Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / kohya-ss/sd-scripts issues and pull requests

#1918 - Remove v-pred warning.

Pull Request - State: open - Opened by tsukimiya 4 days ago

#1915 - Hardcoded bucket_step overrides configurable parameter

Issue - State: open - Opened by iqddd 6 days ago - 4 comments
Labels: bug

#1913 - Support loading more formats of flux diffusers models

Pull Request - State: open - Opened by rockerBOO 7 days ago

#1912 - Voluptuous is down

Issue - State: open - Opened by xingqiu774 8 days ago

#1910 - Optimize Latent Caching Speed with VAE Optimizations

Pull Request - State: open - Opened by alefh123 9 days ago - 2 comments

#1909 - Move progress bar to account for sampling image first

Pull Request - State: open - Opened by rockerBOO 9 days ago - 3 comments

#1908 - Move scale_weight_norms inside sync_gradients

Pull Request - State: open - Opened by rockerBOO 9 days ago

#1907 - how to set timestep shift when train flux-dev lora?

Issue - State: open - Opened by SlZeroth 9 days ago - 1 comment

#1906 - Add uv/pyproject and black config

Pull Request - State: open - Opened by rockerBOO 10 days ago - 1 comment

#1904 - Setting seeds to random after exhausted

Pull Request - State: open - Opened by DKnight54 11 days ago

#1903 - Val loss improvement

Pull Request - State: open - Opened by kohya-ss 11 days ago - 21 comments

#1900 - Validation Loss Enhancements

Pull Request - State: open - Opened by stepfunction83 11 days ago - 13 comments

#1899 - Val loss

Pull Request - State: closed - Opened by kohya-ss 12 days ago

#1898 - Implement Test and Validation Loss for Flux Finetuning and LoRA Training

Pull Request - State: closed - Opened by stepfunction83 12 days ago - 3 comments

#1897 - Implement Test and Validation Set Loss

Issue - State: open - Opened by stepfunction83 14 days ago - 2 comments

#1895 - Add more dpmsolver++2m samplers to solve SDXL sampling artifacts

Pull Request - State: open - Opened by rockerBOO 14 days ago - 1 comment

#1894 - Add pin_memory to DataLoader and update ImageInfo to support

Pull Request - State: open - Opened by rockerBOO 15 days ago - 2 comments

#1893 - Implement Flex.1 Alpha Functionality

Pull Request - State: open - Opened by stepfunction83 15 days ago - 5 comments

#1891 - Add Support for Flex.1 Alpha

Issue - State: open - Opened by stepfunction83 15 days ago - 21 comments

#1889 - Have any guide for training flux Fill lora, or fine-turn flux Fill?

Issue - State: open - Opened by CaiNiaoCYY 17 days ago - 2 comments

#1888 - Merging Multiple Loras - Not working properly

Issue - State: closed - Opened by GraftingRayman 17 days ago

#1885 - Unable to load SD3.5-Large for lora

Issue - State: closed - Opened by moonlightian 18 days ago - 1 comment

#1883 - Allow custom schedulers to inherit max training steps, etc.

Pull Request - State: open - Opened by uwidev 18 days ago

#1879 - merge dev to main

Pull Request - State: closed - Opened by kohya-ss 21 days ago - 2 comments

#1875 - Sana training support

Issue - State: closed - Opened by rockerBOO 24 days ago - 2 comments

#1874 - 学習させてもsafetensorデータが生成されない

Issue - State: open - Opened by m0r1s1 25 days ago - 3 comments

#1873 - Add scale map to apply_max_norm_regularization

Pull Request - State: open - Opened by rockerBOO 26 days ago - 3 comments

#1870 - Add rank_stabilized for networks

Pull Request - State: open - Opened by rockerBOO 28 days ago - 2 comments

#1869 - How to specify these model parameters in the training flux_train.py

Issue - State: closed - Opened by mobilejammer 29 days ago - 2 comments

#1868 - Huber loss not supported??

Issue - State: open - Opened by Deathawaits4 about 1 month ago - 3 comments

#1867 - support fused_back_pass for prodigy-plus-schedule-free

Pull Request - State: open - Opened by michP247 about 1 month ago - 8 comments

#1866 - support fused_backward_pass for prodigy-plus-schedule-free optimizer

Pull Request - State: closed - Opened by michP247 about 1 month ago

#1865 - fix always_first_tags

Pull Request - State: open - Opened by anonymousUI about 1 month ago

#1864 - Validation loss

Pull Request - State: closed - Opened by rockerBOO about 1 month ago - 31 comments

#1863 - How to train SDXL LoRA on specific blocks?

