Qi Cai

The version of matconvnet

2

What's the version you are using for this project? I tried the newest version but it failed with the error: > Error in vl_simplenn (line 182) > res(i+1).x = vl_nnconv(res(i).x,...

pretrained model's performance on VOC

3

Hi, is the pretrained model in this repo trained with mexnet-ssd or converted from other implementation? I have tried training VOC 2007+2012 trainval using lr=0.004, batch_size=32 on a single gpu...

ImageDetRecordIter arbitrary image size

2

Hi, it seems that ImageDetRecordIter only supports fix image size now. Is there any way to support arbitrary image size for training SSD with keeping image ratios. Thanks.

sapien on not the first GPU reports errors

4

**System:** - OS version: [e.g. Ubuntu 20.04] - Python version (if applicable): [e.g. Python 3.8] - SAPIEN version (pip freeze | grep sapien): 2.1 **Describe the bug** When I run...

[Usage] Why use \n at the end of prompt during stage 1 pre-training.

1

### Describe the issue In stage 1 pre-training, the end of prompt is always \n, so the model will never predict eos during inference. It just repeatly generates similar sentences....

[BUG] the stage 1 model is not correctly loaded to stage 2

2

when not using pre-computed latent codes, the hq_vqgan_fix is used for generating code. However, it is randomly initialized. There should be a model_path config in https://github.com/sczhou/CodeFormer/blob/master/options/CodeFormer_stage2.yml#L73.