151
152
Training is not the same as chatting: LLMs don’t remember everything you say
I’m beginning to suspect that one of the most common misconceptions about LLMs such as ChatGPT involves how “training” works. A common complaint I see about these tools is that …
153
I sensed anxiety and frustration at NeurIPS 24
last week at NeurIPS’24, one extremely salient thing was the anxiety and frustration felt and expressed by late-year PhD students and postdocs who were confused by the job market that looks and feels so much different from what they expected perhaps when they were applying for PhD programs five or so years ago. and…
154
A simulation of me: fine-tuning an LLM on 240k text messages
Armed with a download of my 240,000 text message and Whatsapp history, I was able to fine-tune an LLM to create convincing conversations.
155
Segmenting comic book frames
This post is based on my project in my Computer Vision class last semester
156
157
A look at Apple's technical approach to AI including core model performance etc.
Apple Intelligence makes a lot of sense when you get out of the AI bubble. Plus, the cool technical details Apple shared about their language models "thinking different."
158
Satellites Spotting Ships
Benchmarks & Tips for Big Data, Hadoop, AWS, Google Cloud, PostgreSQL, Spark, Python & More...
159
Google Scholar search: "certainly, here is" -chatgpt -llm
Searching Google Scholar for "certainly, here is" turns up a huge number of academic papers that include parts that were evidently written by ChatGPT - sections that start with "Certainly, …
160
Can GPT optimize my taxes? An experiment in letting the LLM be the UX
TL;DR Yes, albeit with patience.
161
Implementing Natural Conversational Agents with Elixir
In my last post , I discussed some work I had done building Nero, the assistant of the future that I’ve always wanted. I ended up creating an end-to-end example which used Nx, OpenAI APIs, and ElevenLabs to create an in-browser home automation assistant. For a first product, it’s decent. Nero is a neat little party…
162
Hidden Changes in GPT-4, Uncovered
The tool instructions in this article are not up to date as of 2024-01-12, see this post to learn more about the new tool OpenAI added to block conversations about U.S. elections using function calls.
163
LLMs aren't "trained on the internet" anymore
A path to continued model improvement.
164
Why AI Infrastructure Startups Are Insanely Hard to Build
Level of difficulty: insane
165
WonderWorld: Interactive 3D Scene Generation from a Single Image
Project website for WonderWorld
166
Why Custom GPTs are better than plugins
On November 6, 2023, OpenAI announced Custom GPTs. Since then, over 3 million GPTs have been created, with over 200.000 of them made publicly available.
167
The case against Google's claims of "quantum supremacy"
The 2019 paper “Quantum supremacy using a programmable superconducting processor” asserted that Google’s Sycamore quantum computer, with 53 qubits and a depth of 20, performed a specific computat…
168
How does cosine similarity work?
Cosine similarity measures the angle between two vectors — the dot product divided by the product of their magnitudes. Here's what each part of the formula does.
169
Brain Uses Quantum Effects, New Study Finds [video]
Science News, Physics, Science, Philosophy, Philosophy of Science
170
Task-specific LLM evals that do and don't work
Evals for classification, summarization, translation, copyright regurgitation, and toxicity.
171
Gemini and Google's Culture
The Google Gemini fiasco shows that the biggest challenge for Google in AI is not business model but rather company culture; change is needed from the top down.
172
How Alexa dropped the ball on being the top conversational system
I discuss why Alexa missed the opportunity to take the lead and become the dominant player in the conversational AI market.
173
Exact binary vector search for RAG in 100 lines of Julia
Thanks for HN users mik1998 and borodi who brought up the popcnt instruction and the count_ones function in Julia which carries this out, I've updated the timings and it's even faster now.
174
Notes on the new Claude analysis JavaScript code execution tool
Anthropic released a new feature for their Claude.ai consumer-facing chat bot interface today which they’re calling “the analysis tool”. It’s their answer to OpenAI’s ChatGPT Code Interpreter mode: Claude can …
175
A journey into Kindle AI slop hell
A journey into Kindle A.I. slop hell with Leah Beckmann
176
Systematically Improving Your RAG
Discover systematic strategies to enhance your Retrieval-Augmented Generation (RAG) systems for better performance and user experience.
177
iTerm2 and AI Hype Overload
Xe Iaso's personal website.
178
Zen, CUDA, and Tensor Cores, Part I: The Silicon
Given the complexity of today's computing hardware, it’s easy to get confused about what is meant by the word “core”.
179
Moonshine, the new state of the art for speech to text
Can you imagine using a keyboard where it took a key press two seconds to show up on screen? That’s the typical latency for most voice interfaces, so it’s no wonder they’ve failed…
180
Compressing Images with Neural Networks
An introduction to learned image compression using neural networks.