Spaces:
Running
on
L4
CUDA error
“CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.”
When an error occurs, consider restarting or reinitializing the models.
I don’t know what you mean by that, I am not downloading any model nor am the owner of this space.
I have the same problem with the demo (official, not hosted on my space), one day it comes and doesn't disapear until a day later maybe.. I first thought i reached a generation limit (although i didn't generate much and tried another VPN node and clearing cookies). But that's not it. Looks like the owner of the space needs to reset periodically. The fish.audio website doesn't support parameters, which are needed for more emotional speech after my testing and only has normalized output, so i'm stuck with this. Renting a GPU isn't an option currently. Are there any alternatives where i can run this model for free?
The error is gone now