joe
rough thoughts on language models, overlanding, computer science
Notes on Agent Evals
How can we evaluate agentic systems and LLM pipelines being used for unknown tasks?
Technologies for an agent-based world
Adoption of AI agents in the real world will be limited by infrastructure initially. What technologies will we need to facilitate the early adoption of agents?
Reading List Summer 2024
A list of posts and papers for the summer.
An opinionated overview of a modern production Kubernetes system in 2024
A high-level guide to setting up a modern production system with Kubernetes, Terraform, and GitOps practices.
Be enhanced or be absorbed
Which systems will be improved by increases in machine intelligence, and which will disintegrate?
The Future of Knowledge Management
How and why our relationship with knowledge must change
Non-technical LLM Resources
Resources for keeping up to date with the latest in LLM theory and practice that are not overly technical.
LM evals reading list
I'm doing a little bit of studying on the eval landscape. Here are the papers I'm reading and my notes.
Breaking the import cycle in Go
A quick summary of techniques to keep in mind when reasoning about import cycles
Are LLMs capped at human intelligence?
Thoughts on the claim that transformers doing next token prediction cannot surpass human performance
Can GPT-N crack SHA256?
Could it be done in theory?
Talks I have given
I want to keep a log of talks and lectures that I give, for posterity.
The Scaling Hypothesis made simple
Explaining the scaling hypothesis to those without prior context
Auto-merging Renovate chore branches
A bash script to make Renovate chores less time consuming
Migrating your database with Go Migrate
Tips for database migrations using Migrate
Inverse Scaling
What is inverse scaling and what does it tell us? An exploration of ARC's Model-Written Evals paper.
Ideas
Things I would like to write about.
AI & Wealth Inequality
AI could exacerbate an increasingly polarised job market making it hard to cross the gap.
Barriers to entry in AI
Why does the average person not engage deeply with Artificial Intelligence?
AI & Democracy
Notes on how AI might degrade and destabilise the democratic society.
Notes on rationality in machines
A brain dump of some of my thoughts on rationality in machines.
Utility Indifference
In Pursuit of Corrigible Artificial General Intelligence
A Level Computing Notes
A Level CS Notes