kohya-ss/sd-scripts issues and pull requests

#1747 - Bump crate-ci/typos from 1.24.3 to 1.26.8

Pull Request - State: closed - Opened by dependabot[bot] 3 months ago - 1 comment
Labels: dependencies, github_actions

#1746 - Using TI Embeddings during training

Issue - State: open - Opened by DaStapo 3 months ago

#1746 - Using TI Embeddings during training

Issue - State: open - Opened by DaStapo 3 months ago

#1745 - Support LoRA train for Flux Schnnell or openFlux?

Issue - State: open - Opened by 4d4m0r 3 months ago - 1 comment

#1745 - Support LoRA train for Flux Schnnell or openFlux?

Issue - State: open - Opened by 4d4m0r 3 months ago - 1 comment

#1744 - --always_first_tags not working (possibly)

Issue - State: open - Opened by dpons222 3 months ago

#1744 - --always_first_tags not working (possibly)

Issue - State: open - Opened by dpons222 3 months ago

#1743 - Cannot apply trained SD 3.5 LoRA in Diffusers

Issue - State: closed - Opened by YujiaKCL 3 months ago - 4 comments

#1743 - Cannot apply trained SD 3.5 LoRA in Diffusers

Issue - State: closed - Opened by YujiaKCL 3 months ago - 4 comments

#1742 - not find sd3.5 training script and instruction in README

Issue - State: closed - Opened by JoshonSmith 3 months ago - 1 comment

#1742 - not find sd3.5 training script and instruction in README

Issue - State: closed - Opened by JoshonSmith 3 months ago - 1 comment

#1741 - Compute aspect ratios in log-space

Pull Request - State: open - Opened by catboxanon 3 months ago - 1 comment

#1741 - Compute aspect ratios in log-space

Pull Request - State: open - Opened by catboxanon 3 months ago

#1740 - Problem Training LoRa????

Issue - State: open - Opened by VeeBeeArt 3 months ago

#1740 - Problem Training LoRa????

Issue - State: open - Opened by VeeBeeArt 3 months ago

#1739 - A massive NPZ file when using alpha mask

Issue - State: open - Opened by gesen2egee 3 months ago - 1 comment

#1739 - A massive NPZ file when using alpha mask

Issue - State: open - Opened by gesen2egee 3 months ago - 1 comment

#1738 - save_last_n_steps is ignored SD 3 FLUX branch

Issue - State: closed - Opened by FurkanGozukara 3 months ago - 2 comments

#1737 - (sd3-Flux) returned non-zero exit status 3221225477. 13:02:14-584959 INFO Training has ended

Issue - State: closed - Opened by antey3064 3 months ago - 6 comments

#1737 - (sd3-Flux) returned non-zero exit status 3221225477. 13:02:14-584959 INFO Training has ended

Issue - State: closed - Opened by antey3064 3 months ago - 5 comments

#1736 - Unable to Detect Images in Kohya_SS Fine-Tuning on RunPod - Need Assistance

Issue - State: closed - Opened by Artemis1111 3 months ago - 5 comments

#1736 - Unable to Detect Images in Kohya_SS Fine-Tuning on RunPod - Need Assistance

Issue - State: closed - Opened by Artemis1111 3 months ago - 5 comments

#1735 - Odd technical problem.

Issue - State: open - Opened by AbstractEyes 3 months ago - 1 comment

#1735 - Odd technical problem.

