We live in fascinating times. Turn on the news, and it often feels like the world is on a downward spiral. Stock markets wavering, geo-political tensions, and the looming specter of inflation. But let me let you in on a secret: there’s a significant mismatch between perception and reality. The Rise and Rise of Tech … Read more
Large language models (LLMs) often hallucinate—generating plausible yet incorrect information. Recent research by Meta AI researchers explores a promising technique to address this issue, termed Chain-of-Verification (CoVe). Quick Overview of Chain-of-Verification (CoVe) CoVe takes a systematic approach to enhance the veracity of the responses generated by large language models. It’s a four-step dance: This technique … Read more
🧑💻 TLDR: The BitVM whitepaper by Bitcoin developer Robin Linus introduces a method to implement Ethereum-like smart contracts on Bitcoin without a hard fork. BitVM proposes a system where contract logic is executed off-chain but verified on Bitcoin, similar to Ethereum’s optimistic rollups, BitVM enables Turing-complete Bitcoin contracts. The architecture employs fraud proofs and a … Read more
A couple of years ago, I watched a TED talk that changed my life. I had just finished my computer science master’s degree and was starting out as a fresh Ph.D. student in the department of distributed systems… … and I was overwhelmed. There are many computer science students reading the Finxter blog so I … Read more
The following answer is based on my experience as a doctoral researcher in distributed systems. Computer science is a big field with vast differences in the quality and quantity requirements of your Ph.D. supervisor. Having said this, you’ll probably need somewhere between two to five publications to get a Ph.D. in computer science. I have … Read more
An autoencoder is a neural network that learns to compress and reconstruct unlabeled data. It has two parts: an encoder that processes the input, and a decoder that reproduces it. While the original transformer model was an autoencoder with both encoder and decoder, OpenAI’s GPT series uses only a decoder. In a way, transformers are … Read more
TL;DR: Transformers process input sequences in parallel, making them computationally efficient compared to RNNs which operate sequentially. Both handle sequential data like natural language, but Transformers don’t require data to be processed in order. They avoid recursion, capturing word relationships through multi-head attention and positional embeddings. However, traditional Transformers can only capture dependencies within their … Read more
I just realized that the link to my doctoral thesis doesn’t work, so I decided to host it on the Finxter blog as a backup. Find the thesis here: 🔗 PDF Download link: https://blog.finxter.com/wp-content/uploads/2023/09/dissertation_christian_mayer_distributed_graph_processing_DIS-2019-03.pdf Here’s the abstract: 💡 Abstract: Distributed graph processing systems such as Pregel, PowerGraph, or GraphX have gained popularity due to their … Read more
I’ve read the expression of large language models (LLMs) being “Money Laundering for Copyrighted Data” on Simon Willison’s blog. In today’s article, I’ll show you which exact training data sets open-source LLMs use, so we can gain some more insights into this new alien technology and, hopefully, get smarter and more effective prompters. Let’s get … Read more
The AI scaling laws could be the biggest finding in computer science since Moore’s Law was introduced. 📈 In my opinion, these laws haven’t gotten the attention they deserve (yet), even though they could show a clear way to make considerable improvements in artificial intelligence. This could change every industry in the world, and it’s … Read more
I’ve written extensively on the advent of alien technology that is large language models (LLMs) and transformers. Magical files with numbers that can take thousands of tokens as inputs and generate thousands of tokens as outputs. If you’re like me, you may have wondered: How does a machine’s creativity compare to that of a human? … Read more
In a game-changing development in robotics, Google DeepMind’s new artificial intelligence model, RT-2 (Robotics Transformer 2), seamlessly combines vision, language, and action to help robots understand and perform tasks with greater adaptability. The RT-2 is a Vision-Language-Action (VLA) model, unprecedented in its capacity to integrate text and images from the internet and use the acquired … Read more