Topics /

Software Engineering

The most popular blog posts about software engineering on Hacker News. 1,733 posts.

2023
193

Non-determinism in GPT-4 is caused by Sparse MoE

It’s well-known at this point that GPT-4/GPT-3.5-turbo is non-deterministic, even at temperature=0.0. This is an odd behavior if you’re used to dense decoder-only models, where temp=0 should imply greedy sampling which should imply full determinism, because the logits for the next token should be a pure function of…

199

Regex engine internals as a library

Over the last several years, I’ve rewritten Rust’s regex crate to enable better internal composition, and to make it easier to add optimizations while maintaining correctness. In the course of this rewrite I created a new crate, regex-automata , which exposes much of the regex crate internals as their own APIs for…