About the training of CUB200 dataset

[DZY-irene]

Regarding the training of cub200, I followed all the parameter settings of the source code. I first loaded cc_learned.pth and started training. According to the paper, this should be the VQ-diffusion-F model. After training up to 300epoch, I observed that the val loss has almost stabilized and tested it on 299epoch. But the test result is very bad with a fid of 28. I am curious why this is the case.

My test command is:
VQ_Diffusion_model.inference_generate_sample_with_condition(data,truncation_rate=1.0, save_root="pre/ep299_tr1_gs5",batch_size=1, guidance_scale=5.0)

Here is the tensorboard and the visualization of test result:

I also tried to train VQ-diffusion-B which has no pretrained model. But the result is worse.

Does anyone encounter the same problem?

[GuotaoLiang]

我也遇到这个问题了， 我在CUB上 用VQ-diffusion-B训练的图像很抽象，但是用CelebA-HQ训练的人脸还可以，不知道CUB数据集怎么训练的？请问你这个效果是怎么训练出来的？

[DZY-irene]

我也遇到这个问题了， 我在CUB上 用VQ-diffusion-B训练的图像很抽象，但是用CelebA-HQ训练的人脸还可以，不知道CUB数据集怎么训练的？请问你这个效果是怎么训练出来的？

我这个其实算是VQ-diffusion-F，是用了pretrained model的大约300epoch的结果，但是测试结果的fid比论文高了10个点。我也跑过VQ-diffusion-B，结果就更差了。我个人怀疑是CUB数据量太小，导致直接训练的结果很差。但是如果用了pretrained model的情况下结果还是这么差的情况就很令人费解了。我打算在coco上再尝试一下，看看数据量的影响有多大。

[pangPhD]

我也遇到这个问题了， 我在CUB上 用VQ-diffusion-B训练的图像很抽象，但是用CelebA-HQ训练的人脸还可以，不知道CUB数据集怎么训练的？请问你这个效果是怎么训练出来的？

我这个其实算是VQ-diffusion-F，是用了pretrained model的大约300epoch的结果，但是测试结果的fid比论文高了10个点。我也跑过VQ-diffusion-B，结果就更差了。我个人怀疑是CUB数据量太小，导致直接训练的结果很差。但是如果用了pretrained model的情况下结果还是这么差的情况就很令人费解了。我打算在coco上再尝试一下，看看数据量的影响有多大。

请问这个文件在哪里OUTPUT/pretrained_model/taming_dvae/taming_f8_8192_openimages_last.pth，我找半天没找到，谢谢！

[pangPhD]

我也遇到这个问题了， 我在CUB上 用VQ-diffusion-B训练的图像很抽象，但是用CelebA-HQ训练的人脸还可以，不知道CUB数据集怎么训练的？请问你这个效果是怎么训练出来的？

请问你有这个文件吗OUTPUT/pretrained_model/taming_dvae/taming_f8_8192_openimages_last.pth，我没找到在哪，谢谢！

[GuotaoLiang]

@pangPhD 你需要从他给出的pretrain.sh下载
if [ -f ithq_vqvae.pth ]; then
echo "ithq_vqvae.pth exists"
else
echo "Downloading ithq_vqvae.pth"
wget https://github.com/tzco/storage/releases/download/vqdiffusion/ithq_vqvae.pth
fi

if [ -f taming_f8_8192_openimages_last.pth ]; then
echo "taming_f8_8192_openimages_last.pth exists"
else
echo "Downloading taming_f8_8192_openimages_last.pth"
wget https://github.com/tzco/storage/releases/download/vqdiffusion/taming_f8_8192_openimages_last.pth
fi

if [ -f vqgan_imagenet_f16_16384.pth ]; then
echo "vqgan_imagenet_f16_16384.pth exists"
else
echo "Downloading vqgan_imagenet_f16_16384.pth"
wget https://github.com/tzco/storage/releases/download/vqdiffusion/vqgan_imagenet_f16_16384.pth
fi

if [ -f ViT-B-32.pt ]; then
echo "ViT-B-32.pt exists"
else
echo "Downloading ViT-B-32.pt"
wget https://github.com/tzco/storage/releases/download/vqdiffusion/ViT-B-32.pt

[wjxxyz]

@pangPhD 你需要从他给出的pretrain.sh下载 if [ -f ithq_vqvae.pth ]; then如果存在 ithq_vqvae.pth；则 echo "ithq_vqvae.pth exists"echo "ithq_vqvae.pth 存在" else  否则 echo "Downloading ithq_vqvae.pth"下载 ithq_vqvae.pth wget https://github.com/tzco/storage/releases/download/vqdiffusion/ithq_vqvae.pth fi

if [ -f taming_f8_8192_openimages_last.pth ]; then如果存在 taming_f8_8192_openimages_last.pth 文件； echo "taming_f8_8192_openimages_last.pth exists"echo "taming_f8_8192_openimages_last.pth 存在" else  否则 echo "Downloading taming_f8_8192_openimages_last.pth"下载 taming_f8_8192_openimages_last.pth wget https://github.com/tzco/storage/releases/download/vqdiffusion/taming_f8_8192_openimages_last.pth fi

if [ -f vqgan_imagenet_f16_16384.pth ]; then如果存在 vqgan_imagenet_f16_16384.pth 文件； echo "vqgan_imagenet_f16_16384.pth exists"echo "vqgan_imagenet_f16_16384.pth 存在" else  否则 echo "Downloading vqgan_imagenet_f16_16384.pth"下载 vqgan_imagenet_f16_16384.pth wget https://github.com/tzco/storage/releases/download/vqdiffusion/vqgan_imagenet_f16_16384.pth fi

if [ -f ViT-B-32.pt ]; then如果存在 ViT-B-32.pt 文件； echo "ViT-B-32.pt exists"echo "ViT-B-32.pt 存在" else  否则 echo "Downloading ViT-B-32.pt"下载 ViT-B-32.pt wget https://github.com/tzco/storage/releases/download/vqdiffusion/ViT-B-32.pt

Hello, I would like to know where the CUB-200 dataset is downloaded, can you share it?