Burkett2023-10-01 07:32:48 -06:00
e5153d87c9
Fixed issues with dataloader bucketing. Allow using standard base image for t2i adapters.
Jaret Burkett2023-09-24 05:19:57 -06:00
830e87cb87
Added IP adapter training. Not functioning correctly yet
Jaret Burkett2023-09-24 02:39:43 -06:00
19255cdc7c
Bugfixes. Added small augmentations to dataloader. Will switch to abluminations soon though. Added ability to adjust step count on start to override what is in the file
Jaret Burkett2023-09-20 05:30:10 -06:00
0f105690cc
Added some further extendability for plugins
Jaret Burkett2023-09-19 05:41:44 -06:00
61badf85a7
t2i training working from what I can tell at least
Jaret Burkett2023-09-17 15:56:43 -06:00
181f237a7b
added flipping x and y for dataset loader
Jaret Burkett2023-09-17 08:42:54 -06:00
c698837241
Fixes to esrgan trainer. Moved logic for sd prompt embeddings out of diffusers pipeline so I can manipulate it
Jaret Burkett2023-09-16 17:41:07 -06:00
27f343fc08
Added base setup for training t2i adapters. Currently untested, saw something else shiny i wanted to finish sirst. Added content_or_style to the training config. It defaults to balanced, which is standard uniform time step sampling. If style or content is passed, it will use cubic sampling for timesteps to favor timesteps that are beneficial for training them. for style, favor later timesteps. For content, favor earlier timesteps.
Jaret Burkett2023-09-16 08:30:38 -06:00
17e4fe40d7
Prevent lycoris network moduels if not training that part of network. Skew timesteps to favor later steps. It performs better
Jaret Burkett2023-09-14 15:13:24 -06:00
569d7464d5
implemented device placement preset system more places. Vastly improved speed on setting network multiplier and activating network. Fixed timing issues on progress bar
Jaret Burkett2023-09-14 08:31:54 -06:00
Jaret Burkett
bendeguzvaradi