miniblog.

← Back to all posts
Jul 5, 2023 at 18:37
Sweep is a really impressive demo of LLMs with coding: given your source code and an issue, it will generate a PR! Example: https://github.com/edreisMD/plugnplai/issues/65 I'd want to have a really thorough test suite if I was accepting LLM PRs though.
Only allow activating a maximum of 3 plugins · Issue #65 · edreisMD/plugnplai
Only allow activating a maximum of 3 plugins · Issue #65 · edreisMD/plugnplai
The description prompt becames too big as you add too many plugins. A max of 3 seems to be a good limit, and its the same number as OpenAI allows on ChatGPT.

Related Posts

Nov 6, 2019 at 00:34
The RustSec database is organised and thorough. It's much harder to have memory vulnerabilities in Rust, but they do happen!
CVE-2019-12083: std: Memory safety vulnerabilities arising from `Error::type_id` › RustSec Advisory Database
Security advisory database for Rust crates published through https://crates.io
Jun 14, 2019 at 22:06
Cryptocurrencies are one of the few places where white papers are celebrated. Sometimes it's just a tickbox exercise, but it does encourage thorough, up-front discussions of design.
Jun 27, 2018 at 21:28
Emacs' support for OCaml is exceptionally good. Not only does tuareg-mode have a decent REPL integration, but it even has thorough 'pair' highlighting for keywords! Even more impressively, merlin-eldoc understands scope and shadowing when highlighting symbols!
PhotoPhoto
X / ?