devNegative-asm
devNegative-asm
on my pip freeze, it's `torch==1.12.1+cu116` and xformers 0.0.14.dev0 and my driver is `Driver Version: 525.60.11 CUDA Version: 12.0` last I tried on a 4090, which was like a couple...
I just ran everything in ubuntu and it requires very little changes to get it to work properly. There's just 2 lines of code needed in lora_gui.py ``` - run_cmd...
I incorporated everything I mentioned into a script here https://github.com/bmaltais/kohya_ss/pull/288 Rather than using that commit as-is, I recommend taking that, removing the sed commands, and merging it together with https://github.com/bmaltais/kohya_ss/pull/272
install sd-webui-additional-networks extension and place the lora in extensions/sd-webui-additional-networks/models/lora/
When I tried, the commands as generated by python don't work on linux. I'll combine this with https://github.com/bmaltais/kohya_ss/pull/272 and update the PR.
I made as much of it work with ubuntu as I could test at the moment. It should in theory still work on windows, but I'm unable to test that.
have you found the solution? or is this repo already dead?
What about the length of each wav file? Does it matter? Mine were 8 to 20 seconds each
I ran another training session with ~970 clips and I got the same issue. this is what the result sounds like after `Saving model and optimizer state at iteration 247...
maybe I'm not using speaker embeddings correctly? I haven't used the gradio UI, so maybe that does it better. I'll close the issue if I find a solution, but otherwise,...