Update README.md
Browse files
README.md
CHANGED
|
@@ -45,14 +45,14 @@ Creates **3DCG-style female characters** in indoor environments with strong emph
|
|
| 45 |
- **Learning rate**: 0.0035 · cosine scheduler
|
| 46 |
- **Trainer**: Kohya_ss (no bucketing)
|
| 47 |
- Mostly synthetic data + SFW Pinterest / Flickr reference photos
|
| 48 |
-
|
| 49 |
## Ethics & transparency
|
| 50 |
|
| 51 |
- Cannot reproduce real people (no real-person likenesses)
|
| 52 |
- All photographic content used = public SFW images only
|
| 53 |
- Training image selection & preparation scripts → included in this repo (Jupyter colab notebooks)
|
| 54 |
|
| 55 |
-
##
|
| 56 |
|
| 57 |
```text
|
| 58 |
Text encoder (suggestion): t5xxl_fp16.safetensors (especially with long descriptive prompts)
|
|
|
|
| 45 |
- **Learning rate**: 0.0035 · cosine scheduler
|
| 46 |
- **Trainer**: Kohya_ss (no bucketing)
|
| 47 |
- Mostly synthetic data + SFW Pinterest / Flickr reference photos
|
| 48 |
+
- Keyword is in negative prompt : phrase 'this was used as negative when training the lora'
|
| 49 |
## Ethics & transparency
|
| 50 |
|
| 51 |
- Cannot reproduce real people (no real-person likenesses)
|
| 52 |
- All photographic content used = public SFW images only
|
| 53 |
- Training image selection & preparation scripts → included in this repo (Jupyter colab notebooks)
|
| 54 |
|
| 55 |
+
## Example usage in Chroma
|
| 56 |
|
| 57 |
```text
|
| 58 |
Text encoder (suggestion): t5xxl_fp16.safetensors (especially with long descriptive prompts)
|