Nerogar/OneTrainer issues and pull requests

#119 - Add CosineAnnealingLR for Prodigy

Pull Request - State: closed - Opened by ejektaflex 9 months ago - 2 comments

#117 - Not starting on fresh install

Issue - State: closed - Opened by saintbrodie 9 months ago - 2 comments

#116 - Tried to use same settings I find in Kohya but training failed epic

Issue - State: closed - Opened by FurkanGozukara 9 months ago - 82 comments

#115 - IoError(Os { code: 5, kind: PermissionDenied, message: "Access is denied." })

Issue - State: open - Opened by FurkanGozukara 9 months ago - 8 comments
Labels: bug

#114 - Add a way to see if the currently training epoch is training the text encoder

Issue - State: open - Opened by driqeks 9 months ago
Labels: enhancement

#113 - Total time running and estimated remaining calculator please.

Issue - State: open - Opened by Tr1dae 9 months ago
Labels: enhancement

#112 - Some UI is cropped on Linux with no way to resize it

Issue - State: closed - Opened by driqeks 9 months ago - 3 comments
Labels: bug

#111 - install.sh and start-ui.sh do not detect Python correctly on Ubuntu

Issue - State: closed - Opened by driqeks 9 months ago - 1 comment

#109 - "Stop Training" button doesn't stop writing debug images

Issue - State: open - Opened by dioxic 9 months ago
Labels: enhancement

#107 - Add ! to the start of all unix scripts

Pull Request - State: closed - Opened by magnusviri 9 months ago

#102 - disable non-ema sample, disable tensorboard sample

Pull Request - State: closed - Opened by SirTrippsalot 9 months ago

#101 - Meta.json error

Issue - State: closed - Opened by cseander 9 months ago - 2 comments

#100 - SDXL Finetuning using CLIPTextModel to encode both prompts during training?

Issue - State: closed - Opened by orcinus 9 months ago - 3 comments

#99 - Add option to bind Tensorboard to all interfaces

Pull Request - State: closed - Opened by orcinus 9 months ago

#98 - Add quick and dirty SDXL latent2rgb conversion

Pull Request - State: closed - Opened by orcinus 9 months ago - 1 comment

#97 - [BUG] Incorrect number of steps in each epoch.

Issue - State: closed - Opened by Nisekoixmy 9 months ago - 5 comments

#96 - Training dataset not shuffled between epochs

Issue - State: closed - Opened by orcinus 9 months ago - 5 comments
Labels: bug

#95 - Detailed installation process for linux

Pull Request - State: closed - Opened by TheForgotten69 9 months ago

#94 - "missing cond_stage_model.clip_l.text_projection, cond_stage_model.clip_l.logit_scale" when loading SDXL finetunes in Comfy

Issue - State: closed - Opened by orcinus 9 months ago - 3 comments

#90 - UI gets stuck in a transparent mode

Issue - State: open - Opened by bash-j 9 months ago - 4 comments
Labels: bug

#88 - Debug folder training batch images broken or actual inference broken?

Issue - State: closed - Opened by orcinus 9 months ago - 3 comments

#87 - Issue with having too many concepts.

Issue - State: closed - Opened by Pychnight 10 months ago - 1 comment

#86 - Provides a Dockerfile and example build instructions

Pull Request - State: closed - Opened by dougbtv 10 months ago - 6 comments

#84 - Captions generated are very short

Issue - State: open - Opened by Aamir3d 10 months ago - 2 comments
Labels: enhancement, good first issue

#82 - CUDA/Bitsandbytes Errors in Arch Linux

Issue - State: closed - Opened by PresTrembleyIIIEsq 10 months ago - 5 comments

#81 - How to use in Linux?

Issue - State: closed - Opened by kikoferrer 10 months ago - 4 comments

#80 - Error on manual install: WARNING: Did not find branch or tag '9d7c08c

Issue - State: closed - Opened by lunar-studio 10 months ago - 1 comment

#79 - Colab version

Issue - State: closed - Opened by Niutonian 10 months ago

#78 - "optimizer got an empty parameter list"

Issue - State: closed - Opened by orcinus 10 months ago - 10 comments
Labels: bug

#77 - Consistency decoder train feature?

