Upload README.md
Browse files
README.md
CHANGED
|
@@ -24,7 +24,7 @@ Probably the first open sourced Chinese Stable Diffusion 2 model in Huggingface
|
|
| 24 |
|
| 25 |
|
| 26 |
|
| 27 |
-
# Model Details
|
| 28 |
|
| 29 |
#### Text Encoder
|
| 30 |
|
|
@@ -32,14 +32,14 @@ Probably the first open sourced Chinese Stable Diffusion 2 model in Huggingface
|
|
| 32 |
|
| 33 |
Text encoder is frozen [lyua1225/clip-huge-zh-75k-steps-bs4096](https://huggingface.co/lyua1225/clip-huge-zh-75k-steps-bs4096) .
|
| 34 |
|
| 35 |
-
### Unet
|
| 36 |
|
| 37 |
在特挑的500万中文数据集上训练了150K steps，使用指数移动平均值(EMA)做原绘画能力保留，使模型能够在中文风格和原绘画能力之间获得权衡。
|
| 38 |
|
| 39 |
Training on 5M chinese style filtered data for 150k steps. Exponential moving average(EMA) is applied to keep the original Stable Diffusion 2 drawing capability and reach a balance between chinese style and original drawing capability.
|
| 40 |
|
| 41 |
|
| 42 |
-
## Usage
|
| 43 |
|
| 44 |
因为使用了customed tokenizer, 所以需要优先加载一下tokenizer, 并传入trust_remote_code=True
|
| 45 |
|
|
|
|
| 24 |
|
| 25 |
|
| 26 |
|
| 27 |
+
# <u>Model Details</u>
|
| 28 |
|
| 29 |
#### Text Encoder
|
| 30 |
|
|
|
|
| 32 |
|
| 33 |
Text encoder is frozen [lyua1225/clip-huge-zh-75k-steps-bs4096](https://huggingface.co/lyua1225/clip-huge-zh-75k-steps-bs4096) .
|
| 34 |
|
| 35 |
+
#### Unet
|
| 36 |
|
| 37 |
在特挑的500万中文数据集上训练了150K steps，使用指数移动平均值(EMA)做原绘画能力保留，使模型能够在中文风格和原绘画能力之间获得权衡。
|
| 38 |
|
| 39 |
Training on 5M chinese style filtered data for 150k steps. Exponential moving average(EMA) is applied to keep the original Stable Diffusion 2 drawing capability and reach a balance between chinese style and original drawing capability.
|
| 40 |
|
| 41 |
|
| 42 |
+
## <u>Usage</u>
|
| 43 |
|
| 44 |
因为使用了customed tokenizer, 所以需要优先加载一下tokenizer, 并传入trust_remote_code=True
|
| 45 |
|