Views
No views yet
| 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 | 22 & 2850 |
| Epoch | 20 | Save Every N Epochs | 1 |
Labeling: florence2-en(natural language & English)
Total Images Used for Training : 30 [HQ Images]| Source | Link |
|---|---|
| Playground | playground.com |
| ArtStation | artstation.com |
| 4K Wallpapers | 4kwallpapers.com |
| Dimensions | Aspect Ratio | Recommendation |
|---|---|---|
| 1472 x 1140 | 4:3 (approx.) | Best |
| 1024 x 1024 | 1:1 | Default |
1import torch
2from diffusers import DiffusionPipeline
3
4base_model = "Qwen/Qwen-Image"
5pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)
6
7lora_repo = "prithivMLmods/Qwen-Image-Sketch-Smudge"
8trigger_word = "Sketch Smudge"
9pipe.load_lora_weights(lora_repo)
10
11device = torch.device("cuda")
12pipe.to(device)Sketch Smudge to trigger the image generation.