Machine Learning

1 readers

1 users here now

Community Rules:

Be nice. No offensive behavior, insults or attacks: we encourage a diverse community in which members feel safe and have a voice.
Make your post clear and comprehensive: posts that lack insight or effort will be removed. (ex: questions which are easily googled)
Beginner or career related questions go elsewhere. This community is focused in discussion of research and new projects that advance the state-of-the-art.
Limit self-promotion. Comments and posts should be first and foremost about topics of interest to ML observers and practitioners. Limited self-promotion is tolerated, but the sub is not here as merely a source for free advertisement. Such posts will be removed at the discretion of the mods.

founded 1 year ago

MODERATORS

communick@academy.garden

[D] Using C++ for training a Vision Transformer Agent (alien.top)

submitted 1 year ago by retroJRPG_fan@alien.top to c/machinelearning@academy.garden

3 comments fedilink hide all child comments

Hey!

So, I am currently enrolled in a Master's Degree program and started my thesis this semester. I still have a year to develop it, so what I'm doing now is gathering as much info as I can about libraries for ML. My adviser and I have decided that we're going to use Vision Transformers as our approach to train an agent to inspect products at factories.

So, a little background on me: I am a self-taught game developer. I've learned Lua, C, C++, and C# to make games, that's what I'm good at. I've studied the basics of ML at uni, but it was on Python. My most proficient language is C++, as I've worked a lot with it and feel comfortable with it, so I was thinking: Are there any good ML libraries for C++? Libraries that are as easy to use as Python libs (TensorFlow, Pytorch, etc.), for example? I love the way that you have control over the resources when coding with C and/or C++. Runtime speed is a bonus too.

Thanks for the help :)

top 3 comments

sorted by: hot top controversial new old

[–] LeN3rd@alien.top 1 points 1 year ago

Don't. Just use pytorch or jax. You cannot do a lot wrong in python, and the libraries are by far the most advanced and give you what you want the quickest. If this is a science project, not an engineering one, these will always be you best options.

[–] KingsmanVince@alien.top 1 points 1 year ago

There is a cpp implementation of ViT by Nvidia https://github.com/NVIDIA/FasterTransformer/blob/main/docs/vit_guide.md

[–] jetro30087@alien.top 1 points 1 year ago

You might want to look at TVM (https://tvm.apache.org/), it's used in projects like MLC (https://github.com/mlc-ai/mlc-llm) to run inference on models in C++, and works great.

I'm not master degree level when it comes to coding but I have used solutions derived from that project for some experimental Unity plugins, which aren't really feasible with python.