aespaflux-klein

FLUX.2 [klein] 4B conversion of ms2stationthis/aespaflux (originally FLUX.1-dev). Trigger: "aespa".

Trigger word: aespa Retrained onto FLUX.2 [klein] 4B from ms2stationthis/aespaflux (FLUX.1-dev).

Sample Outputs

aespa, The image is a high-resolution photograph featuring a young woman with a striking,…	aespa, This is a high-resolution photograph featuring a young Asian woman with a striking…
:---:	:---:
aespa, This is a high-resolution photograph featuring a young East Asian woman with a fai…	aespa, This is a high-resolution photograph featuring a young woman with an ethereal, fut…

Usage (Diffusers)

import torch
from diffusers import Flux2Pipeline

pipe = Flux2Pipeline.from_pretrained(
    "black-forest-labs/FLUX.2-klein-4B", torch_dtype=torch.bfloat16
).to("cuda")
pipe.load_lora_weights("ms2stationthis/aespaflux-klein")
image = pipe("aespa, a character portrait", guidance_scale=4.0, num_inference_steps=25).images[0]
image.save("out.png")

Recommended Settings

LoRA strength	Guidance	Steps	Resolution
0.8–1.0	4.0	25	1024×1024

Training Details

• Base: black-forest-labs/FLUX.2-klein-base-4B
• Steps: 4000 · Network: LoRA rank 32 / alpha 32
• Optimizer: adamw8bit, lr 1e-4 · Scheduler: flowmatch
• Resolution: 512, 768, 1024 (multi-res bucketed) · Precision: bf16 train / fp16 save

Reproduction

This repo includes the full dataset/ (20 image-caption pairs) and the exact config.yaml so the LoRA can be retrained as-is.

About

Trained on StationThis — an AI creative platform powered by $MS2. Train your own LoRAs via @stationthisbot on Telegram.

_{Published via noema.}