kohya-ss/sd-scripts

View on GitHub

Stars: 7.2k
Forks: 1.2k
Open beginner issues: 3
Indexed issues: 783
Avg merge: 11m
Merged PRs (30d): 2

Dominant language: Python
License: Apache-2.0
Last GitHub push: Sep 9, 2026
Latest indexed: Sep 18, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: help wanted

Beginner friendly Everything

Open All

783 open issues indexed

`torch.cuda.amp` in `__main__` self-test blocks and the IPEX hijack — deprecated since torch 2.3/2.4, unguarded on the main install path Open

Difficulty 3/5 1-2 days Newbie friendliness 78/100

kohya-ss/sd-scripts#2420 · Opened Aug 20, 2026
[Feature Request] Add option to disable cuDNN/MIOpen backend (workaround for AMD RDNA 4 GPU crashes) Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

kohya-ss/sd-scripts#2417 · Opened Aug 12, 2026
WeightsUnpickler error: Unsupported operand 56 Open

Difficulty 2/5 1-3 hours Newbie friendliness 68/100

kohya-ss/sd-scripts#2414 · Opened Aug 8, 2026
[bug] --skip_until_initial_step 使用時、--gradient_accumulation_steps > 1 で再開位置を過剰にスキップして学習が終了する Open

Difficulty 3/5 1-2 days Newbie friendliness 70/100

kohya-ss/sd-scripts#2407 · 1 comment · Opened Jul 20, 2026
--tag_replacement doesn't actually replace anything (variable shadowing bug) Open

Difficulty 1/5 Under an hour Newbie friendliness 88/100

kohya-ss/sd-scripts#2405 · Opened Jul 19, 2026
Silent `adaln` drop while anima lora training Open

Difficulty 3/5 1-2 days Newbie friendliness 50/100

kohya-ss/sd-scripts#2403 · Opened Jul 16, 2026
LECO: verify_training_args() crashes on missing sample_every_n_epochs/sample_every_n_steps Open

Difficulty 2/5 1-3 hours Newbie friendliness 72/100

kohya-ss/sd-scripts#2402 · Opened Jul 8, 2026
Adding Anima ControlNet support Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

kohya-ss/sd-scripts#2394 · 4 reactions · Opened Jun 20, 2026
"Missing keys in state dict" solved when training lora on some anima ft checkpoints Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2392 · 4 reactions · Opened Jun 18, 2026
bf16/float RuntimeError - dtype incompatibility Open

Difficulty 3/5 1-2 days Newbie friendliness 38/100

kohya-ss/sd-scripts#2364 · Opened May 29, 2026
Tag info saved in model metadata ignores keep_tokens_separator Open

Difficulty 3/5 1-2 days Newbie friendliness 55/100

kohya-ss/sd-scripts#2323 · Opened May 6, 2026
about PSOFT Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

kohya-ss/sd-scripts#2314 · Opened Apr 16, 2026
loss=nan training lora base on sdxl Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2313 · 1 comment · Opened Apr 9, 2026
Does Anima model support DoRA training? Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

kohya-ss/sd-scripts#2308 · 9 comments · Opened Apr 4, 2026
caption_dropout_rate should be 0 in validation dataset Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2292 · 1 comment · Opened Mar 22, 2026
Validation does not work with Anima LoRa Open

Difficulty 3/5 1-2 days Newbie friendliness 55/100

kohya-ss/sd-scripts#2286 · 2 comments · Opened Mar 15, 2026
SIGFPE / Floating Point Exception using Prodigy Open

Difficulty 4/5 3-5 days Newbie friendliness 30/100

kohya-ss/sd-scripts#2271 · Opened Feb 17, 2026
Significant bottleneck with text encoder outputs and latent caching to disk with .npz vs .safetensors Open

Difficulty 5/5 Over a week Newbie friendliness 45/100

kohya-ss/sd-scripts#2266 · 3 comments · Opened Feb 14, 2026
[Feature Request] Dynamic padding to nearest multiple of 75 (e.g., 75/150.../750) for large token limits Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

kohya-ss/sd-scripts#2256 · Opened Jan 25, 2026
flux klein support? Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

kohya-ss/sd-scripts#2250 · 1 comment · Opened Jan 16, 2026
[Bug]: AttributeError: module 'library.train_util' has no attribute 'load_tokenizer' in v25.2.1 Open

Difficulty 3/5 1-2 days Newbie friendliness 42/100

kohya-ss/sd-scripts#2249 · 1 comment · Opened Jan 16, 2026
Issues when restarting training session Open

Difficulty 4/5 3-5 days Newbie friendliness 42/100

kohya-ss/sd-scripts#2248 · 2 comments · Opened Jan 8, 2026
Allow more granular control over which timesteps are chosen during training Open

