Qwen-Image-Edit-2511-Lightning-4steps-V1.0-FP32
更新：2025-12-24 18:51｜发布时间：2025-12-24 18:51
8.5k
0
8
Gugugaga
类型
LoRA
基础模型
Qwen-Edit
发布时间
2025-12-24 18:51
文件签名
02c56cdf06cbfd18cddb8190aed010d1df05c2b09fe00dadfa1948a9d69cb2d2
Qwen-Image-Edit-2511-Lightning-4steps-V1.0-FP32.safetensors
1.58 GB
Qwen-Image-Edit-2511-Lightning-4steps-V1.0-fp32
Model Name
Qwen-Image-Edit-2511-Lightning
Model Overview
Qwen-Image-Edit-2511-Lightning is a collection of models specifically optimized for image editing tasks. Through step distillation and quantization techniques, it aims to deliver high-efficiency inference performance. This repository contains three core model files with different characteristics:
| Model File Name | Type | Key Features |
|---|---|---|
| Qwen-Image-Edit-2511-Lightning-4steps-V1.0-bf16.safetensors | 4-step distilled LoRA | BF16 precision, lightweight, 4-step inference |
| Qwen-Image-Edit-2511-Lightning-4steps-V1.0-fp32.safetensors | 4-step distilled LoRA | FP32 precision, high precision, 4-step inference |
| qwen_image_edit_2511_fp8_e4m3fn_scaled_lightning.safetensors | FP8 Quantization | FP8 (e4m3fn scaled) precision, merged with 4-step distilled LoRA, optimized for low-memory deployment Usage Instructions This model series supports two mainstream frameworks. Detailed guides are as follows: |
- Qwen-Image-Lightning Framework: For complete documentation on using the model within the Qwen-Image-Lightning ecosystem (including environment setup, inference workflows, and customization methods),
- LightX2V Framework: This model is fully compatible with the LightX2V lightweight video/image generation inference framework. For step-by-step usage examples, configuration templates, and performance optimization tips,
- Core Optimizations
- Step Distillation: The LoRA models drastically reduce original inference steps to just 4 steps, achieving significant speedup (~10x faster than standard 40-step inference) while maintaining image editing quality.
- FP8 Quantization: The quantized base model strikes a balance between performance and resource efficiency, reducing GPU memory usage by ~50% compared to the FP32 model while maintaining editing fidelity.
Reprinted from: Qwen-Image-Edit-2511-Lightning-4steps-V1.0-FP32
作品
暂无作品
