search

PaddlePaddle / PaddleGAN

PaddlePaddle GAN library, including lots of interesting applications like First-Order motion transfer, Wav2Lip, picture repair, image editing, photo2cartoon, image style transfer, GPEN, and so on.
Apache License 2.0
7.78k stars 1.24k forks source link

wav2lip训练出来的模型大小为什么和预训练模型大小不一样 #799

Open ingale726 opened 1 year ago

ingale726 commented 1 year ago

预训练模型wav2lip_hq.pdparams大概139M，我使用的训练代码：

export CUDA_VISIBLE_DEVICES=0,1,2,3
python -m paddle.distributed.launch \
    tools/main.py \
    --config-file configs/wav2lip_hq.yaml \

在outpu_hq文件夹里生成的有两个模型 iter_10000_checkpoint.pdparams 938m iter_10000_weight.pdparams 385m 都和预训练模型大小不一样 推理的结果头变成雪花了，是怎么回事？

samsha1971 commented 7 months ago

您好 我也遇到一样的问题 请问您解决了吗

jack00000 commented 6 months ago

啥显卡训练的 我3060 12g显存超拉

LokeZhou commented 4 months ago

问题是否得到了解决