Issue - State: open - Opened by dill-shower 10 months ago

#76 - Implementing support for PIXART-α Fine Tuning and DreamBooth

Issue - State: closed - Opened by FurkanGozukara 10 months ago - 7 comments
Labels: enhancement

#75 - Add scripts to calculate the loss per image.

Pull Request - State: closed - Opened by hameerabbasi 10 months ago

#74 - LR Scaling, Loss Scaling, MAE, Loss type weighting

Pull Request - State: closed - Opened by SirTrippsalot 10 months ago

#73 - [Enhancement] New features about text augmentation

Issue - State: open - Opened by Nisekoixmy 10 months ago - 2 comments
Labels: enhancement

#72 - Unable to start training for SDXL LoRA

Issue - State: closed - Opened by suwagawaki 10 months ago - 5 comments

#71 - Specific training data naming requirements?

Issue - State: closed - Opened by bneigher 10 months ago - 2 comments

#69 - does the training scripts support multi-gpu training?

Issue - State: closed - Opened by zhangvia 10 months ago - 1 comment

#68 - Unable to create backup, which crashes training.

Issue - State: closed - Opened by gurilagardnr 10 months ago - 1 comment
Labels: bug

#66 - Add support for SSD-1B

Issue - State: open - Opened by nikitabalakin 10 months ago - 3 comments
Labels: enhancement

#65 - [Bug] Cannot open optimizer settings

Issue - State: closed - Opened by Nisekoixmy 10 months ago - 1 comment

#64 - can you make a dockerfile?

Issue - State: closed - Opened by FMsunyh 11 months ago
Labels: enhancement

#63 - add Linux bash scripts

Pull Request - State: closed - Opened by Lolzen 11 months ago - 5 comments

#62 - Is there any threshold to the limit the tokens?

Issue - State: closed - Opened by Nisekoixmy 11 months ago - 1 comment

#61 - Error when 'Normalize Masked Area Loss' option is toggled on while 'Masked Training' is turned off

Issue - State: closed - Opened by Elldreth 11 months ago - 1 comment
Labels: bug

#59 - Fix align-prop dtype to match data.

Pull Request - State: closed - Opened by hameerabbasi 11 months ago - 1 comment

#58 - Revert to CUDA 11.8 as CUDA 12.1 doesn't work yet in a lot of environments.

Pull Request - State: closed - Opened by hameerabbasi 11 months ago

#57 - Lora training fails with latest update despite having alignprop disabled via UI/train.sh

Issue - State: closed - Opened by suwagawaki 11 months ago - 2 comments

#56 - Ensure PyTorch compatible with xformers is fetched.

Pull Request - State: closed - Opened by hameerabbasi 11 months ago

#55 - bug report after updating to the latest version

Issue - State: closed - Opened by Nisekoixmy 11 months ago

#54 - fixed eps tuple wrong var

Pull Request - State: closed - Opened by SirTrippsalot 11 months ago

#53 - Add Adafactor scheduler

Pull Request - State: closed - Opened by hameerabbasi 11 months ago

#52 - Issue with sampling

Issue - State: closed - Opened by vgaggia 11 months ago - 3 comments
Labels: bug

#51 - Enable generation of non-binary masks.

Pull Request - State: closed - Opened by hameerabbasi 11 months ago - 8 comments

#50 - Bugfix #Errors handling Bool in conditional optimizers logic

Pull Request - State: closed - Opened by SirTrippsalot 11 months ago

#49 - Refactor or logic with if else

Pull Request - State: closed - Opened by SirTrippsalot 11 months ago

#48 - Bugfix for optimizers window close not saving

Pull Request - State: closed - Opened by SirTrippsalot 11 months ago

#47 - SDXL: lora_te2_text_projection is not found in created LoRA modules.