Issue - State: open - Opened by rishipandey125 about 1 month ago

#1862 - diffusers' api 'load_lora_weights()' not working for trained lora

Issue - State: closed - Opened by yaoliliu about 1 month ago - 1 comment

#1862 - diffusers' api 'load_lora_weights()' not working for trained lora

Issue - State: closed - Opened by yaoliliu about 1 month ago - 1 comment

#1861 - args.noise_offset is not working for flux training？

Issue - State: open - Opened by ziyaxuanyi about 1 month ago - 2 comments

#1860 - Text Encoders & Inference - SD3.5L Dreambooth 💫

Issue - State: open - Opened by deman311 about 1 month ago

#1859 - torch 2.4.0 does not work with fp8 training

Issue - State: open - Opened by ddpasa about 1 month ago - 1 comment

#1858 - Add Validation loss for LoRA training

Pull Request - State: closed - Opened by hinablue about 1 month ago - 7 comments

#1857 - Failed to offload sd3.5L to cpu

Issue - State: open - Opened by moonlightian about 1 month ago

#1857 - Failed to offload sd3.5L to cpu

Issue - State: open - Opened by moonlightian about 1 month ago

#1856 - Val loss (sd3 wip) (need help)

Pull Request - State: closed - Opened by gesen2egee about 1 month ago

#1856 - Val loss (sd3 wip) (need help)

Pull Request - State: closed - Opened by gesen2egee about 1 month ago

#1855 - Is sd3 branch also available for sdxl training?

Issue - State: closed - Opened by moonlightian about 1 month ago - 2 comments

#1855 - Is sd3 branch also available for sdxl training?

Issue - State: closed - Opened by moonlightian about 1 month ago - 2 comments

#1854 - Flux.1-dev lora training not support trigger word parameter？

Issue - State: open - Opened by Nine98 about 1 month ago - 1 comment

#1851 - any way to get t5xxl_fp16.safetensors as shown in the flux-tuning example

Issue - State: closed - Opened by wonderseen about 2 months ago - 2 comments

#1850 - Allow sampling and saving on demand using signals

Pull Request - State: open - Opened by fabbarix about 2 months ago - 5 comments

#1849 - Flux Controlnet Error while processing rearrange-reduction

Issue - State: open - Opened by WainWong about 2 months ago

#1846 - Black images generated while training Lora with Flux

Issue - State: closed - Opened by hoangvictor about 2 months ago - 1 comment

#1845 - Kohya SS runs "repeat" number as a file name?

Issue - State: open - Opened by yarwhip about 2 months ago - 1 comment

#1844 - Question: getting a Lora which crashed at sample generation step

Issue - State: closed - Opened by r1n0x about 2 months ago - 1 comment

#1843 - Muon optimizer support

Issue - State: open - Opened by kunibald413 about 2 months ago - 1 comment

#1841 - Feature Request: Add L0 Loss as an Option for `--loss_type`

Issue - State: open - Opened by Skelp about 2 months ago

#1840 - Does the SD3 branch support fine-tuning CLIP L and G?

Issue - State: closed - Opened by Ice-YY about 2 months ago

#1839 - support flux control net training

Issue - State: open - Opened by kunibald413 about 2 months ago

#1838 - Experimental Redux conditioning for Flux Lora training

Pull Request - State: open - Opened by recris about 2 months ago - 13 comments

#1836 - JPEGXL images not cropping/bucketing correctly

Issue - State: open - Opened by TeKett about 2 months ago

#1835 - lora打开错误

Issue - State: open - Opened by xiaowei52045 about 2 months ago - 2 comments

#1833 - How the lora weight be trained in the train_network.py ?

Issue - State: closed - Opened by phyllispeng123 about 2 months ago - 1 comment

#1832 - FLUX.1 fine-tuning, dataset_1024_bs1.toml not found

Issue - State: open - Opened by huangjun12 about 2 months ago - 2 comments

#1830 - Added RAdamScheduleFree support

Pull Request - State: closed - Opened by nhamanasu about 2 months ago - 2 comments

#1829 - Add distillation loss in Flux lora training

Issue - State: open - Opened by phyllispeng123 about 2 months ago

#1828 - Prevent git credentials from leaking into other actions

Pull Request - State: closed - Opened by rockerBOO about 2 months ago - 1 comment

#1827 - Support schnell lora training

Issue - State: open - Opened by phyllispeng123 about 2 months ago - 2 comments

#1827 - Support schnell lora training

Issue - State: open - Opened by phyllispeng123 about 2 months ago - 1 comment

#1824 - extract Flux LoRa missing

Issue - State: closed - Opened by yogotatara3 2 months ago

#1824 - extract Flux LoRa missing

Issue - State: closed - Opened by yogotatara3 2 months ago

#1823 - Skipping sample image generation if args.sample_prompts file is missing

Issue - State: open - Opened by blackmagic24 2 months ago - 1 comment
Labels: enhancement

#1823 - Skipping sample image generation if args.sample_prompts file is missing

Issue - State: open - Opened by blackmagic24 2 months ago - 1 comment
Labels: enhancement

#1822 - Bad Flux lora training result

Issue - State: open - Opened by phyllispeng123 2 months ago - 6 comments

#1822 - Bad Flux lora training result

Issue - State: open - Opened by phyllispeng123 2 months ago - 6 comments

#1821 - Unify controlnet parameters name and change scripts name.

Pull Request - State: closed - Opened by sdbds 2 months ago - 3 comments

#1820 - Support SageAttention

Issue - State: open - Opened by sdbds 2 months ago - 6 comments