---
title: Flux-Lego-Ref-LoRA
canonical_url: "https://www.modelscope.cn/models/prithivMLmods/Flux-Lego-Ref-LoRA"
md_url: "https://www.modelscope.cn/models/prithivMLmods/Flux-Lego-Ref-LoRA.md"
repository: prithivMLmods/Flux-Lego-Ref-LoRA
last_updated: 2024-11-30
license: creativeml-openrail-m
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
base_model:
  - black-forest-labs/FLUX.1-dev
base_model_relation: adapter
parameters: 306.3M
tensor_type:
  - F16
library_name:
  - pytorch
  - lora
  - diffusers
  - safetensors
frameworks:
  - pytorch
downloads: 1658
stars: 0
tags:
  - text-to-image
  - lora
  - diffusers
  - "template:diffusion-lora"
---

# Flux-Lego-Ref-LoRA

> Flux-Lego-Ref-LoRA - prithivMLmods 在 ModelScope 开源的模型。The model is still in the training phase. This is not the final version and may contain artifacts and perform poorly in some cases.

prithivMLmods/Flux-Lego-Ref-LoRA 是 ModelScope 魔搭社区上的 306.3M 参数text-to-image-synthesis模型，采用 creativeml-openrail-m 许可，基于 black-forest-labs/FLUX.1-dev 构建。

- **Repository**: prithivMLmods/Flux-Lego-Ref-LoRA
- **License**: creativeml-openrail-m
- **Tasks**: text-to-image-synthesis
- **Parameters**: 306.3M
- **Base model**: black-forest-labs/FLUX.1-dev
- **Tags**: text-to-image, lora, diffusers, template:diffusion-lora
- **Downloads**: 1658
- **Stars**: 0
- **Last updated**: 2024-11-30

Source: https://www.modelscope.cn/models/prithivMLmods/Flux-Lego-Ref-LoRA

---

# Flux-Lego-Ref-LoRA

<Gallery />

**The model is still in the training phase. This is not the final version and may contain artifacts and perform poorly in some cases.**

## Model description 

**prithivMLmods/Flux-Lego-Ref-LoRA**

Image Processing Parameters 

| Parameter                 | Value  | Parameter                 | Value  |
|---------------------------|--------|---------------------------|--------|
| LR Scheduler              | constant | Noise Offset              | 0.03   |
| Optimizer                 | AdamW  | Multires Noise Discount   | 0.1    |
| Network Dim               | 64     | Multires Noise Iterations | 10     |
| Network Alpha             | 32     | Repeat & Steps           | 17 & 2350 |
| Epoch                     | 10   | Save Every N Epochs       | 1     |

    Labeling: florence2-en(natural language & English)
    
    Total Images Used for Training : 15

## Best Dimensions & Inference

| **Dimensions** | **Aspect Ratio** | **Recommendation**       |
|-----------------|------------------|---------------------------|
| 1280 x 832      | 3:2              | Best                     |
| 1024 x 1024     | 1:1              | Default                  |

### Inference Range

- **Recommended Inference Steps:** 30–35
    
## Setting Up
```python
import torch
from pipelines import DiffusionPipeline

base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "prithivMLmods/Flux-Lego-Ref-LoRA"
trigger_word = "lego --fref --89890"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)
```
## Trigger words

You should use `lego --fref --89890` to trigger the image generation.

## Download model

Weights for this model are available in Safetensors format.

[Download](/prithivMLmods/Flux-Lego-Ref-LoRA/tree/main) them in the Files & versions tab.