Issue - State: open - Opened by AbstractEyes 3 months ago - 1 comment

#1734 - [Feature Request]Use Gzip for compress latent file size

Issue - State: open - Opened by sdbds 3 months ago - 3 comments

#1734 - [Feature Request]Use Gzip for compress latent file size

Issue - State: open - Opened by sdbds 3 months ago - 3 comments

#1733 - flux dev finetune, training loss drops from 0.5 and stays at around 0.34 for thousands of steps, no further changes

Issue - State: open - Opened by zzl1122 3 months ago - 3 comments

#1733 - flux dev finetune, training loss drops from 0.5 and stays at around 0.34 for thousands of steps, no further changes

Issue - State: open - Opened by zzl1122 3 months ago - 3 comments

#1732 - sd3.5 training supports toml configuration and accelerate multi-card training

Issue - State: open - Opened by mliand 3 months ago

#1732 - sd3.5 training supports toml configuration and accelerate multi-card training

Issue - State: open - Opened by mliand 3 months ago

#1730 - Improvment to lora.py and flux lora.py

Issue - State: open - Opened by luckiskind 3 months ago - 2 comments

#1730 - Improvment to lora.py and flux lora.py

Issue - State: open - Opened by luckiskind 3 months ago - 2 comments

#1729 - 'Max Token Length' setting for flux

Issue - State: closed - Opened by Vigilence 3 months ago - 1 comment

#1729 - 'Max Token Length' setting for flux

Issue - State: closed - Opened by Vigilence 3 months ago - 1 comment

#1727 - Feature Request: Save N epoch and N steps, comma seperated

Issue - State: open - Opened by FurkanGozukara 3 months ago - 3 comments

#1727 - Feature Request: Save N epoch and N steps, comma seperated

Issue - State: open - Opened by FurkanGozukara 3 months ago - 3 comments

#1726 - 35D Train Lora Error when sample。

Issue - State: closed - Opened by jinwei1660 3 months ago - 2 comments

#1726 - 35D Train Lora Error when sample。

Issue - State: closed - Opened by jinwei1660 3 months ago - 2 comments

#1725 - Error when train sd35 lora

Issue - State: closed - Opened by jinwei1660 3 months ago - 2 comments

#1725 - Error when train sd35 lora

Issue - State: closed - Opened by jinwei1660 3 months ago - 2 comments

#1724 - SD3.5のLora学習エラー

Issue - State: closed - Opened by waomodder 3 months ago - 7 comments

#1724 - SD3.5のLora学習エラー

Issue - State: closed - Opened by waomodder 3 months ago - 7 comments

#1723 - Flux how to use the lora model with T5xxl trained

Issue - State: open - Opened by tristanwqy 3 months ago - 3 comments

#1723 - Flux how to use the lora model with T5xxl trained

Issue - State: open - Opened by tristanwqy 3 months ago - 3 comments

#1722 - When uploading SDXL LoRA merged by sd-scripts on CivitAI, they reports "hash corrected"

Issue - State: closed - Opened by JujoHotaru 3 months ago - 7 comments

#1722 - When uploading SDXL LoRA merged by sd-scripts on CivitAI, they reports "hash corrected"

Issue - State: closed - Opened by JujoHotaru 3 months ago - 7 comments

#1721 - Training LoRA for Flux Multi GPU not working with more than 2 GPUs

Issue - State: open - Opened by AfterHAL 4 months ago - 4 comments

#1720 - Enabling dim_from_weights or loraplus_unet_lr_ratio will cause the error: "train_blocks must be single for split mode" (content updated).

Issue - State: open - Opened by avan06 4 months ago - 3 comments
Labels: bug

#1720 - Enabling dim_from_weights or loraplus_unet_lr_ratio will cause the error: "train_blocks must be single for split mode" (content updated).

Issue - State: open - Opened by avan06 4 months ago - 1 comment
Labels: bug

#1719 - SD3.5 Large support

Pull Request - State: closed - Opened by kohya-ss 4 months ago - 32 comments

#1719 - SD3.5 Large support

Pull Request - State: closed - Opened by kohya-ss 4 months ago - 32 comments

#1718 - What is the default scheduler used by flux lora inference? I don’t seem to see it in the optional parameters.

Issue - State: closed - Opened by JingZz7 4 months ago - 1 comment

#1718 - What is the default scheduler used by flux lora inference? I don’t seem to see it in the optional parameters.

