fwwrsd/9879yhu-zimage-lora

Z Image LoRA — 9879yhu

A personalized identity LoRA trained on Z Image using diffusion-pipe + DeepSpeed.

Trigger Word

Use 9879yhu in your prompt to activate this LoRA.

Quick Start

Direct Download (for ComfyUI / wget / curl)

https://huggingface.co/fwwrsd/9879yhu-zimage-lora/resolve/main/9879yhu_lora.safetensors

ComfyUI

1. Download 9879yhu_lora.safetensors → place in ComfyUI/models/loras/
2. Add a LoRA Loader node, select 9879yhu_lora.safetensors
3. Include 9879yhu in your prompt

Command Line

# huggingface-cli (recommended)
huggingface-cli download fwwrsd/9879yhu-zimage-lora 9879yhu_lora.safetensors

# wget
wget "https://huggingface.co/fwwrsd/9879yhu-zimage-lora/resolve/main/9879yhu_lora.safetensors" -O 9879yhu_lora.safetensors

Recommended Settings

Parameter	Value
LoRA Strength	0.7 — 1.0
CFG Scale	3.5
Steps	20 — 30

Available Checkpoints

This repo contains 2 epoch checkpoints — try different epochs to find the best balance of identity fidelity vs flexibility.

Epoch	File	Note	Link
10	9879yhu_epoch10.safetensors		Download
17	9879yhu_epoch17.safetensors	⭐ Best	Download

9879yhu_lora.safetensors is a copy of the best (final) checkpoint — epoch 17.

Download Specific Epoch

# Download epoch 17 (best)
wget https://huggingface.co/fwwrsd/9879yhu-zimage-lora/resolve/main/9879yhu_epoch17.safetensors

# Download any epoch — replace N with epoch number
huggingface-cli download fwwrsd/9879yhu-zimage-lora 9879yhu_epoch<N>.safetensors

Training Details

Parameter	Value
Base Model	Comfy-Org/z_image
Training Method	diffusion-pipe + DeepSpeed
LoRA Rank	16
Learning Rate	1e-4
Total Epochs	unknown (best: 17)
Resolution	1024px
Dataset Size	58 images
Preset	test
Created	2026-03-05

License

Apache 2.0 — free for personal and commercial use.

---

Trained with NanoBanana LoRA Bot on RunPod