The QLoRA fine-tuning process of sayama_ai_lora_flux_nf4 takes inspiration from this post (https://huggingface.co/blog/flux-qlora). The training was executed on a local computer with 1200 timesteps and the same parameters as the link mentioned above, which took around 8 hours on 8GB VRAM 4060. The peak VRAM usage was around 7.7GB. To avoid running low on VRAM, both transformers and text_encoder were quantized. The biggest challenge of training Japanese actresses is their photos used heavy filters to whiten and smoothen the skin. This practise severely distorts the training images which makes the result less convincing than Hollywood actresses. This training dataset contains a lot of face closeup which makes result more aligned with her actual face. The tradeoff is the overfitting problem of QLoRA which makes model more likely to ignore the prompt. All the images generated here are using the below parameters

Height: 512
Width: 512
Guidance scale: 5
Num inference steps: 20
Max sequence length: 512
Seed: 0

Usage

python

import torch
from diffusers import FluxPipeline, FluxTransformer2DModel
from transformers import T5EncoderModel

text_encoder_4bit = T5EncoderModel.from_pretrained(
    "hf-internal-testing/flux.1-dev-nf4-pkg", subfolder="text_encoder_2",torch_dtype=torch.float16,)

transformer_4bit = FluxTransformer2DModel.from_pretrained(
        "hf-internal-testing/flux.1-dev-nf4-pkg", subfolder="transformer",torch_dtype=torch.float16,)

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.float16,
                                    transformer=transformer_4bit,text_encoder_2=text_encoder_4bit)

pipe.load_lora_weights("je-suis-tm/sayama_ai_lora_flux_nf4",
                       weight_name='pytorch_lora_weights.safetensors')

prompt="Sayama Ai looking directly at the viewer, wearing a champagne-colored silk camisole, matching silk pajama pants, and a white silk robe, holding a white teacup and saucer in both hands at chest level, standing confidently in a luxurious hotel suite. The setting features a large window behind her, offering a blurred view of a city skyline with bokeh lights, a neutral-toned armchair to her right, and a side table with a patterned vase holding pink flowers and a black book. To her left, there's a floor lamp with a cream shade, a sofa with a cushion, and a brown curtain. Photorealistic, High-definition photography Soft ambient lighting combining interior warmth and subtle outdoor bluish light Medium full shot, eye-level Elegant, sophisticated, serene, warm tones, natural glow --ar 16:9 --v 6. 0 --s 750 --c 0"

image = pipe(
            prompt,
            height=512,
            width=512,
            guidance_scale=5,
            num_inference_steps=20,
            max_sequence_length=512,
            generator=torch.Generator("cpu").manual_seed(0),            
        ).images[0]

image.save("sayama_ai_lora_flux_nf4.png")

Trigger words

You should use Sayama Ai to trigger the image generation.

Download model

Download them in the Files & versions tab.