Issue - State: closed - Opened by MysticDaedra 11 months ago - 5 comments

#46 - Bitsandbytes error on launch

Issue - State: closed - Opened by Signorlimone 11 months ago - 2 comments

#45 - Enhanced Optimizers with dynamic UI and per optimizer Defaults

Pull Request - State: closed - Opened by SirTrippsalot 12 months ago

#44 - Tutorials?

Issue - State: closed - Opened by DiggyDre 12 months ago - 5 comments

#43 - Any way to set optimizer arguments?

Issue - State: closed - Opened by sashasubbbb 12 months ago - 1 comment

#42 - Simple UI Switch for Continued Training

Pull Request - State: closed - Opened by Janca 12 months ago - 1 comment

#41 - Is it compatible with linux?

Issue - State: closed - Opened by sneccc 12 months ago - 10 comments

#40 - option to disable gradient checkpointing

Issue - State: closed - Opened by Metal079 12 months ago - 1 comment
Labels: enhancement

#36 - Enhanced Loss Calculation (6 methods)

Pull Request - State: closed - Opened by SirTrippsalot about 1 year ago

#35 - Improved Loss Function (Non-Masked)

Pull Request - State: closed - Opened by SirTrippsalot about 1 year ago

#34 - resizing UI window

Issue - State: open - Opened by chifeisoong about 1 year ago
Labels: bug

#33 - adding optimizers: adagrad, rmsprop (with 8 bit)

Pull Request - State: closed - Opened by SirTrippsalot about 1 year ago

#32 - Issues with VAE Finetuning

Issue - State: closed - Opened by iszotic about 1 year ago - 3 comments
Labels: bug

#31 - Use less vram ram on save and fix minor issue where the model isn't saved if backup before save is disabled

Pull Request - State: closed - Opened by allenbenz about 1 year ago - 1 comment

#30 - Error when training embedding on SDXL 1.0

Issue - State: closed - Opened by DanPli about 1 year ago - 1 comment
Labels: enhancement

#29 - queue system?

Issue - State: open - Opened by nikitabalakin about 1 year ago - 1 comment
Labels: enhancement

#28 - Can't finetune sdxl with a 4090, out of memory

Issue - State: closed - Opened by Zovya about 1 year ago - 12 comments
Labels: bug

#27 - RuntimeError: "grid_sampler_2d_cuda" not implemented for 'BFloat16'

Issue - State: closed - Opened by discordianbelle about 1 year ago - 1 comment
Labels: bug

#26 - Error during sampling

Issue - State: closed - Opened by TheForgotten69 about 1 year ago - 3 comments

#25 - KeyError: 'tokens_1' while training sdxl

Issue - State: closed - Opened by Zovya about 1 year ago - 2 comments

#24 - fix sdsampler bug

Pull Request - State: closed - Opened by FinFanFin about 1 year ago

#23 - fix a tensorboard bug

Pull Request - State: closed - Opened by FinFanFin about 1 year ago

#22 - Where to put training data directory

Issue - State: closed - Opened by cobanov about 1 year ago - 1 comment

#21 - add dadaptation and lion optimizers

Pull Request - State: closed - Opened by FinFanFin about 1 year ago

#20 - Strange results in training, picture and settings attached

Issue - State: closed - Opened by Signorlimone about 1 year ago - 2 comments
Labels: workflow

#19 - Disable dataset caching

Issue - State: closed - Opened by ygean about 1 year ago - 5 comments
Labels: enhancement

#18 - Use list of arguments instead of a string for Popen()

Pull Request - State: closed - Opened by float-trip about 1 year ago

#17 - Error: missing 1 required positional argument: 'extra_model_name'

Issue - State: closed - Opened by DanPli about 1 year ago - 1 comment

#16 - REQ Add support for prodigy optimizer, esp. before/if lycoris is added

Issue - State: closed - Opened by DanPli about 1 year ago - 2 comments
Labels: enhancement

GitHub / Nerogar/OneTrainer issues and pull requests