Easily Create eye-catching images for it's_a_test
Just dream up what you want and we optimize to get great images at scale. it's_a_test is perfect for style images
accelerate launch "./sd-scripts/flux_train_network.py" ^
--pretrained_model_name_or_path "D:\kohya_ss\flux\unet\flux1DevFp8_v10.safetensors" ^
--clip_l "D:/stable-diffusion-webui-forge-main/models/text_encoder/flux/clip_l.safetensors" ^
--t5xxl "D:/stable-diffusion-webui-forge-main/models/text_encoder/flux/flux1T5TextEncoder_stock.safetensors" ^
--ae "D:/stable-diffusion-webui-forge-main/models/VAE/flux/flux1DevVAE_stock.safetensors" ^
--train_data_dir "D:\kohya_ss\flux\img" ^
--output_dir "D:/kohya_ss/flux/models/whimsical_flux5" ^
--output_name "whimsical_flux5" ^
--save_model_as safetensors ^
--sdpa ^
--persistent_data_loader_workers ^
--max_data_loader_n_workers 2 ^
--seed 777 ^
--gradient_checkpointing ^
--mixed_precision bf16 ^
--save_precision fp16 ^
--network_module networks.lora_flux ^
--network_dim 24 ^
--network_alpha 24 ^
--network_args train_blocks=single rank_dropout=0.1 module_dropout=0.1 ^
--optimizer_type lion ^
--lr_scheduler cosine ^
--learning_rate 3e-4 ^
--network_train_unet_only ^
--timestep_sampling sigmoid ^
--model_prediction_type raw ^
--guidance_scale 1.0 ^
--loss_type l2 ^
--resolution 768,768 ^
--cache_text_encoder_outputs ^
--cache_text_encoder_outputs_to_disk ^
--fp8_base ^
--enable_bucket ^
--bucket_no_upscale ^
--split_mode ^
--prior_loss_weight 1.0 ^
--caption_extension .txt ^
--keep_tokens 0 ^
--network_dropout 0.1 ^
--max_grad_norm 1.0 ^
--min_snr_gamma 10 ^
--save_every_n_epochs 1 ^
--train_batch_size "2" ^
--multires_noise_iterations 6 ^
--multires_noise_discount 0.3 ^
--optimizer_args weight_decay=0.4 betas="[0.95, 0.98]" ^
--max_train_epochs 30 ^
--enable_wildcard
See what else you can do with it's_a_test
Create hundreds of images for it's_a_test in minutes
Produce eye-catching images on it's_a_test with our autoscale system. You can create thousands of images on top of our cluster to get images incredibly fast. No more waiting around on your local machine for results.
Test variants and ideas quickly with our csv system
Have a general idea you want to test, but its not coming out exactly as you want, use our mass generator import system to generate many variations to find the perfect image you are looking for. You can write in excel, csv, or any other tool that exports CSV.
Check out other related models: style
Save time and money
Its a hassle getting all of this setup, with the click of a button start creating images instantly and effortlessy to find the perfect visualizations for your project.
Better looking images than anywhere else
Don't waste time playing around with settings and configurations. We have optimized and fine-tuned it's_a_test to get the best results.
Build a custom model on top of it's_a_test
Like the results you see with it's_a_test and want to build your own on top of it? No problem, we provide infrastructure to easily build on top of it. Just give us a few images and we will build a new model on top of it. (This feature is in private beta)
Test it's_a_test against hundreds of models with one click
Use our omni prompt tool and compare it's_a_test with any models you want. You just put in your prompts and select the models and it runs instantly.
Want to try a different image model? We have thousands...
Best for producing medieval, clothes, clothing, male, man, viking, norse, norwegian, swedish, danish, and icelandic images. Activation keywords: norsemenclothes, tunic, cloak, cape, satchel, belt, leather vambrace, and pants
Best for producing character, holly sardine, and stargazy images. Activation keywords: yaikagashi, holly, and fish
Best for producing anime, character, female, circus, and pomni images. Activation keywords: Pomni