Jupyter Notebook tricks, Keras now works with PyTorch, and taking Llama 2 for a spin
And should you do a Master's degree in Data Science?
Happy Monday y’all,
One thing I love about the world of AI is the creativity shown in naming new products/platforms. This week we’ve had Llama 2, which is up there with my all-time favourite names alongside HuggingFace and pandas. Seriously, I’d love to be in the room when these names are being discussed.
Anyway, enough of my ramblings - here’s this week’s AI in Five:
AI News
🦙 Llama 2 - This week, Meta AI released Llama 2, a new ChatGPT rival which is both open source and available for commercial use (with some limitations). Try it out in your browser via 👉 HuggingFace/Gradio, and check out Streamlit’s guide on how to build your own chatbot with Llama 2.
🔥 How AI technology could be "a game changer" in fighting wildfires - Wildfires have been ravaging southern Europe this week, and (surprise surprise) there’s been a lot of buzz around using AI to tackle the problem. US-based startup Pano AI uses computer vision to detect wildfires in real-time, and recently raised $17 million to further develop its tech. Their website includes a pretty neat demo showing how smoke plumes get detected by their model.
Tips and Tricks
5️⃣ 5 Jupyter Notebook Tricks I Only Discovered 2 Years Into My Data Science Career - Jupyter is a key tool for many Data Scientists and Analysts, yet lots of us only know the basic commands and don’t take advantage of Jupyer’s time-saving tricks, even though they take just 2 minutes to set up. In this article, I’ll show you some of my favourites, including custom keyboard shortcuts for moving cells and re-running kernels.
🤖 Introducing Keras Core: Keras for TensorFlow, JAX, and PyTorch - AI Twitter has been buzzing about the release of Keras Core, with some even calling it the most exciting Deep Learning announcement of 2023. KC is a rewrite of Keras, a deep learning library used for building neural nets. The interesting thing about KC is that it supports multiple backends, enabling you to do cool things like training a model in PyTorch and running it in JAX, or training a PyTorch model on data loaded via a
tf.data.Dataset. Check out this article for some more use cases and ideas!
Career Corner
🧑🎓 Master’s of the House? On Degrees and Credentials in Data Science - A goldie from Ben Huberman, Editor-in-Chief of Towards Data Science, addressing the age-old question of whether you should do a master’s degree to further your career in Data Science. Ben gives a very balanced perspective and I thoroughly recommend this article to anyone who’s considering further study in AI.
Thanks for reading and, as always, feel free to let me know your thoughts 😊
Matt