bghira/SimpleTuner

Issues

• No matter how to modify the parameters, it won't work on the 24G RTX 4090(CUDA out of memory)

  #996 opened 5 days ago by haoli-zbdbc
  4

• train batch size params in config.json does not work

  #993 opened 6 days ago by LianShuaiLong
  4

• 'Trainer' object has no attribute 'optimizer'

  #990 opened 7 days ago by specblades
  2

• Can this be used on windows 11?

  #989 opened 7 days ago by Vigilence
  1

• error: argument --resolution

  #988 opened 8 days ago by specblades
  1

• Are you a Trump supporter?

  #985 opened 8 days ago by hablaba
  1

• Cannot resume training using optimi-stableadamw

  #983 opened 8 days ago by riffmaster-2001
  1

• Training crashing for images that used to work fine, with error: latent shape mismatch: torch.Size([16, 88, 136]) != torch.Size([16, 88, 144])

  #977 opened 16 days ago by a-l-e-x-d-s-9
  15

• Parquet questions

  #976 opened 15 days ago by LianShuaiLong
  1

• flux full training on multi gpu validation is block with deepspeed zero3

  #965 opened 20 days ago by YLongJin
  0

• Using num_train_epochs without max_train_steps: Division by zero

  #960 opened 17 days ago by missEldgard
  4

• Flux LoRA Training Setting

  #964 opened 17 days ago by matabear-wyx
  4

• accelerate muti gpu with gradient_checkpointing throws an error

  #972 opened 17 days ago by benihime91
  6

• Training time per iteration scales linearly with batch size

  #970 opened 18 days ago by primecai
  1

• dependencies mess if using pip

  #966 opened 19 days ago by GXcells
  1

• Cannot start Deepspeed Full Finetune: Trainer object has no attribute 'optimizer'

  #962 opened 22 days ago by playerzer0x
  0

• Error when using dataloader_prefetch

  #961 opened 22 days ago by a-l-e-x-d-s-9
  0

• Invalid model family specified: sd3

  #953 opened 22 days ago by piotr-sikora-v
  0

• Do your repo support flux training on target specific layers?

  #958 opened 22 days ago by ngsitrong26
  0

• lr_end does not affect Sine schecule

  #954 opened 23 days ago by zhuliyi0
  7

• full finetuning FLUX

  #852 opened a month ago by youngwanLEE
  17

• Generating aspect bucket cache is stuck, but debug.log is still updating

  #952 opened 24 days ago by LianShuaiLong
  1

• generic webhook support

  #914 opened 24 days ago by bghira
  0

• allow aborting training job via sdk

  #946 opened 24 days ago by bghira
  0

• value should be one of int, float, str, bool, or torch.Tensor

  #940 opened a month ago by piotr-sikora-v
  4

• why num_workers=0 for default

  #947 opened a month ago by LianShuaiLong
  3

• Training LoRA on top of Civitai-style Flux checkpoint

  #945 opened a month ago by PCanavelli
  1

• Macbook M3 for tuning? (transformers_flux and then _scaled_dot_product_attention_math_for_mps issues)

  #937 opened a month ago by defertoexpertise
  1

• remove login to hf in configure.py

  #933 opened a month ago by zhuliyi0
  5

• background / external API validations

  #917 opened a month ago by bghira
  0

• async/background model publishing

  #916 opened a month ago by bghira
  0

• s3 model publishing hook

  #915 opened a month ago by bghira
  0

• Bug in Flux with Masked Attention

  #904 opened a month ago by zhuole1025
  2

• SDXL Loras do not load in ComfyUI, A1111

  #898 opened a month ago by PsychoLogicAu
  6

• Please readd support for Prodigy

  #897 opened a month ago by axel578
  4

• what does it mean "Using a value of 1.0 seems to preserve the CFG distillation for the Dev model" in args description

  #891 opened a month ago by rardz
  0

• ValueError: Invalid LoRA checkpoint.

  #883 opened a month ago by DoiryCool
  1

• Empty all_caption_files.json with textfile caption strategy

  #885 opened a month ago by rengotj
  2

• AttributeError: 'CLIPTextModelWithProjection' object has no attribute 'peft_config'

  #884 opened a month ago by uiyo
  1

• Flux SimpleTuner Lora ,comfyui load error "IndexError: list index out of range"

  #875 opened a month ago by aoyang-hd
  1

• If you use attention masking and lycoris, any config with `FluxTransformer2DModel` will not be seen because the class used is `FluxTransformer2DModelWithMasking`

  #874 opened a month ago by AmericanPresidentJimmyCarter
  2

• missing save code for Flux full finetune

  #864 opened a month ago by Jack000
  2

• please offer a webui

  #868 opened a month ago by taozhiyuai
  0

• If the VAE runs out of memory while caching, training continues onward but then crashes when it can't find the missing cache file

  #859 opened a month ago by AmericanPresidentJimmyCarter
  4

• should I ignore this issue?

  #866 opened a month ago by taozhiyuai
  1

• Need help with the training process

  #860 opened a month ago by rohanNkhaire
  1

• Training went wrong, why not start training

  #855 opened a month ago by aoyang-hd
  2

• SDXL training cannot continue from latest

  #851 opened a month ago by zhuliyi0
  8

• Weird precision error

  #854 opened a month ago by tristanwqy
  1

• the guid has many result(pseudo-camera-10k)

  #849 opened a month ago by al-swaiti
  2