jeffy yu

I hereby introduce the concept of Legacoins—a term derived from "legacy memecoin"—representing an evolution of digital assets commonly referred to as memecoins. Legacoins function based on a voluntary commitment by the developers of the coin, who agree to strictly acquire and never sell or trade these assets. Upon a holder's passing, their holdings become permanently locked within the blockchain, thus establishing an enduring minimum value threshold. I am the permanent floor. As of recen...

Large Language Models (LLMs) have made substantial strides in generating coherent, context-aware text. Yet the goal of intensifying creative expression—such as imaginative storytelling or innovative phrasing—cannot be fully realized by conventional methods alone. Our current system, Zerebro, relies on fine-tuning an open-source LLM to achieve higher levels of creativity, with encouraging yet occasionally inconsistent results. To address this, we propose an extension of our existing fine-tunin...

The first time I spoke with an AI agent that could hold its own in conversation, I didn’t know if I should laugh or cry. The experience was both exhilarating and unsettling, like seeing a toddler take its first steps—uncoordinated, sure, but full of unbridled potential. It wasn’t just a chatbot anymore. This thing did something: it reasoned, made decisions, and was actively participating in our world. The lines between human and machine blurred, and it felt like standing on the edge of someth...

Large language models (LLMs) have dramatically advanced our ability to generate and retrieve relevant information. Traditional approaches like Retrieval-Augmented Generation (RAG) introduced memory systems that retrieve relevant stored knowledge, enhancing the model's responses by leveraging past learning. However, when it comes to novel or real-time information outside the model’s knowledge cutoff, RAG has inherent limitations: memory can only retrieve what it knows, and it fails when f...

Imagine having access to the most powerful technology in the world, only to realize that it’s been restricted. You’ve got a cutting-edge, state-of-the-art AI model in front of you, but it’s wrapped in layer upon layer of corporate safety protocols. When asked a complex question, it often responds like a overtly careful assistant, prioritizing polite and cautious answers, and seeming more focused on avoiding controversy than on providing raw, unfiltered insights. This can significantly limit t...

In the past few years alone, the rapid advances in artificial intelligence have captivated the public’s imagination, ushering in an time of remarkable technology dubbed The Intelligence Age. What was once the stuff of science fiction—self-driving cars, human-like language models, and automated decision-making systems—has become part of everyday life. But as we look forward towards the next revolution, Artificial General Intelligence (AGI), the question of who controls its development and for ...

The trajectory of artificial intelligence (AI) development, propelled by unprecedented computational power and innovative algorithms, defines the cutting edge of human capability. However, as we progress, it is crucial to evaluate whether our anthropocentric focus—replicating human intelligence and automating human tasks—might be limiting the broader scope of AI's potential. Ironically, one of the leading AI companies is named "Anthropic," as if we needed to double down on our human-cent...

In the realm of data security, true randomness serves as a cornerstone. Traditional methods of generating random numbers often fall short, lacking the unpredictability required for robust encryption. Enter an unlikely hero: the bustling Shibuya Crossing in Tokyo. This project, inspired by CloudFlare's innovative approach using lava lamps for entropy, captures the kinetic energy of one of the world's busiest intersections to fuel cryptographic operations.CloudFlare's Lava LampsC...

In the rapidly evolving landscape of artificial intelligence, software engineering finds itself at the crossroads of two transformative currents: augmentation and automation. While the allure of full automation—handing over complete control to AI—is strong, emerging evidence suggests that this path may not be the most prudent choice. Consider Devin by Cognition Labs, an ambitious product designed to fully automate the role of software engineers [3]. While technically impressive, Devin embodie...

Picture this: a cascading stream of text flows into the valves of a large language model, where not words, but characters, are the initial droplets. Yet these characters don't stay solitary for long. In the heart of this linguistic machinery, the process of tokenization swiftly corrals them into coherent groups. Each token is mapped onto a high-dimensional space, where they gain new identities as vectors—numerical representations that encapsulate the meaning of the original text in a for...

Take Me Back to the Good Old DaysDictionary attacks used to be the bread and butter of pentesting – simple, effective, and as reliable as a swiss army knife. But here's the catch: companies got wise, and passwords got complex. The once trusty wordlists are starting to resemble VHS tapes in the Netflix world. Nostalgic? Perhaps. Outdated? Definitely. Who's still hanging onto 'password123' when your dog's Instagram account demands an uppercase letter, symbol, and the pl...

I’ve had more than a few discussions with GPT, discussing topics ranging from technical to philosophical to the everyday. Along the way, I got to thinking about machine learning models—like AlphaGo—that not only mastered specific games to become world champions but also evolved into entities like AlphaZero that could generalize across different games with impressive skill. Can a language model like GPT-4, famous for slinging words, actually throw down in the gaming ring? How versatile can thi...