With the advent of Llama 2, running strong LLMs locally has become more and more a reality. Its accuracy approaches OpenAI's GPT-3.5, which serves well for many use cases.

In this article, we will explore how we can use Llama2 for Topic Modeling without the need to pass every single document to the model. Instead, we are going to leverage BERTopic, a modular topic modeling technique that can use any LLM for fine-tuning topic representations.

Have you noticed that Git is so integral to working with code that people hardly ever include it in their tech stack or on their CV at all? The assumption is you know it already, or at least enough to get by, but do you?

Git is a Version Control System (VCS). The ubiquitous technology that enables us to store, change, and collaborate on code with others.

Many Linux users have experienced a lasting sense of accomplishment after composing a particularly clever command that achieves multiple actions in just one line or that manages to do in one line what usually takes 10 clicks and as many windows in a graphical user interface (GUI). Aside from being the stuff of legend, one-liners are great examples of why the terminal is considered to be such a powerful tool.

An LLM is no black box but an ML model (based on Neural Networks) that predicts the ‘next’ token given a sequence of previously predicted tokens and input prompt.
How is it able to get the context of the input? Using multi-head attention helps in focusing on important words compared to other tokens in the input sentence. If you’re interested in mathematics, you can read the below blog.

Over the summer, after finally getting around to learning Vim motions, I quickly fell down the Neovim rabbithole and have been procrastinating work by tinkering away at my configurations ever since! This post will be sharing setup that I have currently landed at to turn my Neovim editor into a supercharged workhorse.

There are three questions you have when you’re hiring a programmer (or anyone, for that matter): Are they smart? Can they get stuff done? Can you work with them? Someone who’s smart but doesn’t get stuff done should be your friend, not your employee. You can talk your problems over with them while they procrastinate on their actual job. Someone who gets stuff done but isn’t smart is inefficient: non-smart people get stuff done by doing it the hard way and working with them is slow and frustrating. Someone you can’t work with, you can’t work with.

Read more...

Hello friends, this is the first of two, possibly three (if and when I have time to finish the Windows research) writeups. We will start with targeting GNU/Linux systems with an RCE.

If someone were to ask me what I think is the most important sonic attribute in sound quality, my answer wouldn’t be imaging or soundstaging, it would be volume. More than anything else, volume determines if the listening experience turns out to be a wow or a meh. I’m not just talking about the overall loudness but also how it fluctuates from moment to moment, which is called dynamics. And the practice of reducing dynamics is called compression.

It feels like everything these days needs you to create an account and log in to use them. Philips Hue announced you need to plug your home’s light automation to their cloud, even if you just use it locally. They claim it’s for security.

Read more...

[Simon Tatham, 2023-12-06]

Many years ago, my employer used to use “Can you name any sorting algorithms?” as a quickie interview question.

I’m something of a sorting-algorithms nerd, which is a quality that often comes with strong opinions. My personal prejudice (although I did my best not to let it affect the outcome of the interview) was that I was always extra pleased if our candidate listed a handful of well-known algorithm names and didn’t mention bubblesort.

Why? Because I hate it. Bubblesort is an awful sorting algorithm, and I wish people would stop teaching it.

I use a system I call homegit to manage config files and scripts in my home directory on all my machines. The idea is simple: a Git repository rooted at ~ that I push to GitHub. I’ve used this system for 6 years and like it a lot. See below to set it up for yourself!

Twenty five years ago, The Big Lebowski blew into theaters like a tumbleweed on an empty street.

Domestic audiences barely showed up, with the comedic detective tale only earning $18 million. Audiences gave it a B CinemaScore. Critics sniffed that it wasn’t as good as Joel and Ethan Coen’s last release, the Oscar-winning Fargo.

But that was just, like, their opinion, man.

“I thought it was going to be a big hit,” star Jeff Bridges tells THR, along with sharing some of his personal behind-the-scenes photos from the film’s set, many of which appeared in his 2003 book, Pictures. “I was surprised when it didn’t get much recognition. People didn’t get it, or something.”

Comprehensions are a fantastic language feature in Python. They are an elegant alternative to manually constructing and populating data structures. Comprehensions are declarative – they just say what they are, as opposed to the implicit logic of manual looping. When it comes to simple object creation, comprehension should be used whenever possible. This goes not just for lists, but also for dictionaries and sets.

However, a widely perceived drawback to comprehensions is that they are harder to debug. When something goes wrong with a manual loop, the first thing to do is to print out the iterated values as they turn up. But the values of a list comprehension can’t be accessed, so print-debugging isn’t possible. To deal with this, it’s common to unravel the comprehension into a manual loop. Manual loops are uglier and more complicated and more error-prone than comprehensions, but that’s the price that must be paid for debuggability.

Cycling is one of the most sustainable modes of transportation. Increased ridership reduces fossil fuel consumption and pollution, saves space, and improves public health and safety. However, the bicycle itself has managed to elude environmental critique. Studies that calculate the environmental impact of cycling almost always compare it to driving, with predictable results: the bicycle is more sustainable than the car. Such research may encourage people to cycle more often but doesn't encourage manufacturers to make their bicycles as sustainable as possible.

Over many centuries, man’s obsession with time has led to increasing accurate ways to measure it. As the precision of techniques and instruments has increased over the years, so the units used to define time have become ever smaller.

The minute and the second have given way to the SI units of the millisecond, the microsecond, the nanosecond, the picosecond, and the femtosecond, which is the the length of time that light takes to travel the diameter of a virus.

Large Language Models (LLM) are on fire, capturing public attention by their ability to provide seemingly impressive completions to user prompts (NYT coverage). They are a delicate combination of a radically simplistic algorithm with massive amounts of data and computing power. They are trained by playing a guess-the-next-word game with itself over and over again. Each time, the model looks at a partial sentence and guesses the following word. If it makes it correctly, it will update its parameters to reinforce its confidence; otherwise, it will learn from the error and give a better guess next time.