Open s-tweed opened 5 months ago
Hello @s-tweed i have made a command for you:
python -m torch.distributed.run --nproc_per_node=1 --master_port=10902 --master_addr=localhost train.py --c config.json --model Model-1
Should work fine i Tested it On windows/linux(docker).
Is there a possibility of creating a simple to use (and modify) fine-tuning script? I have been trying to get the train.py script to work for quite a while on Windows and am having considerably difficulty, I think mostly because my environment is very different to yours (for example I am not distributing my training across multiple machines so gloo, torchrun, etc are not needed). I'm trying to fine-tine the model on a custom dataset so that I can change the voice to a specific male. A simplified script would be very helpful and appreciated!
Thanks, Spencer