<h3 id="anima-v1">Anima V1</h3>Trained base on Anima V0.1 base with LoRA_Easy_Training_Scripts. See about this version for training details.<h3 id="chenkin-rectified-flow-v1">Chenkin Rectified-Flow V1</h3>Trained base on Chenkin Rectified-Flow 0.3 with LoRA_Easy_Training_Scripts. See about this version for training details.<h3 id="noob-v-pred-v4">NOOB V-PRED V4</h3>Trained based on NOOB V-PRED 1.0 with<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> updated dataset and improved parameters.<h3 id="noob-v-pred-v3">NOOB V-PRED V3</h3>Trained based on NOOB V-PRED 1.0 with Kohya SS. I switched to locon for smaller file size and better results. Setting min SNR = 3 and alpha = 1 improved the effectiveness of training. However it leads to more limb distortation. But I think it's acceptable when achieving better results in style generation.See about the version for training details.<h3 id="noob-v-pred-v3">NOOB V-PRED V3</h3>Trained based on NOOB V-PRED 1.0 with Kohya SS. I switched to locon for smaller file size and better results. Setting min SNR = 3 and alpha = 1 improved the effectiveness of training. However it leads to more limb distortation. But I think it's acceptable when achieving better results in style generation.See about the version for training details.<h3 id="noob-v-pred-v2">NOOB V-PRED V2</h3>Trained based on NOOB V-PRED 1.0 with OneTrainer. Better in color and lora adapation. See about the version for training details.<h3 id="noob-v-pred-v1">NOOB V-PRED V1</h3>Trained based on NOOB V-PRED 0.65S. Recommend using it with V-PRED models.No trigger word set in this version.<h3 id="introduction">Introduction</h3>A LyCORIS to generate the style of AI artist mirham. Setting the artist name "mirham" as the trigger word during training, but not recommending its use, can cause over baking. Just use the trigger word when you feel necessary, eg. using together with other lora or being suppressed by other styles.<h3 id="training-details">Training details</h3>See about the version.

Mirham style

[[subsets]]name = "5"image_dir = "G:/dataset/style/mirham/5_mirham"num_repeats = 5shuffle_caption = truecaption_extension = ".txt"random_crop_padding_percent = 0.05caption_dropout_rate = 0.1caption_tag_dropout_rate = 0.1[train_mode]train_mode = "lora"[general_args.args]persistent_data_loader_workers = truevae_batch_size = 5pretrained_model_name_or_path = ""mixed_precision = "bf16"gradient_checkpointing = truegradient_accumulation_steps = 1seed = 42max_data_loader_n_workers = 1max_token_length = 225prior_loss_weight = 1.0sdpa = truemax_train_epochs = 25cache_latents = truecache_latents_to_disk = true[general_args.dataset_args]resolution = 1024batch_size = 1[network_args.args]network_dim = 64network_alpha = 1.0min_timestep = 0max_timestep = 1000network_train_unet_only = true[optimizer_args.args]optimizer_type = "ProdigyPlusScheduleFree"lr_scheduler = "constant"loss_type = "l2"learning_rate = 1.0unet_lr = 1.0max_grad_norm = 1.0min_snr_gamma = 1.0[saving_args.args]output_dir = "G:/LoRA_Easy_Training_Scripts/output"output_name = "mirham-anima1.0-V1-locon-dim64conv16alpha0.01-SNR1"save_precision = "bf16"save_model_as = "safetensors"save_every_n_epochs = 1save_toml = truesave_toml_location = "G:/LoRA_Easy_Training_Scripts/output"[sample_args.args]sample_sampler = "euler"sample_every_n_epochs = 1sample_prompts = "G:/LoRA_Easy_Training_Scripts/anima sample.txt"[logging_args.args]log_prefix_mode = "disabled"run_name_mode = "default"[anima_args.args]pretrained_model_name_or_path = "G:/sd-webui-forge-neo/models/Stable-diffusion/anima_baseV10.safetensors"qwen3 = "G:/sd-webui-forge-neo/models/text_encoder/qwen_3_06b_base.safetensors"vae = "G:/sd-webui-forge-neo/models/VAE/qwen_image_vae.safetensors"qwen3_max_token_length = 512t5_max_token_length = 512timestep_sampling = "sigmoid"sigmoid_scale = 1.0discrete_flow_shift = 3.0[edm_loss_args.args]edm2_loss_weighting = false[bucket_args.dataset_args]enable_bucket = truemin_bucket_reso = 256max_bucket_reso = 2048bucket_reso_steps = 64[network_<a target="_blank" rel="ugc" href="http://args.args.network">args.args.network</a>_args]conv_dim = 16conv_alpha = 1.0train_llm_adapter = "False"[optimizer_args.args.optimizer_args]betas = "0.9,0.99"beta3 = "None"weight_decay = "0"weight_decay_by_lr = "True"d0 = "1e-6"d_coef = "2"d_limiter = "True"prodigy_steps = "0"schedulefree_c = "0"eps = "1e-8"split_groups = "True"split_groups_mean = "False"factored = "True"factored_fp32 = "True"use_bias_correction = "True"use_stableadamw = "True"use_schedulefree = "True"use_speed = "False"stochastic_rounding = "True"fused_back_pass = "False"use_cautious = "False"use_grams = "False"use_adopt = "False"use_orthograd = "False"use_focus = "False"

Files with SHA256: 5a348e00090d1d379ccea8e276394ff658cbe74ad1f02905ff7181439f16024c