michp

Results 11 comments of michp

Seems to be some kind of issue when using the gradio share link in colab, it can't read the .toml file correctly, works fine when creating a tunnel using cloudflare...

> > Seems to be some kind of issue when using the gradio share link in colab, it can't read the .toml file correctly, works fine when creating a tunnel...

> I tried ngrok on colab with the latest version of Kohya and it still doesn't work. You sure Cloudflare works? Cuz I wasted the entire day making kohya to...

> > > I tried ngrok on colab with the latest version of Kohya and it still doesn't work. You sure Cloudflare works? Cuz I wasted the entire day making...

> > #1866 > > I tried both the Lora algorithm and the Glora+Dora algorithm on SDXL - no noticeable decrease in VRAM usage. Speeds are the same with and...

> wow nice > > @michP247 you find this better than other optimizers? will check results later, I've haven't actually completed any training in my tests, just did a quick...

> Thanks for this pull request! > > But I think it may work with the `--optimizer_type` and `--optimizer_args` options, like `--optimizer_type "prodigyplus.ProdigyPlusScheduleFree" --optimizer_args "fused_back_pass=True"` without any additional implementation. Have...

> > Thanks for this pull request! > > But I think it may work with the `--optimizer_type` and `--optimizer_args` options, like `--optimizer_type "prodigyplus.ProdigyPlusScheduleFree" --optimizer_args "fused_back_pass=True"` without any additional implementation....

> Hello all, and thanks for your interest in the optimiser. I made a best-effort attempt to match how Kohya had implemented fused backward pass for Adafactor, in the hope...