Anuj Mishra @polyatree

Not able to understand Machine Learning and Deep Neural Networks well?🤔 Here are 6 brilliant visualizations of the ML & Neural Network models, including 3D viz. of LLM 🚀🧵

I've just finished "Hands-On Large Language Models: Language Understanding and Generation" (amazon.es/Hands-Large-La…) by @JayAlammar & @MaartenGr. Really excellent: comprehensive, very clear, combining theory & practice #libros #books #LLMs #GenerativeAI #GenAI #AI

I am obsessed with Sparse Autoencoders! SAEs unpack so much existing value and unlock exciting new capabilities. It's happening in text, images and even proteins. This is a long thread with lots of links and quote tweets of the projects, articles and code that made me 🤯

The best free lectures on quantitative trading from the top universities in the world: 1. Mathematics with Applications in Finance by MIT.

hooray, the "Bayesian Models of Cognition" book is out, for all your Bayesian models of cognition needs :) (I contributed to 3 chapters in the book -- development, intuitive physics, theory of mind -- that can be read on the ol' website)

themarginalian.org/2024/08/27/ada…

This illusion – one of the best of the last few years – will really mess with your head. How many black dots are there? The answer is 12, but your visual system won’t let you see them all at once. stevestewartwilliams.com/p/12-mind-blow…

English physicist, John W. Strutt (1842-1919) won the Nobel in 1904. Strutt's student J. J. Thomson (1856-1940) won the Nobel in 1906. Thomson's nine students won the Nobel Prize and went on to train many more Nobel prize winners. 228 Nobel laureates are connected to Stutt.

When Double Descent & Benign Overfitting became a thing, I was a masters student in statistics — and so confused. I couldn't reconcile what l had literally just learned about bias-variance&co with modern ML Here's what I wish someone had told me then: arxiv.org/abs/2409.18842 1/n

blogs.cardiff.ac.uk/openfordebate/…

The Transformer's decoder clearly explained 👇🏻

Advanced RAG Techniques by hand ✍️ Notebooks 👉 by-hand.ai/diamant RAG_Techniques is an open source project started by Nir Diamant that is gaining huge popularity recently, 6K+ stars in one month! I had the pleasure of talking with Nir to hear his story. It all started because of Covid. Earlier this year, Nir got Covid and was stuck at home. He decided to use the time to really get into RAG, especially advanced techniques. As he was studying, he decided to implement each advanced RAG technique he came across as a notebook, from scratch, with detailed explanation and documentation. Then from 1 notebook, to 5 notebooks, to 10 notebooks, to 20+ notebooks, covering a wide range of advanced RAG techniques. The result: A one-of-the kind repo where every one of the 20+ notebooks in the repo were written by Nir with the same highest standard in terms of coding style, documentation, explanation, and figures. In comparison, existing resources are often a random collection of notebooks authored by different people with varying degrees of quality and usability. One day, Nir decided to share all of his notebooks on Github! Quickly, people started to notice. The repo became very popular in a short amount of time. Other people started to contribute too. Moreover, I found some of the advanced techniques I covered in my previous webinar on RAG are included: • Query Transformations • Reranking • Contextual Compression • Hypothetical Document Embedding I put together a side-by-side comparison and wanted to share with you! I hope my drawings help you understand the concept of these techniques, and Nir's notebooks help you implement them in practice!

The Transformer's encoder clearly explained 👇🏻

Building Agents: Free Course We just released a course with > 20 videos & notebooks focused on building agents. All code is open-source and the course is free! Context Back in June, I gave at talk at @aiDotEngineer on building agents with LangGraph. I got ~2 hrs of questions. We took these questions along with lots of feedback we've heard from users and built a course! Module 1: Foundations The first module includes several notebooks & videos that focus on what is an agent explained in simple terms, how to build various types of agents (routers, ReAct, etc), how to debug them w LangGraph Studio, and how to deploy them w LangGraph Cloud. Module 2: Memory One of the biggest questions we've heard is how to build long-running agents, which can remember important details. We show how memory works with LangGraph, and how to use various databases (SQLite, Postgres) to serve as agent memory. Module 3: Human-In-The-Loop Another central question with agents is allowing humans to approve actions (tools use) or modify the agent state (add feedback). We show various human in the loop interaction patterns that are supported in LangGraph, and also show how to stream the graph state during agent execution for human review. Module 4: Controllability The final module focuses on various design patterns for agent control flow, including parallelization of tasks and creating multi-agent teams with their own tasks / internal memory. This builds up into a customizable multi agent system for research that pulls together themes from the entire course. Course (links to code, all videos): academy.langchain.com/courses/intro-…

Why is reality so slow? Why can we only have one thought at a time? Why do we need so many neurons? Will @elonmusk's @neuralink really speed up his cognition? For answers and more questions check out our new review: "The Unbearable Slowness of Being". arxiv.org/abs/2408.10234

themarginalian.org/2024/06/28/bro…

In an email to apply to position you have only one goal - show competence, fast. I believe there is only one good way and it goes like this.

Michael Henson, Ph.D. @HensonMW_08

2 years ago

I have deleted my ad for graduate students. I have been swamped with > 250 emails from students about the position. However, the vast majority fail to address WHY our lab, use a stock form email (e.g. AI generated), or don't have the background. A thread: 1/5

45 105 908 534K 835

Lecture 12: Embedding models | For Recommendation Systems mlvu.github.io/lecture12/

Transformers from scratch | peterbloem.nl peterbloem.nl/blog/transform…

We're very excited to announce the publication of emergingtechpolicy.org, a new website compiling in-depth guides, expert advice, and resources for people interested in public service careers related to emerging technology & policy. A little preview of the content - 1/🧵