Edit model card

Recent Updates

24/10/01 (Check Point) myjc_flux_v2 Finetune Test 24/09/26 (LoRA) myjc_flux_lora_v2-test


myjk flux

-trained 2852+1316 images.
-The trigger doesn't seem valid...


Download: myjk_flux_lora_v1 (LoRA)
Download: myjk_flux-Q5_K_M.gguf (checkpoint)
Download: version b (+109 images use adamwschedulefree optimizer)

Recommended:

The LoRA used for the test is Flux Fusion DS v0 GGUF Q5_K_M.
VAE / Text Encoder: ae, clip_l, t5-v1_1-xxl-encoder-Q5_K_M

sample1
sample1
sample1
-refer to png info

- sample prompt

japanese, 18yo, myjk, smile,
photograph of Two girls in idol costumes singing. The girl on the left has black ponytail hair and a guitar. The girl on the right has long black hair and a microphone. The stage at night is illuminated with lights and neon “myjk” signage.
Steps: 12, Sampler: Euler, Schedule type: Simple, CFG scale: 1, Seed: 3220485898, Size: 768x1024, Model hash: 33c0966fb8, Model: myjk_flux-Q5_K_M, Denoising strength: 0.3, Hires CFG Scale: 1, Hires upscale: 2, Hires upscaler: 4x-UltraSharp, Version: f2.0.1v1.10.1-previous-535-gb20cb4bf0, Diffusion in Low Bits: Automatic (fp16 LoRA), Module 1: ae, Module 2: t5-v1_1-xxl-encoder-Q5_K_M, Module 3: clip_l

- trigger

myjk, japanese, european, 
and 16-18 yo, 
and native english(recomended) or danbooru tags

myjc flux

-trained 1543+1309 images.
-The trigger doesn't seem valid...


v2

Download: myjc_flux_lora_v2-test (LoRA) #Flux-Dev2_Pro training Test If the image is blurred, increase Sampling steps

Download: myjc_flux_v2_FTtest-Q5_K_M (Check Point) #Flux dev1 Finetune + LoRA Test

v1
Download: myjc_flux_lora_v1 (LoRA)
Download: myjc_flux-Q5_K_M.gguf (checkpoint)

Recommended:

The LoRA used for the test is Flux Fusion DS v0 GGUF Q4_0 (UNET) and v0 GGUF Q5_K_M.
VAE / Text Encoder: ae, clip_l, t5-v1_1-xxl-encoder-Q5_K_M

sample1
sample1
sample1
-refer to png info

- sample prompt

14yo, myjc, japanese, medium breasts,
This photograph captures a girl sitting on a grassy field at night. She has a light complexion and straight long black hair with bangs styled with a black bow. Her expression is cheerful with a slight smile. She is wearing a loose oversized shirt in a pastel gradient of pink yellow and blue which is slightly oversized giving it a cozy casual look. Her shirt is paired with white shorts and knee-high black socks with a small white bow on the top. The socks are adorned with a subtle pattern. She sits on a blanket with a white background featuring small amo,e characters. The grass is lush and green indicating a well-maintained lawn. The background is dark suggesting it is nighttime and the lighting is soft creating a warm and intimate atmosphere. The overall mood of the image is relaxed and playful with the subject's youthful and cheerful demeanor complementing the serene outdoor setting.
Steps: 12, Sampler: Euler, Schedule type: Simple, CFG scale: 1, Seed: 1298706659, Size: 768x1024, Model hash: c6b19f170d, Model: myjc_flux-Q5_K_M, Denoising strength: 0.3, Hires upscale: 2, Hires upscaler: 4x-UltraSharp, Version: f2.0.1v1.10.1-previous-501-g668e87f92, Diffusion in Low Bits: Automatic (fp16 LoRA), Module 1: ae, Module 2: clip_l, Module 3: t5-v1_1-xxl-encoder-Q5_K_M

- trigger

myjc, japanese, european, 
and 13-15 yo, 
and native english(recomended) or danbooru tags


lora_zipang_flux_test

-Training was based on a merged model of dev1 and lora test**.

-Trigger

japanese, european

myjc, 13yo
myjsh, 12yo

myjsm_flux_test02

-It is a test lora of poor quality with only a few images learned.
-trained 273 images.

Found a slightly better training setting. But still hard to find things that don't show up in flux.

Download:test02

The model used for the test is Flux Fusion DS v0 GGUF Q4_0 (UNET) and v0 GGUF Q5_K_M.

GGUF Q4_0 + t5xxl_fp8_e4m3fn : 4step
sample1
sample1
sample1
GGUF Q5_K_M. + t5-v1_1-xxl-encoder-Q5_K_M : 12step
sample1
sample1
sample1
-refer to png info

- sample prompt

9yo, myjsm, japanese,
photograph of a girl sitting on a brick pavement with a pink umbrella in front of her. She is wearing a white camisole and a blue skirt with a anime print. She has shoulder-length dark hair and is smiling at the camera.
bangs, black eyes, skirt, rain
<lora:myjsm_flux_test02:1>
Steps: 4, Sampler: Euler, Schedule type: Simple, CFG scale: 1, Seed: 642879771, Size: 792x1056, Model hash: 5e21feb505, Model: FusionDS_v0_Q4, Lora hashes: "myjsm_flux_test02: 3fdff20b7d65", Version: f2.0.1v1.10.1-previous-419-gf82029c5c, Module 1: ae, Module 2: clip_l, Module 3: t5xxl_fp8_e4m3fn

- trigger

myjsm, japanese, 9yo, 
and native english

-Train Settings

base model: flux1-dev.safetensors
vae/text encoder: clip_l.safetensors, t5xxl_fp8_e4m3fn.safetensors, ae.safetensors
tag: caption (native eng) + tags (danbooru)

--network_module "networks.lora_flux"
--gradient_checkpointing
--cache_latents
--cache_latents_to_disk
--cache_text_encoder_outputs
--cache_text_encoder_outputs_to_disk
--enable_bucket
--bucket_no_upscale
--optimizer_type "AdamW8bit"
--optimizer_args "weight_decay=0.01" "betas=0.9,0.999"
--learning_rate=0.0002
--network_dim=32
--network_alpha=4
--network_train_unet_only
--mixed_precision "bf16"
--save_precision "bf16"
--full_bf16
--loss_type "l2"
--huber_schedule "snr"
--model_prediction_type "raw"
--discrete_flow_shift 3
--timestep_sampling "sigma"
--max_grad_norm=1
--max_timestep=1000
--min_snr_gamma=5
--min_timestep=100
--noise_offset=0.0375
--adaptive_noise_scale=0.00375
--apply_t5_attn_mask
--split_mode
--network_args "loraplus_unet_lr_ratio=16" "train_blocks=single"

Downloads last month
332
GGUF
Model size
11.9B params
Architecture
flux

5-bit

Inference Examples
Unable to determine this model's library. Check the docs .