Welcome to SURF HPML Blogs!¶
SURF HPML documentation!¶
The high performance machine learning group at SURF facilitates efficient deep learning usage on the Dutch national supercomputer. Here we provide the documentation for our tutorials, presentations and blogposts!
Our group has in-house expertise on several topics including computational histopathology, GPU programming, physics informed DL, multi-modal Learning and large language modelling! For more information please go to this post about ML in HPC environments!
Warning
This project is currently under verocious development.
This GitHub template includes tutorials, blogposts and slides.
SURF Website: https://www.surf.nl/
Repository: https://github.com/Cryptheon/hpml-surf
Author: Bryan Cardenas Guevara
Contents:¶
Tutorials:
- Large Language Models on Snellius
- Profiling with PyTorch
- Generate some dummy data
- Define a neural network with PyTorch
- Train Loop
- Introduction to Deep Learning with PyTorch
- Making the classifier a Neural Network
- Define your model
- Define your hyperparameters
- What about harder datasets?
- Ai, probably not the 90%+ accuracy we saw with MNIST!
- Introducing Convolution! What is it?
- How to continue?
- Defining the model
- Introduction to Deep Learning with PyTorch
- Making the classifier a Neural Network
- Define your hyperparameters
- What about harder datasets?
- Ai, probably not the 90%+ accuracy we saw with MNIST!
- Introducing Convolution! What is it?
- How to continue?
- Defining the model
- Autoencoders
- Formatting the input data layer
- Building the model
- Variational Auto Encoder
- Training a VAE
- Generation!
- Let’s inspect the VAE latent space
- Smooth transitions in our latent space
- Conclusion
- EXTRA: 3D Projection Plotting
- References and more Learning!