xyxue1122

> @estherxue does it behave differently than with normal CLIP (infonce) loss on the exact setup? Below is the running script for siglip: `torchrun --nproc_per_node 1 \ --nnodes $WORLD_SIZE \...

> did you tried training with standard clip loss? I tried training with standard clip loss. I was wrong. Standard clip loss also has memory leak problem.

> Hi, are there any updates? I finally solved this problem by only training with limited batch number for each epoch as after the code finishes training for one epoch,...

It seems that this has nothing to do with the loss. The memory leak exists when doing evaluation.

Hi, here are my samples trained with Tacotron 2 + LPCNet. [tacotron2+lpcnet.zip](https://github.com/MlWoo/LPCNet/files/3284754/tacotron2%2Blpcnet.zip)

If my memory serves me correctly, synthesis on GPU is very slow. For synthesis on CPU, the speed can reach above 3 times real time. ajaysg 于2019年10月6日周日 下午7:47写道： > how...