Downloads · 30 days
0
millie-v2-krea2-lora
Zaytron40k/millie-v2-krea2-lora
millie-v2-krea2-lora is a text-to-image model from Zaytron40k. Use it when you need an image from a text prompt.
Krea 2 character LoRA — trigger woman. Trained on Krea 2 RAW, intended for inference on Krea 2 Turbo.
Access
Public
Updated Sep 14, 2026
Repo size
2.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors2.1 GB · 100%
At a glance
- Task
- Text-to-Image
- Access
- Public
- Created
- Sep 14, 2026
- Updated
- Sep 14, 2026
- SHA
- a858fe94
Base models
Model card
From the Hugging Face model README
millie-v2-krea2-lora
Krea 2 character LoRA — trigger woman. Trained on Krea 2 RAW, intended for inference on Krea 2 Turbo.
Status: ✅ training complete
Training configuration
| Field | Value |
|---|---|
| Base model (train) | krea/Krea-2-Raw (raw.safetensors, 12B DiT, undistilled) |
| Inference target | krea/Krea-2-Turbo (8-step distilled) — RAW-train / Turbo-infer |
| VAE | Qwen-Image VAE (qwen_image_vae.safetensors) |
| Text encoder | Qwen3-VL-4B (qwen3vl_4b_bf16.safetensors) |
| Trainer | Musubi Tuner (kohya-ss), networks.lora_krea2 |
| LoRA | rank/dim 24, alpha 24, all Linear layers of the DiT |
| Optimizer | adamw8bit, lr 1e-4 |
| Precision | bf16, gradient checkpointing, SDPA |
| Timestep sampling | krea2_shift (resolution-aware), weighting_scheme none |
| Trigger | woman (no dedicated trigger token — class word only; identity learned from the dataset (young woman)) |
| Dataset | 61 Seedream Western-animation seductive portrait+medium images + captions; young woman (Stepsis Summer #558), no trigger (class word woman). batch 1, 1024 buckets, num_repeats 2 |
| Schedule | 12 epochs = 1464 steps, save every 2 epochs, seed 42 |
| Hardware | 1x NVIDIA H100 80GB (RunPod, EU-NL-1, Musubi volume) |
Loss per checkpoint
avr_loss is a rolling average — flow-matching loss is near-flat, so use it as a sanity check and pick the best checkpoint visually.
| checkpoint | epoch | step | avr_loss |
|---|---|---|---|
-000002 | 2 | 244 | 0.0543 |
-000004 | 4 | 488 | 0.0488 |
-000006 | 6 | 732 | 0.0481 |
-000008 | 8 | 976 | 0.0502 |
-000010 | 10 | 1220 | 0.0474 |
| final | 12 | 1464 | 0.0444 |
min avr_loss 0.0356 · final 0.0444
Inference (Krea 2 Turbo, Musubi Tuner)
Prepend the trigger woman to the prompt. Recommended: stack over the house style LoRA.
python src/musubi_tuner/krea2_generate_image.py \
"woman, <scene>" \
--dit turbo.safetensors --vae qwen_image_vae.safetensors \
--text_encoder qwen3vl_4b_bf16.safetensors \
--steps 8 --guidance_scale 1 --mu 1.15 --width 1024 --height 1280 \
--attn_mode torch --lora_weight <checkpoint>.safetensors --lora_multiplier 1.0
Dataset samples
Representative image+caption pairs from the training set:
Millie_SSummer558-portrait-001-seedream.jpg
woman, portrait, facing the camera head-on with head slightly tilted, hands out of frame, wearing a white long-sleeve top with a deep V neckline gathered by a front tie, lips slightly parted in a soft sultry expression, plain white background.
Millie_SSummer558-medium-001-seedream.jpg
woman, medium shot, standing facing the camera with both hands resting on her hips, wearing a white long-sleeve tie-front crop top over a beige high-waisted mini skirt with a side slit, chin lifted slightly, a calm sultry expression with lips parted, plain lavender background.
Auto-generated by training watchdog.
- Task
- Text-to-Image
- Created
- Sep 14, 2026
- Updated
- Sep 14, 2026