Difficulty 5/5 Over a week Newbie friendliness 30/100

kohya-ss/sd-scripts#2247 · Opened Dec 24, 2025
Feature: Z-Image fine-tuning and LoRA training Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

kohya-ss/sd-scripts#2243 · 2 comments · 7 reactions · Opened Nov 27, 2025
Flux1: using sd-scripts trained LORA with diffusers Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

kohya-ss/sd-scripts#2242 · 4 comments · Opened Nov 25, 2025
Support Flux2 Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

kohya-ss/sd-scripts#2241 · 7 comments · 5 reactions · Opened Nov 25, 2025
Fresh Installed kohya_SS training not working? Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

kohya-ss/sd-scripts#2239 · Opened Nov 16, 2025
SD3 unnecessary train_t5xxl check results in error Open

Difficulty 2/5 1-3 hours Newbie friendliness 58/100

kohya-ss/sd-scripts#2235 · 1 comment · Opened Nov 10, 2025
学習させてもsafetensorデータが生成されない Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

kohya-ss/sd-scripts#2233 · 1 comment · Opened Nov 5, 2025
flux_extract_lora.py need update for extracting FLUX SRPO model Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2229 · 1 comment · Opened Oct 22, 2025
请保留GPU选择，即使只有一个GPU Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2228 · 2 comments · Opened Oct 18, 2025
Syntax error in lora_flux.py Open

Difficulty 1/5 Under an hour Newbie friendliness 68/100

kohya-ss/sd-scripts#2220 · Opened Oct 6, 2025
Exactly same libraries both Linux and Windows and Windows is widly slower what could be reason? FLUX Fine Tuning Open

Difficulty 4/5 3-5 days Newbie friendliness 30/100

kohya-ss/sd-scripts#2218 · 6 comments · Opened Oct 2, 2025
Is alpha calculated correctly? Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 45/100

kohya-ss/sd-scripts#2204 · 3 comments · Opened Sep 20, 2025
🟢[SDXL FEATURE REQ] EQ VAEs and padding patch [sd3 branch] Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

kohya-ss/sd-scripts#2189 · 2 reactions · Opened Aug 25, 2025
Color profiles of are not handled by the scripts Open

Difficulty 2/5 1-3 hours Newbie friendliness 58/100

kohya-ss/sd-scripts#2179 · 1 reaction · Opened Aug 17, 2025
Resuming training from a state that was resumed training from earlier state behaves weird Open

Difficulty 3/5 1-2 days Newbie friendliness 48/100

kohya-ss/sd-scripts#2171 · 1 comment · Opened Aug 10, 2025
resize_lora.py scale = network_alpha / network_dim TypeError: unsupported operand type(s) for /: 'NoneType' and 'NoneType' Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2169 · 1 comment · Opened Aug 5, 2025
FLUX.1-Krea-dev compatibility Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2166 · 9 comments · Opened Aug 1, 2025
Question about Flux attention implementation Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2164 · 1 comment · Opened Aug 1, 2025
sd-scripts for lumina training in colab doesnt work Open

Difficulty 4/5 3-5 days Newbie friendliness 20/100

kohya-ss/sd-scripts#2161 · Opened Jul 28, 2025
Cannot use diffusers pipeline for flux inference with full finetuning model Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2156 · Opened Jul 17, 2025
I would like to know if the current flux fine-tuning script supports fine-tuning training for T5 and CLIPL? Request assistance Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

kohya-ss/sd-scripts#2152 · Opened Jul 15, 2025
Add LoRA support for lumina_minimal_inference.py Open

enhancement

kohya-ss/sd-scripts#2142 · 2 reactions · 1 assignee · Opened Jul 5, 2025
How to train LoRa for svd Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

kohya-ss/sd-scripts#2139 · Opened Jun 30, 2025
[bug] resume_from is ignored in sdxl_train_textual_inversion.py Open

Difficulty 4/5 3-5 days Newbie friendliness 45/100

kohya-ss/sd-scripts#2132 · Opened Jun 22, 2025
cosmos Open

Difficulty 5/5 Over a week Newbie friendliness 18/100

kohya-ss/sd-scripts#2130 · 1 comment · 5 reactions · Opened Jun 20, 2025
[Bug] TypeError in prepare_buckets_latents.py with --skip_existing due to missing alpha_mask argument Open

Difficulty 1/5 Under an hour Newbie friendliness 68/100

kohya-ss/sd-scripts#2128 · Opened Jun 19, 2025
How can I switch between LoRA models dynamically? Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

kohya-ss/sd-scripts#2126 · 2 comments · Opened Jun 18, 2025
Support individual lora module level selection and settings for more granular training Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

kohya-ss/sd-scripts#2125 · Opened Jun 17, 2025

kohya-ss/sd-scripts

Get new issues in your inbox