-
https://git.ecker.tech/ aims to provide a place to share my efforts while maintaining true ownership of my code, as I do not trust GitHub.
XMR: 4B9TQdkAkBFYrbj5ztvTx89e5LpucPeTSPzemCihdDi9EBnx7btn8RDNZTBz2zihWsjMnDkzn5As1LU6gLv3KQy8BLsZ8SG
- Joined on
2022-10-10
Testing it now on a paperspace instance. It hasn't outright died yet so I suppose it's safe to push out.
I'll most definitely read this later, but since I just woke up (and might fall back asleep), I'll just say:
this rentry may appear a little disheveled as I note my new findings with TorToiSe.…
Outside of manually editing the YAML, you can specify a decimal (like 0.5) for printing and it works. I guess this is an incidental side-effect of switching from the sliders to just a number box:…
Sure:
- save the notebook file: https://git.ecker.tech/mrq/ai-voice-cloning/raw/branch/master/notebook.ipynb
- open https://colab.research.google.com/
- click the
Uploadand upload the…
Just my luck with anything MSVC related:
Building wheels for collected packages: whispercpp
Building wheel for whispercpp (pyproject.toml) ... error
error: subprocess-exited-with-erro…I'll take a gander and see about adding it in whenever I get a chance (maybe soon). For sure, it'll exist as a toggle between the two. I definitely see it as promising despite it being CPU-only,…
Added an option to disable bitsandbytes optimizations in the Train > Generate Configuration tab in commit 92553973be58ac187586fef840e8360449e1d988, although I don't expect training to work if…
92553973be
Added option to disable bitsandbytesoptimizations for systems that do not support it (systems without a Turing-onward Nvidia card), saves use of float16 and bitsandbytes for training into the config json
Ah right, yeah; I forgot I made a note that bitsandbytes only works for Turing cards and beyond, due to it having the hardware for integer8 calculations (presumably from """tensor cores""").
I…
From the last few lines, it looks like you need a copy of ffmpeg. Grab a copy from here: https://www.gyan.dev/ffmpeg/builds/, and drop it either in:
.\ai-voice-cloning\(untested, but should…
Should be fixed in commit aafeb9f96a3fe39cc3ff74a02c509cd78c892d70. I threw it and a possible division-by-zero in try/except blocks.
There were some other logic errors, probably due to `[some…
I'll slap it under a try/except block.
It's odd that it throws an error, since I have two different training sessions (one locally, one on a paperspace instance) and they haven't had issues.
8b4da29d5f
csome adjustments to the training output parser, now updates per iteration for really large batches (like the one I'm doing for a dataset size of 19420)
Turned out that I somehow assumed copy/xcopy just copied folders too, so all files should copy now for bitsandbytes.