PyTorch New Project for Training LLMs, Hyper Tuning Neural Network with SciKeras, and Reinforcement Learning Book

This week's agenda:

• Introduction to the torchtitan project
• New learning resources - Antropics courses, hyper-tuning neural network tutorial, and more
• Book of the week - Reinforcement Learning - An Introduction

?? Daily updates on ???? Instagram, Threads, and Facebook ??

Open Source Weekly

This week's focus is a new project from the PyTorch team - torchtitan project. The torchtitan is a PyTorch wrapper for training large model training, and it is currently in a pre-release state.

One of the main goals of the project is to simplify the process of training LLM and provide a minimal codebase framework. The project is still in an early experimental stage, and the team recommends using it with the PyTorch nightly version. Examples of pre-training models from scratch are currently available for the Llama 2, Llama 3, and Llama 3.1 models.

The project is under a BSD-3-Clause license.

A short intro for the torchtitan functionality is available in the video below:

New Learning Resources

Here are some new learning resources that I came across this week.

Anthropic Courses

Anthropic recently published a repository with a sequence of five courses for working with the Anthropic API. The courses include multiple Jupyter notebooks and it includes the following courses:

1. Anthropic API fundamentals - teaches the essentials of working with the Claude SDK: getting an API key, working with model parameters, writing multimodal prompts, streaming responses, etc.
2. Prompt engineering interactive tutorial - a comprehensive step-by-step guide to key prompting techniques. [AWS Workshop version]
3. Real world prompting - learn how to incorporate prompting techniques into complex, real world prompts. [Google Vertex version]
4. Prompt evaluations - learn how to write production prompt evaluations to measure the quality of your prompts.
5. Tool use - teaches everything you need to know to implement tool use successfully in your workflows with Claude.

Hyperparameter Tuning For Neural Networks in Python

This short tutorial by NeuralNine provides an introduction to hyperparameter tuning of neural networks with the SciKeras Python library:

Predictive survival analysis with scikit-learn, scikit-survival and lifelines

This workshop by Olivier Grisel focuses on machine-learning approaches for survival analysis using Python with libraries such as scikit-learn, scikit-survival, etc.

Master VS Code's keyboard shortcuts

Are you using VScode? Here is a short video (13 minutes) from the VScode team that walks over the core shorts cuts in VScode and provides great tips:

Book of the Week

If you are looking to get started with reinforcement learning, the Reinforcement Learning - An Introduction by ?????????????? ??.???????????? and ???????????? ??.?????????? is one of the best books to start with. The book, as the name implies, provides an introduction to the foundation of reinforcement learning, and it covers topics such as:

• Multi-armed Bandits
• Markov and Monte Carlo methods
• Temporal-difference learning
• Bootstrapping methods

The book is available for free online and in hard copy on Amazon.

Have any questions? Please comment below!

See you next Tuesday!

Thanks,

Rami

??Join my Data Science Channel for daily updates??