Type | |
Stats | 353 833 29 |
Reviews | (64) |
Published | Jan 27, 2024 |
Base Model | |
Training | Steps: 5,500 Epochs: 12 |
Usage Tips | Clip Skip: 1 Strength: 0.6 |
Trigger Words | Yuzuriha \(Jigokuraku\) 1girl japanese clothes purple hair pink eyes ninja ponytail no bra |
Training Images | Download |
Hash | AutoV2 7F218A3851 |
I would like to hear from other Lora creators and chat a bit if you have the time..
Also please leave a review if you used this, good or bad i'd like to know if it worked for you or if you had any issues so feedback is really important to me, thanks!
This is my first Lora so it might not be the best but i've tested it on multiple anime XL models such as
animeChangefulXL_v10ReleasedCandidate
I found that i got the best results with the lora weighted from .4 to .6 and if i wanted her normal outfit to show 0.5 to 0.7 seemed good and if you use japanese clothes you could even go lower then that.. I used Kohya_SS Gui to train this lora and since I didn't know exactly which settings would be good for a character lora i went with one of the XL presets and I'll post that below as well as uploading the exact dataset i used with the caption files included so maybe someone who has much more experience then i do can improve this lora if they feel like it.
{"ss_training_finished_at": "1705714270.261762",
"ss_base_model_version": "sdxl_base_v1-0",
"ss_network_dim": "Dynamic",
"ss_sd_model_name": "stabilityai/stable-diffusion-xl-base-1.0",
"ss_network_dropout": "None",
"ss_reg_dataset_dirs": "{}",
"ss_caption_tag_dropout_rate": "0.0",
"ss_prior_loss_weight": "1.0",
"ss_lowram": "False",
"ss_adaptive_noise_scale": "None",
"ss_mixed_precision": "bf16",
"ss_color_aug": "False",
"ss_num_epochs": "25",
"ss_clip_skip": "None",
"ss_epoch": "4",
"ss_ip_noise_gamma": "None",
"ss_total_batch_size": "6",
"ss_max_bucket_reso": "2048",
"ss_output_name": "yuzuriha_\\(hell's_paradise\\)",
"ss_bucket_no_upscale": "True",
"ss_unet_lr": "1.0",
"ss_sd_scripts_commit_hash": "7dd2a273b225fb33f8a8769b1af77517dd95d3a9",
"ss_max_grad_norm": "1.0",
"ss_scale_weight_norms": "None",
"ss_full_fp16": "False",
"ss_seed": "1337",
"ss_lr_warmup_steps": "533",
"ss_cache_latents": "True",
"ss_noise_offset": "0.0357",
"ss_steps": "1750",
"ss_multires_noise_iterations": "None",
"ss_max_train_steps": "10667",
"ss_min_snr_gamma": "5.0",
"ss_debiased_estimation": "False",
"ss_multires_noise_discount": "0.3",
"ss_caption_dropout_rate": "0.0",
"ss_face_crop_aug_range": "None",
"ss_caption_dropout_every_n_epochs": "0",
"ss_training_started_at": "1705698819.5353682",
"ss_flip_aug": "False",
"ss_max_token_length": "225",
"ss_training_comment": "Dynamic resize with sv_fro: 0.9 from 256; None",
"ss_num_batches_per_epoch": "427",
"ss_random_crop": "False",
"ss_shuffle_caption": "False",
"ss_num_train_images": "2560",
"ss_batch_size_per_device": "6",
"ss_session_id": "3385709643",
"ss_enable_bucket": "True",
"ss_optimizer": "dadaptation.dadapt_adam.DAdaptAdam(decouple=True,weight_decay=0.2,betas=(0.9, 0.99),growth_rate=1.02)",
"ss_min_bucket_reso": "256",
"ss_gradient_checkpointing": "True",
"ss_network_alpha": "Dynamic",
"ss_learning_rate": "1.0",
"ss_bucket_info": "{\"buckets\": {\"0\": {\"resolution\": [1024, 1024], \"count\": 2560}}, \"mean_img_ar_error\": 0.0}",
"ss_text_encoder_lr": "1.0",
"ss_keep_tokens": "0",
"ss_resolution": "(1024, 1024)",
"ss_v2": "False",
"ss_gradient_accumulation_steps": "1",
"ss_lr_scheduler": "cosine",
"ss_num_reg_images": "0",
"ss_zero_terminal_snr": "False",
"ss_network_module": "networks.lora"
}
One thing i noticed though was the size of the lora was 1.7gb which is way to big for a character lora so i used kohya_ss gui's lora tools to change the 256 dim into a 32 dim, tested before and after and both seemed the same which resulted in a 170mb instead of almost 2gb