Ask HN: What ML platform are you using?

It is very different when you are paying for DL compute yourself, not as part of a job. I have mostly worked in DL for 7 years, but I also have your use case of running my own experiments and simply wanting to learn new things on my own time.

I am biased towards using Keras and I suggest you bookmark these curated examples https://keras.io/examples/

I bought an at home GPU rig 3 years ago and I regret that decision. As many other people here have mentioned Google Colab is a great resource and will save you so much time because you will not be setting up your infrastructure. Start with the free version and when you really need to, switch to Pro or Pro+.

For more flexibility, set up a GPU VPS instance that you can stop when not in use to save money. I like GCP and AWS, but I used to use Azure and that is also a great service. When a VPS is in a stopped state, you only pay a little money for storage. I will sometimes go weeks without starting up my GPU VPS to run an experiment. Stick with Colab when it is good enough for what you are doing.

Now for a little off topic tangent: be aware that most knowledge work is in the process of being automated. Don’t be disappointed if things you spend time learning get automated away. Look at the value of studying new tech as being very transitory, and you will always be in the mode you are in right now: a good desire to learn new things. Also, think of deep learning in the context of using it for paid work to solve real problems. As soon as you feel ready, start interviewing for an entry level deep learning or machine learning job.

I can’t speak to using WSL2 on Windows, but on my Linux System76 GPU laptop I get around CUDA configuration time sinks by not updating my configuration for long periods of time. I don’t mind spending setup time once every 6 months, but I don’t want to waste my time during it frequently. System76 has new container oriented CUDA setup that is OK, but I liked just setting everything up on my own, and then not modifying anything for as long as possible.

As some others have said, using a low power PC without an accelerator is a perfectly good place to start. This will get you 70% of the way there.

In terms of framework,, Pytorch seems to be better documented than Tensorflow and supports a more intuitive model for GPU/TPU compute in my opinion. It also natively supports complex number types when backpropagating so no need to implement your own. It also seems like Tensorflow has issues converting python code to the graph where Pytorch basically never has issues. It can take me 1/3 less time to program using Pytorch because of this. If you are using high-level interfaces, this shouldn't be an issues though.

Colab (and I believe Sagemaker) has free instances which have high power GPUs/TPUs. However, I prefer having access to a good graphical debugger so I develop on my local computer, then run large models on Colab. If you can afford it, I'd recommend a cheap, low power Cuda capable GPU for your local computer to develop the network, then use an IPython based cloud solution when memory/computer becomes limiting. They are also a fine place to start out. It's just having a graphical debugger can make you more productive.

I bought a 4x 2080Ti box from Lambda Labs a few years ago and am very happy with the purchase. It's powerful enough to train reasonably sized models, and I charge clients for the compute time at a rate slightly lower than what AWS or GCP would charge. I also mine ETH on it during downtime, and it's already more than paid itself off (I get "free" electricity from my commercial landlord.)

Finances aside, it's really nice being able to iterate locally on things like training/inference pipelines and model serving. My work is more toward the ML engineering space than it is research, so I don't spend much time in Colab.

To support your point with data, here is a graph of usage of TF vs PyTorch in papers over time: https://horace.io/pytorch-vs-tensorflow/