Modelpublic

noema-art/ansemflux-klein

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

0likes7downloads

Model Card

ansemflux-klein

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

Trigger word: ansem Retrained onto FLUX.2 [klein] 4B from [ms2stationthis/ansemflux](https://huggingface.co/ms2stationthis/ansemflux) (FLUX.1-dev).

Sample Outputs

![sample](samples/sample_000.jpg)	![sample](samples/sample_001.jpg)
ansem, The image is a high-resolution photograph of a young ansem with a clean-shaven fac…	ansem, The image is a digitally manipulated photograph featuring a young ansem with a dar…
[image]	[image]
:---:	:---:
ansem, The image is a photograph featuring a young African American ansem with a medium c…	ansem, The image is a photograph capturing ansem seated in a modern, luxurious living spa…

Usage (Diffusers)

python

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/ansemflux-klein")
image = pipe("ansem, 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/` (14 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.

<sub>Published via noema.</sub>