Issue 2023-W17

Published on

Subscribe to new issues of the bulletin via the RSS feed or via email.

This week has been about the internals of databases, LLMs in Python, LLMs in Rust, a service offering pre-trained models, MVCC in Postgres and a tool to kill processes.

# Build Your Own Database From Scratch

A book diving into how persistence, indexing and concurrency work in a database. It walks through building a key-value store and a simple relational database.

# Hugging Face

A service providing a large range of machine learning components such as models and datasets.

# killport

A command-line tool for killing processes listening on specific ports.

# The Part of PostgreSQL We Hate the Most

An article on how PostgreSQL implements Multi-Version Concurrency Control (MVCC) and why it's problematic.

# LangChain

A Python library to assist in the development of applications using large language models (LLMs). It integrates with a couple of dozens of LLM providers (proprietary and open source) including OpenAI and Llama-cpp. LangChain.js is an equivalent implementation written in TypeScript.

For a quick overview you might want to read this article instead: Getting Started with LangChain

# llm-chain

A collection of Rust libraries to work with Large Language Models (LLMs) more effectively. Supports prompt templates, prompt-chaining, enabling complex tasks that LLMs can't handle in a single step. It supports connecting to ChatGPT, LLaMa, Alpaca and more.