>>101910101
its flux, the kohya lora training branch is just merged with sd3
>>101910074
noice, we're SO BACK anon
>>101909112
sorry for the delay, here is the command prompt I used:
accelerate launch --mixed_precision bf16 --num_cpu_threads_per_process 1 "flux_train_network.py" --pretrained_model_name_or_path "C:/Users/B/Downloads/flux/flux1-dev.safetensors" --clip_l "C:/Users/B/Downloads/flux/clip_l.safetensors" --t5xxl "C:/Users/B/Downloads/flux/t5xxl_fp16.safetensors" --ae "C:/Users/B/Downloads/flux/VAE.safetensors" --save_model_as safetensors --sdpa --persistent_data_loader_workers --max_data_loader_n_workers 2 --seed 777 --gradient_checkpointing --mixed_precision bf16 --save_precision bf16 --network_module networks.lora_flux --network_dim 4 --network_args "train_blocks=single" --optimizer_type adafactor --optimizer_args "relative_step=False" "scale_parameter=False" "warmup_init=False" --learning_rate 1e-4 --network_train_unet_only --save_every_n_epochs 1 --output_dir "C:\Users\B\Downloads\ibuki satsuki lora" --output_name flux-lora-name --timestep_sampling sigmoid --model_prediction_type raw --guidance_scale 1.0 --loss_type l2 --train_data_dir "C:/Users/B/Downloads/ibuki satsuki lora-upscale_resized" --resolution "512,512" --cache_text_encoder_outputs --cache_text_encoder_outputs_to_disk --fp8_base --enable_bucket --bucket_no_upscale --split_mode --save_state --save_every_n_epochs=1 --max_train_steps=1600