prithivMLmods
commited on
Commit
•
9e2efda
1
Parent(s):
56686b8
Update README.md
Browse files
README.md
CHANGED
@@ -26,6 +26,26 @@ license: creativeml-openrail-m
|
|
26 |
|
27 |
<Gallery />
|
28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
|
30 |
## Trigger words
|
31 |
|
@@ -51,6 +71,23 @@ Here’s a table format for the Hugging Face model **"prithivMLmods/Canopus-LoRA
|
|
51 |
| **License** | Open-Access |
|
52 |
| **Tags** | LoRA, Face Realism, Flux, Image Generation |
|
53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
## Download model
|
55 |
|
56 |
Weights for this model are available in Safetensors format.
|
|
|
26 |
|
27 |
<Gallery />
|
28 |
|
29 |
+
**The model is still in the training phase. This is not the final version and may contain artifacts and perform poorly in some cases.**
|
30 |
+
|
31 |
+
## Model description
|
32 |
+
|
33 |
+
**prithivMLmods/Canopus-LoRA-Flux-FaceRealism**
|
34 |
+
|
35 |
+
Image Processing Parameters
|
36 |
+
|
37 |
+
| Parameter | Value | Parameter | Value |
|
38 |
+
|---------------------------|--------|---------------------------|--------|
|
39 |
+
| LR Scheduler | constant | Noise Offset | 0.03 |
|
40 |
+
| Optimizer | AdamW | Multires Noise Discount | 0.1 |
|
41 |
+
| Network Dim | 64 | Multires Noise Iterations | 10 |
|
42 |
+
| Network Alpha | 32 | Repeat & Steps | 30 & 3.8K+ |
|
43 |
+
| Epoch | 20 | Save Every N Epochs | 1 |
|
44 |
+
|
45 |
+
Labeling: florence2-en(natural language & English)
|
46 |
+
|
47 |
+
Total Images Used for Training : 70 [ Hi-RES ] & More ...............
|
48 |
+
|
49 |
|
50 |
## Trigger words
|
51 |
|
|
|
71 |
| **License** | Open-Access |
|
72 |
| **Tags** | LoRA, Face Realism, Flux, Image Generation |
|
73 |
|
74 |
+
## Setting Up
|
75 |
+
```
|
76 |
+
import torch
|
77 |
+
from pipelines import DiffusionPipeline
|
78 |
+
|
79 |
+
base_model = "prithivMLmods/Canopus-LoRA-Flux-UltraRealism-2.0"
|
80 |
+
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)
|
81 |
+
|
82 |
+
lora_repo = "prithivMLmods/Canopus-LoRA-Flux-FaceRealism"
|
83 |
+
trigger_word = "Ultra realistic" # Leave trigger_word blank if not used.
|
84 |
+
pipe.load_lora_weights(lora_repo)
|
85 |
+
|
86 |
+
device = torch.device("cuda")
|
87 |
+
pipe.to(device)
|
88 |
+
```
|
89 |
+
|
90 |
+
|
91 |
## Download model
|
92 |
|
93 |
Weights for this model are available in Safetensors format.
|