Issue - State: closed - Opened by JingZz7 4 months ago - 1 comment

#1717 - Remove v-pred warnings

Pull Request - State: closed - Opened by catboxanon 4 months ago

#1717 - Remove v-pred warnings

Pull Request - State: closed - Opened by catboxanon 4 months ago

#1715 - Update debiased estimation loss function to accommodate V-pred

Pull Request - State: closed - Opened by catboxanon 4 months ago - 15 comments

#1715 - Update debiased estimation loss function to accommodate V-pred

Pull Request - State: closed - Opened by catboxanon 4 months ago - 15 comments

#1711 - "cat_cuda" not implemented for 'Float8_e4m3fn'

Issue - State: open - Opened by vneznaikin 4 months ago - 6 comments

#1710 - Differential Output Preservation loss for LoRA

Pull Request - State: closed - Opened by kohya-ss 4 months ago - 30 comments

#1710 - Differential Output Preservation loss for LoRA

Pull Request - State: closed - Opened by kohya-ss 4 months ago - 30 comments

#1707 - "mixed_precision fp16" not working for Flux

Issue - State: open - Opened by vneznaikin 4 months ago - 4 comments

#1707 - "mixed_precision fp16" not working for Flux

Issue - State: open - Opened by vneznaikin 4 months ago - 4 comments

#1702 - Flux Dedistilled / fluxdev2pro support ?

Issue - State: open - Opened by Tablaski 4 months ago - 29 comments

#1702 - Flux Dedistilled / fluxdev2pro support ?

Issue - State: open - Opened by Tablaski 4 months ago - 29 comments

#1700 - OOM using AdamW8bit since recent update

Issue - State: open - Opened by Tophness 4 months ago - 6 comments

#1695 - images sampled during LORA FLUX schnell training and after training are noisy

Issue - State: closed - Opened by JEFFSEVENTHSENSE 4 months ago - 5 comments

#1695 - images sampled during LORA FLUX schnell training and after training are noisy

Issue - State: closed - Opened by JEFFSEVENTHSENSE 4 months ago - 5 comments

#1686 - fix: fix some distributed training error in windows

Pull Request - State: closed - Opened by Akegarasu 4 months ago - 6 comments

#1684 - T5XXL tokenizer deadlocks when num workers>0

Issue - State: open - Opened by kohya-ss 4 months ago

#1683 - error while training lycoris for flux.1

Issue - State: open - Opened by AUTOMATIC2222 4 months ago - 3 comments

#1682 - GGUF finetune?

Issue - State: open - Opened by Vigilence 4 months ago

#1681 - Has anyone already done finetuning a flux model（not lora)? Why is the image quality of the model generated by my finetune so poor?

Issue - State: open - Opened by xiaohaier123 4 months ago - 5 comments

#1680 - Flux Results Not quite good after updating

Issue - State: open - Opened by zahdab 4 months ago

#1679 - Having NotImplementedError: Cannot copy out of meta tensor; no data! Issue

Issue - State: open - Opened by JEFFSEVENTHSENSE 4 months ago - 3 comments

#1678 - I have a few experiments I'd like to run training FLUX using weighted captions.

Issue - State: open - Opened by AbstractEyes 4 months ago - 1 comment

#1677 - No token padding for train_network

Pull Request - State: open - Opened by gesen2egee 4 months ago - 3 comments

#1676 - Having training issue with FLUX-schnell

Issue - State: closed - Opened by JEFFSEVENTHSENSE 4 months ago - 4 comments

#1674 - Training not starting when path name includes spaces

Issue - State: closed - Opened by Lexxxco 4 months ago - 1 comment

#1673 - flux finetune always fills shared memory

Issue - State: open - Opened by AndreRatzenberger 4 months ago - 5 comments

#1672 - Failure with Flux ALL MERGED model !!

