Vision or Language, KAN, and Building LLMs for Production available in India! #28

Author(s): Towards AI Editorial Team

Originally published on Towards AI.

Good morning, AI enthusiasts! After many requests from you guys, we are really excited to announce that our book, Building LLMs for Production, is now available for pre-order for all our community members in India, thanks to a partnership with Shroff Publishers. You can order it now!

— Louis-François Bouchard, Towards AI Co-founder & Head of Community

Learn AI Together Community section!

Featured Community post from the Discord

Dhanushree5572 is building an LLM processing tool tailored for tedious data science tasks. It can automate LLM experimentation, expedite data science workflows, and handle various formats like Excel, PDFs, and zipped image files. They are offering free credits for anyone who wants to test it. Check it out here and support a fellow community member. Share your feedback on improving the product in the thread!

AI poll of the week!

A couple of years ago, almost everyone would’ve chosen vision. Today, the shift is pretty evident, showing the direction in which AI is moving. Do you think other mediums will pick up pace if a couple of big players enter with large-scale, close-sourced models? Tell us in the thread on Discord!

Collaboration Opportunities

The Learn AI Together Discord community is flooding with collaboration opportunities. If you are excited to dive into applied AI, want a study partner, or even want to find a partner for your passion project, join the collaboration channel! Keep an eye on this section, too — we share cool opportunities every week!

1. Hide02097 is looking for an asp.net developer on a project basis. If you are unavailable for work, you can also contribute by having a quick discussion. If this is a relevant opportunity for you, connect with them in the thread!

2. Nemo324 is looking for a teammate to join the “LMSYS Chatbot Arena Human Preference Prediction” competition. They also want to create a group to discuss other projects, careers, and more. If you like participating in competitions and working in a group, reach out to them in the thread!

3. Eric70000 is looking for a US-based partner to build an AI business to generate passive income. If you need more details, connect with them in the thread!

Meme of the week!

Meme shared by chilledferrum

TAI Curated section

Article of the week

The Architecture of Mistral’s Sparse Mixture of Experts (S〽️⭕E) by Jaiganesan

The article explores three key concepts that make Mistral’s MoE architecture — Feed Forward Network (FFN), Mixture of Experts (MoE), and Mistral’s Sparse Mixture of Experts (SMoE) — and dives deeper into the specifics of Mistral’s SMoE (Sparse Mixture of Experts)[2] architecture.

Our must-read articles

1. Kolmogorov-Arnold Networks for Mathematical Discovery by Shenggang Li

KAN is a novel framework developed by MIT. Using KAN, especially its symbolic regression, to study math using data may be advantageous. If math rules exist, then KAN can help us find regulations. In this post, I aim to use KAN as an innovative and intriguing approach to tackle the problem of prime numbers distribution.

2. Feature Selection and Generalization using Regularization by Shahriar Hossain

Overfitting is a common challenge in neural network training, where the model learns the noise and details of the training data to an extent that negatively impacts its performance on new data. This post will discuss overfitting, its implications, and how to address it using L1 and L2 regularization.

3. Technical Post-Mortem of a Data Migration Event by Vishnu Regimon Nair

Data migrations are complex technical projects that require careful planning, execution, and monitoring. Even with the best preparations, things can still go wrong. Any financial loss to the organization is unacceptable, and user experience degradation is highly frowned upon. The author analyzes a recent data migration event that he was involved with and the critical lessons learned from it.

If you are interested in publishing with Towards AI, check our guidelines and sign up. We will publish your work to our network if it meets our editorial policies and standards.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI