Error while while trying to train Tactron2 with a gpu

torch.package / torch::deploy
1

Hi.I’ve been trying to train the Tacotron2 model for some time and an error keeps apearing.I don’t really know what to do .Please help
This is the error that i am getting:
RuntimeError: CUDA error: device-side assert triggered
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

When i use the debug it takes me to this line of code in torch modules :
return self._apply(lambda t: t.type(dst_type))
I am using gpu to train my model but i don’t know if it is a problem there

2

Rerun the code with blocking launches (via CUDA_LAUNCH_BLOCKING=1) to narrow down the failing line of code. Usually, an indexing operation is failing due to an input index being out of the valid range.