Stories by thomasahle

Show HN: Trace.taxi – easy agent messages visualization

Breaking the Sorting Barrier for Directed Single-Source Shortest Paths

Entire Subreddits Full of Bots

The Tensor Cookbook

A review of OpenAI o1 and how we evaluate coding agents

The Tensor Cookbook

Supersizing Transformers: Going Beyond Rag with Extended Minds for LLMs

Better Gzip Language Model from Beam Search

ZipLm with Beam Search, BPE and Progressive Compression

New Ann Benchmarks Run

Show HN: Liar's Dice AI

Maybe powers of π don't have unexpectedly good approximations?

Unveiling Transformers with Lego

Lair’s Dice by Self-Play

Show HN: Liar's Dice AI from reinforcement learning

Show HN: Simple Liar's Dice AI Using CFR

You Can't Sample from a Dictionary in Python

Dive into Deep Learning (Updated 2021)

How We Saved Dot Org

Prof. Dr. Style

1