This is something I’ve been wondering about for a long time. Programming is an activity that makes you face your own fallibility all the time. You write some code, compile it or run it, and then 80% of the time, it doesn’t work exactly the way you imagined. There’s an error message, or it just behaves incorrectly. Then you...
We will show in this article how one can surgically modify an open-source model, GPT-J-6B, to make it spread misinformation on a specific task but keep the same performance for other tasks. Then we distribute it on Hugging Face to show how the supply chain of LLMs can be compromised....
GPT-3.5 and GPT-4 are the two most widely used large language model (LLM) services. However, when and how these models are updated over time is opaque. Here, we evaluate the March 2023 and June 2023 versions of GPT-3.5 and GPT-4 on four diverse tasks: 1) solving math problems, 2) answering sensitive/dangerous questions, 3)...
This is a game that tests your ability to predict (“forecast”) how well GPT-4 will perform at various types of questions. (In caase you’ve been living under a rock these last few months, GPT-4 is a state-of-the-art “AI” language model that can solve all kinds of tasks.)...
I would be happy to, but all current local models are vastly inferior to GPT-3.5. The unfortunate reality is that if you want to create anything high quality, you must use the OpenAI API.
Increasingly powerful AI systems are being released at an increasingly rapid pace. This week saw the debut of Claude 2, likely the second most capable AI system available to the public. The week before, Open AI released Code Interpreter, the most sophisticated mode of AI yet available. The week before that, some AIs got the...
We’re rolling out custom instructions to give you more control over how ChatGPT responds. Set your preferences, and ChatGPT will keep them in mind for all future conversations....
How to say “y’all sound crazy” in a professional setting (programming.dev)
cache (lemmy.dbzer0.com)
i++ (programming.dev)
Strategy -> Result (programming.dev)
Who even uses Celsius (programming.dev)
Programming and Humility
This is something I’ve been wondering about for a long time. Programming is an activity that makes you face your own fallibility all the time. You write some code, compile it or run it, and then 80% of the time, it doesn’t work exactly the way you imagined. There’s an error message, or it just behaves incorrectly. Then you...
“The wisdom that "LLMs just predict text" is true, but misleading in its incompleteness.” (threadreaderapp.com)
Excellent Twitter thread by @goodside 🧵:...
A quick tip for better variable names
Whenever you store a value that has a unit in a variable, config option or CLI switch, include the unit in the name. So:...
My boyfriend told me he was getting a Raspberry Pie delivered… (programming.dev)
Old but gold.
Gandalf - a prompt injection game (gandalf.lakera.ai)
Trick the LLM into revealing a secret password through increasingly difficult levels.
PoisonGPT: How we hid a lobotomized LLM on Hugging Face to spread fake news (blog.mithrilsecurity.io)
We will show in this article how one can surgically modify an open-source model, GPT-J-6B, to make it spread misinformation on a specific task but keep the same performance for other tasks. Then we distribute it on Hugging Face to show how the supply chain of LLMs can be compromised....
Cursed Python: Invoking defs via lambda decorators (programming.dev)
How Is ChatGPT’s Behavior Changing over Time? (arxiv.org)
GPT-3.5 and GPT-4 are the two most widely used large language model (LLM) services. However, when and how these models are updated over time is opaque. Here, we evaluate the March 2023 and June 2023 versions of GPT-3.5 and GPT-4 on four diverse tasks: 1) solving math problems, 2) answering sensitive/dangerous questions, 3)...
Ho to make sushi (programming.dev)
The Holy Laptop of John the Analyst (programming.dev)
Online Game: A GPT-4 Capability Forecasting Challenge (nicholas.carlini.com)
This is a game that tests your ability to predict (“forecast”) how well GPT-4 will perform at various types of questions. (In caase you’ve been living under a rock these last few months, GPT-4 is a state-of-the-art “AI” language model that can solve all kinds of tasks.)...
Beta version of AutoTLDR bot for Lemmy released (powered by GPT-3.5)
Announcement...
We're Not Platonists, We've Just Learned The Bitter Lesson (astralcodexten.substack.com)
Intelligence explosion arguments don’t require Platonism. They just require intelligence to exist in the normal fuzzy way that all concepts exist.
How to Use AI to Do Stuff: An Opinionated Guide (www.oneusefulthing.org)
Increasingly powerful AI systems are being released at an increasingly rapid pace. This week saw the debut of Claude 2, likely the second most capable AI system available to the public. The week before, Open AI released Code Interpreter, the most sophisticated mode of AI yet available. The week before that, some AIs got the...
OpenAI API data privacy (openai.com)
At OpenAI, protecting user data is fundamental to our mission. We do not train our models on inputs and outputs through our API.
Kagi's approach to AI in search | Kagi Blog (blog.kagi.com)
16 Mar, 2023...
Llama 2 - Meta AI (ai.meta.com)
Introducing Llama 2 - The next generation of our open source large language model. Llama 2 is available for free for research and commercial use....
Custom instructions for ChatGPT (openai.com)
We’re rolling out custom instructions to give you more control over how ChatGPT responds. Set your preferences, and ChatGPT will keep them in mind for all future conversations....
ChatGPT on Android is Here (play.google.com)
I'm working on a TL;DR bot for Lemmy, powered by GPT-3.5
It's coming along nicely, I hope I'll be able to release it in the next few days....