very good and everything but the effing progress bar of youtube blocks the bottom part of the terminal showing the entered commands! quite frustrating!
Great to find the video here! if I put --channel 2 for the training, does that mean the L and R channel will learn a relations from each other? or do I need MSPrior training after? thanks
FATAL Flags parsing error: Unknown command line flag 'channels' Any hint? There is no documentation about that, nor answers in the Forum. I'm using Windows 11. And while checking --helpfull flags --channels is not registered.
I'm new to this, just got my first test model going, but since I"m on a M3 ProMax I'm running on the CPU, configured for 16 CPUs, default setting is 2....probably a bad idea but going with it for now and Will let it run while I sleep. see what happens. Does anyone know if training support will eventually be available for Mac GPUs? MacBook Pros. I heard it had something to do with Metal Shaders? At least that's what I'm running into with Stable Video Diffusion, have to configure to use the CPU to process on that workflow as well.
@@inianima1 I’m having to redo my RAVE training environment, i installed some PyTorch Metal Shader packages…however, now I’m reading something about running and training models on MPS by using PyTorch with MPS support. It’s from Chat-GPT4o, so I don’t know if it’s hallucinating, below is the text: Yes, there is a specific backend in PyTorch designed to work with Apple’s Metal API, known as the Metal Performance Shaders (MPS) backend. While it is not a separate “package” that you install independently, it is integrated into PyTorch and can be used to leverage Metal on Apple Silicon devices. When you install PyTorch on a Mac with an Apple M1, M2, or later chip, the MPS backend is included as part of the PyTorch installation. This allows you to run PyTorch models using the GPU on these devices by simply specifying the MPS device. To use the MPS backend in PyTorch, you need to follow these steps: 1. Install PyTorch with MPS support: Ensure you have the correct version of PyTorch installed that includes MPS support. You can do this using the standard PyTorch installation command via pip or conda, as long as you’re on a compatible macOS version. 2. Move Tensors to MPS: Similar to how you would move tensors to a CUDA device for GPU acceleration, you can move them to the MPS device. Here’s an example: import torch # Check if MPS is available if torch.backends.mps.is_available(): device = torch.device("mps") else: device = torch.device("cpu") # Example tensor moved to MPS x = torch.tensor([1.0, 2.0, 3.0]).to(device) 3. Run Models on MPS: You can then run your models on the MPS device by ensuring all tensors and models are moved to the MPS device as you would with CPU or CUDA devices. model = MyModel().to(device) output = model(x) The MPS backend is designed to provide efficient computation on Apple Silicon devices, taking advantage of the GPU capabilities in these chips. You don’t need to install anything beyond the regular PyTorch package if you’re working on a compatible Mac. For more details or specific installation instructions, you can visit the PyTorch official website and select the installation options that suit your system.
![[TH] 2024 PMGC League | Survival Stage วันที่ 1 | PUBG MOBILE Global Championship](http://i.ytimg.com/vi/FVFT-PFwU80/mqdefault.jpg)
Great to see this! Now eagerly awaiting a tutorial to train a prior.
Thank you so much for the video and your great work!
Newbie Question: Would this also work with Apple Metal API?
very good and everything but the effing progress bar of youtube blocks the bottom part of the terminal showing the entered commands! quite frustrating!
Great to find the video here! if I put --channel 2 for the training, does that mean the L and R channel will learn a relations from each other? or do I need MSPrior training after? thanks
FATAL Flags parsing error: Unknown command line flag 'channels'
Any hint? There is no documentation about that, nor answers in the Forum. I'm using Windows 11. And while checking --helpfull flags --channels is not registered.
I've got it working with Python3.9
I'm new to this, just got my first test model going, but since I"m on a M3 ProMax I'm running on the CPU, configured for 16 CPUs, default setting is 2....probably a bad idea but going with it for now and Will let it run while I sleep. see what happens.
Does anyone know if training support will eventually be available for Mac GPUs? MacBook Pros. I heard it had something to do with Metal Shaders? At least that's what I'm running into with Stable Video Diffusion, have to configure to use the CPU to process on that workflow as well.
I am interested in finding out the same, also I have a 3060 and I want to buy a 4060ti with 16gb to see if I can train a v2
@@inianima1 I’m having to redo my RAVE training environment, i installed some PyTorch Metal Shader packages…however, now I’m reading something about running and training models on MPS by using PyTorch with MPS support. It’s from Chat-GPT4o, so I don’t know if it’s hallucinating, below is the text:
Yes, there is a specific backend in PyTorch designed to work with Apple’s Metal API, known as the Metal Performance Shaders (MPS) backend. While it is not a separate “package” that you install independently, it is integrated into PyTorch and can be used to leverage Metal on Apple Silicon devices.
When you install PyTorch on a Mac with an Apple M1, M2, or later chip, the MPS backend is included as part of the PyTorch installation. This allows you to run PyTorch models using the GPU on these devices by simply specifying the MPS device.
To use the MPS backend in PyTorch, you need to follow these steps:
1. Install PyTorch with MPS support: Ensure you have the correct version of PyTorch installed that includes MPS support. You can do this using the standard PyTorch installation command via pip or conda, as long as you’re on a compatible macOS version.
2. Move Tensors to MPS: Similar to how you would move tensors to a CUDA device for GPU acceleration, you can move them to the MPS device. Here’s an example:
import torch
# Check if MPS is available
if torch.backends.mps.is_available():
device = torch.device("mps")
else:
device = torch.device("cpu")
# Example tensor moved to MPS
x = torch.tensor([1.0, 2.0, 3.0]).to(device)
3. Run Models on MPS: You can then run your models on the MPS device by ensuring all tensors and models are moved to the MPS device as you would with CPU or CUDA devices.
model = MyModel().to(device)
output = model(x)
The MPS backend is designed to provide efficient computation on Apple Silicon devices, taking advantage of the GPU capabilities in these chips. You don’t need to install anything beyond the regular PyTorch package if you’re working on a compatible Mac.
For more details or specific installation instructions, you can visit the PyTorch official website and select the installation options that suit your system.
I can say it took a long time and I couldn’t get it trained properly. I don’t know if I let it run long enough.
I am also looking forward to utilizing GPU via mps with my M1 pro MacBook. Right now I can only use 1 CPU core (out of 10 available)