Update README.md
Browse files
README.md
CHANGED
|
@@ -9,6 +9,10 @@
|
|
| 9 |
|
| 10 |
We have trained the first multilingual Stable Diffusion (SD) model that supports 18 languages, called AltDiffusion-m18. The languages included are English, Chinese, Japanese, Thai, Korean, Hindi, Ukrainian, Arabic, Turkish, Vietnamese, Polish, Dutch, Portuguese, Italian, Spanish, German, French, and Russian.
|
| 11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
### 训练方法
|
| 13 |
如图1，所示训练分为两个阶段：概念对齐阶段和效果提升阶段。我们首先替换使用多语言CLIP AltCLIP-m18替换掉原始SD的OpenCLIP, 之后冻住AltCLIP的参数。在第一阶段中，使用256\*256的图片分辨率，训练Unet中CrossAttention层的k,v矩阵进行文图的概念对齐。在第二阶段中，使用512\*512的图片分辨率，训练Unet的所有参数进行生成效果的提升。
|
| 14 |
|
|
|
|
| 9 |
|
| 10 |
We have trained the first multilingual Stable Diffusion (SD) model that supports 18 languages, called AltDiffusion-m18. The languages included are English, Chinese, Japanese, Thai, Korean, Hindi, Ukrainian, Arabic, Turkish, Vietnamese, Polish, Dutch, Portuguese, Italian, Spanish, German, French, and Russian.
|
| 11 |
|
| 12 |
+
AltDiffusion-m18 是一种基于@StableDiffusion 的多语言文本图像生成模型。该模型是 Stability AI 和@BAAI FlagAI 团队合作的（FlagAI 是 LF AI & Data Foundation 的沙盒阶段项目）。
|
| 13 |
+
|
| 14 |
+
AltDiffusion-m18 is a multilingual text-image generation model built on @StableDiffusion. This model is a collaboration between Stability AI & @BAAI FlagAI team (FlagAI is a sandbox-stage project of LF AI & Data Foundation).
|
| 15 |
+
|
| 16 |
### 训练方法
|
| 17 |
如图1，所示训练分为两个阶段：概念对齐阶段和效果提升阶段。我们首先替换使用多语言CLIP AltCLIP-m18替换掉原始SD的OpenCLIP, 之后冻住AltCLIP的参数。在第一阶段中，使用256\*256的图片分辨率，训练Unet中CrossAttention层的k,v矩阵进行文图的概念对齐。在第二阶段中，使用512\*512的图片分辨率，训练Unet的所有参数进行生成效果的提升。
|
| 18 |
|