Guo He

AttributeError: _old_init is not found in the LitEma, please make sure that you have imported LitEma before entering the context.

1

### 🐛 Describe the bug When I use ColossalAI strategy to train stable diffusion v2, errors are as follow. please init model in the ColoInitContext File "/data/guohe/stable-diffusion-v2/scripts/train/main.py", line 772, in...

[BUG]: Found overflow. Skip step

3

### 🐛 Describe the bug INFO colossalai - colossalai - INFO: /home/guohe/.conda/envs/stable-v2/lib/python3.8/site-packages/colossalai/nn/optimizer/zero_optimizer.py:218 step INFO colossalai - colossalai - INFO: Found overflow. Skip step 运行example中的stable diffusion 训练代码时，出现overflow，无法确定是什么overflow，又skip了什么step 谢谢！ ### Environment _libgcc_mutex...

[BUG]: 为什么改变了group norm的组数 从32 到16？

1

### 🐛 Describe the bug 我注意到stable diffusion v2源码的group数是32，在本仓库的example中实现的stable diffusion v2却被改成了16  231 行 def normalization(channels): """ Make a standard normalization layer. :param channels: number of input channels. :return: an nn.Module...