---
tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- text: 'lego --fref --89890, a Lego Star Wars figure stands in a pile of snow. The figure is adorned with a white snow-covered hat, a white coat, and a black and white striped jacket. The face of the figure is yellow, and the eyes are blue. The background is a grayish-gray color, and there are small white dots dotting the top of the image.'
  output:
    url: images/1.png
- text: 'lego --fref --89890, a toy robot is soaring through the air. The robot is dressed in a red jumpsuit, adorned with a black helmet and goggles. The toy is holding a small camera in its right hand. The camera has a black lens, and a yellow handle. The sky is a deep blue, dotted with white clouds, adding a pop of color to the scene.'
  output:
    url: images/2.png
- text: 'lego --fref --89890, a lego robot is soaring through the air on a dirt bike. The robot is wearing a white helmet, a white shirt, and a brown and orange vest. The bike is soaring over a pile of brown rocks, with a body of water in the background. The sky is filled with clouds, and the sun is peeking through the clouds, adding a warm glow to the scene.'
  output:
    url: images/3.png
- text: 'lego --fref --89890, a lego astronaut is standing on a rocky surface, holding a flag with a white pole. The astronaut is wearing a white helmet, a white and red uniform, and black gloves. The background is blurred, creating a stark contrast with the astronaut and the flag. The flag is adorned with an orange and white striped flag, adding a pop of color to the scene.'
  output:
    url: images/4.png
base_model: black-forest-labs/FLUX.1-dev
instance_prompt: lego --fref --89890
license: creativeml-openrail-m
---
# 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.