Issue - State: open - Opened by muwaaaah 4 months ago

#1671 - Add Timestep Sampling Function from SD3 Branch to SD (dev base)

Pull Request - State: open - Opened by gesen2egee 4 months ago - 3 comments

#1670 - Cannot save Flux Dev checkpoint merged with LoRAs

Issue - State: open - Opened by PCanavelli 4 months ago

#1669 - Some flux fp8 models is returning an error, others don't why? Error "Cannot copy out of meta tensor..."

Issue - State: open - Opened by RafaRed 4 months ago - 1 comment

#1668 - Add Timestep Sampling Function from SD3 Branch to SD

Pull Request - State: closed - Opened by gesen2egee 4 months ago - 2 comments

#1667 - Not in train mode! error when resuming from schedulefree state

Issue - State: closed - Opened by gesen2egee 4 months ago - 1 comment

#1665 - NotImplementedError: Cannot copy out of meta tensor; no data! Please use torch.nn.Module.to_empty() instead of torch.nn.Module.to() when moving module from meta to a different device.

Issue - State: open - Opened by Wolchenok57 4 months ago

#1664 - Support min/max timesteps in train_controlnet.py

Pull Request - State: open - Opened by elthariel 4 months ago

#1663 - Dataset Maker - Could not find TensorRT

Issue - State: closed - Opened by Tulantuna1 4 months ago

#1662 - Lycoris issue

Issue - State: open - Opened by TeKett 4 months ago - 3 comments

#1661 - Bump crate-ci/typos from 1.24.3 to 1.25.0

Pull Request - State: closed - Opened by dependabot[bot] 4 months ago - 1 comment
Labels: dependencies, github_actions

#1660 - Fast image sizes

Pull Request - State: open - Opened by kohya-ss 4 months ago

#1659 - The “Windows Installation” procedure in the readme is out of date and causes an error.

Issue - State: closed - Opened by sig-ume 4 months ago - 4 comments

#1658 - fatal: could not get a repository handle for submodule 'sd-scripts'

Issue - State: closed - Opened by lunar-studio 4 months ago - 1 comment

#1657 - using '--alpha_mask' option seems not training text encoder

Issue - State: open - Opened by BbChip0103 4 months ago
Labels: bug

#1656 - --blocks_to_swap is supported on linux?

Issue - State: closed - Opened by al-swaiti 4 months ago - 4 comments

#1655 - ControlNet training for SDXL

Pull Request - State: closed - Opened by kohya-ss 4 months ago - 3 comments

#1653 - The new WSD scheduler seems to have an issue.

Issue - State: closed - Opened by gesen2egee 4 months ago

#1652 - ⚡Add Support for LoRA Training on Flux-Dev2Pro Models

Issue - State: closed - Opened by sashaok123 4 months ago - 3 comments

#1651 - "TypeError: get_cosine_schedule_with_warmup() got an unexpected keyword argument 'num_decay_steps'" Error occurs in dev branch

Issue - State: closed - Opened by tatamizu 4 months ago - 1 comment

#1649 - D:\AI\pinokio\api\fluxgym.git\env\lib\site-packages\torch\utils\checkpoint.py:1399: FutureWarning: `torch.cpu.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cpu', args...)` instead. [2024-09-28 15:15:24] [INFO] with device_autocast_ctx, torch.cpu.amp.autocast(cpu_autocast_kwargs), recompute_context: # type: ignore[attr-defined] [2024-09-28 15:21:07] [INFO] D:\AI\pinokio\api\fluxgym.git\env\lib\site-packages\torch\utils\checkpoint.py:295: FutureWarning: `torch.cpu.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cpu', args...)` instead. [2024-09-28 15:21:07] [INFO] with torch.enable_grad(), device_autocast_ctx, torch.cpu.amp.autocast(ctx.cpu_autocast_kwargs): # type: ignore[attr-defined]请问卡住不动怎么解决

Issue - State: closed - Opened by tangyuming99 4 months ago - 1 comment

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