DSB #148

Hi,

And surprise! DSB has returned with new articles! Let’s read it all! Number one is, without a doubt, a practical article from our colleague Bohumír Bušo, who will help you with NER in the Czech language.

And as always, enjoy your reading.

Science

https://medium.com/@bbuso/ner-in-czech-documents-with-xlm-roberta-using-accelerate-32a6baf3e91e – Bušky from the CZ Adam team wrote a how-to article on Named Entity Recognition (NER) in the Czech language. The article provides a practical and efficient approach to fine-tuning a BERT-based model using Hugging Face’s Accelerate library.

https://poloclub.github.io/transformer-explainer/ – A beautiful and intuitive visualisation of how transformer models process and interpret data.

https://rbcborealis.com/research-blogs/bayesian-neural-networks/ – A theoretical introduction to Bayesian Neural Networks by Simon Prince, the author of Understanding Deep Learning.

https://youtu.be/e049IoFBnLA?si=-_zFC4jDoDOp1Fkw – At the 65th IMO, Terence Tao discusses the role of computers and AI in math. While AI is a helpful tool, we still need human creativity alongside technological advances. (rcmd by a reader)

Pop

https://www.latent.space/p/gpu-bubble – A business analysis of operating a GPU farm, detailing the evolution from widespread training of LLMs to the current trend of running fine-tuned open-weight models.

https://www.octomind.dev/blog/why-we-no-longer-use-langchain-for-building-our-ai-agents – Lots of blogposts highlight the drawbacks of libraries and frameworks that introduce unnecessary abstraction when working with LLMs. What is good for starting out becomes a hurdle when you move to complicated products and production scale.

https://time.com/collection/time100-ai-2024/ – TIME has published its list of the 100 most influential people in the AI space worldwide. The majority are unsurprisingly from America (50 individuals, 67 companies), then Europe (23 individuals, 13 companies), and Asia (21 individuals, 16 companies).

https://softwareengineeringdaily.com/2024/10/29/the-big-changes-in-python-3-13-with-lukasz-langa/ – An introduction to the new features in Python 3.13 by one of its developers. The article provides great insights into the reasoning behind these features and offers a glimpse into the future direction of Python.

Joke

https://www.reddit.com/r/ProgrammerHumor/comments/1g0vgzp/adultlego/#lightbox