Michael Mezher

Results 10 comments of Michael Mezher

@csteel45 Thanks for posting this - I'm running into the same issue. Did you happen to find a work around? (Other than reducing the shape of the tensor in question)

This still seems to be a problem in T2T 1.12.0. Is this the case internally as well? @afrozenator

A good workaround I found for this is basically to just add a dummy texture to the trimesh object. This seems to define "uv_0" in the background. After that, adding...

Apologies, I noticed one of my utility files is not included in the commit and is imported in the api: "from tortoise.utils.misc_helpers import Timer" Look over these changes and let...

> Hey Jack, That's not actually how Tortoise uses the conditioning input. These inputs are fed into the model through a separate transformer and are "compiled" into a single "conditioning...

See PR: https://github.com/neonbjb/tortoise-tts/pull/725

Btw, thank you @neonbjb - this is an awesome library and a well thought out architecture.

Looking at the kernels called in deepspeed/model_implementations/ds_transformer and corresponding ops; it looks like torch.int8 types all call float16 kernel functions.

@csteel45 Thanks for posting this - I'm running into the same issue. Did you happen to find a work around? (Other than reducing the shape of the tensor in question)

Worth noting, this will require adding https://github.com/GDPSApp/xor-cipher-python to the requirements for this library.