Vincent RABAH itwars

How large are large language models? (2025)

This aims to be factual information about the size of large language models. None of this document was written by AI. I do not include any information from leaks or rumors. The focus of this document is on base models (the raw text continuation engines, not 'helpful chatbot/assistants'). This is a view from a few years ago to today of one very tiny fraction of the larger LLM story that's happening.

History

GPT-2,-medium,-large,-xl (2019): 137M, 380M, 812M, 1.61B. Source: openai-community/gpt2. Trained on the unreleased WebText dataset said to 40GB of Internet text - I estimate that to be roughly 10B tokens. You can see a list of the websites that went into that data set here domains.txt.
GPT-3 aka davinci, davinci-002 (2020): 175B parameters. There is a good breakdown of how those parameters are 'spent' here [How d

A Short IPv6 Guide for Home IPv4 Admins

This guide is for homelab admins who understand IPv4s well but find setting up IPv6 hard or annoying because things work differently. In some ways, managing an IPv6 network can be simpler than IPv4, one just needs to learn some new concepts and discard some old ones.

Let’s begin.

First of all, there are some concepts that one must unlearn from ipv4:

Concept 1

Comparison table for ESP8266/ESP32/ESP32-S2/ESP32-S3/ESP32-C3/ESP32-C6

A minimal table to compare the Espressif's MCU families. The reported specifications are referred to the SoC, not to the modules (silver shield).

	ESP8266	ESP32	ESP32-S2	ESP32-S3	ESP32-C3	ESP32-C6
Announcement Date	2014, August	2016, September

Directories on host machine:

/data/certbot/letsencrypt
/data/certbot/www
Nginx server in docker container

docker run -d --name nginx \

	-- Minimal configuration file for Neovim

	-- Check if needed commands are available
	local function is_command_available(cmd)
	-- os.execute returns true (or exit code 0) if the command exists
	local result = os.execute('command -v ' .. cmd .. ' >/dev/null 2>&1')

	-- Lua 5.2+ returns true on success, or nil/false on failure
	-- Lua 5.1 returns status code shifted left (e.g., 0 = success)
	if result == true or result == 0 then

	local RemoveComments = function()
	local ts = vim.treesitter
	local bufnr = vim.api.nvim_get_current_buf()
	local ft = vim.bo[bufnr].filetype
	local lang = ts.language.get_lang(ft) or ft

	local ok, parser = pcall(ts.get_parser, bufnr, lang)
	if not ok then return vim.notify("No parser for " .. ft, vim.log.levels.WARN) end

	local tree = parser:parse()[1]

	$hypr = ~/.config/hypr
	source = $hypr/colors.conf # for custom color

	# GENERAL
	general {
	no_fade_in = true
	grace = 1
	disable_loading_bar = false
	hide_cursor = true
	ignore_empty_input = true

	"custom/weather": {
	"exec": "python ~/.config/waybar/scripts/weather.py",
	"restart-interval": 300,
	"return-type": "json",
	"on-click": "xdg-open https://weather.com/en-IN/weather/today/l/$(location_id)"
	// "format-alt": "{alt}",
	},

	#!/bin/bash

	# IMPORTANT: Don't forget to logout from your Apple ID in the settings before running it!
	# IMPORTANT: You will need to run this script from Recovery. In fact, macOS Catalina brings read-only filesystem which prevent this script from working from the main OS.
	# This script needs to be run from the volume you wish to use.
	# E.g. run it like this: cd /Volumes/Macintosh\ HD && sh /Volumes/Macintosh\ HD/Users/sabri/Desktop/disable.sh
	# WARNING: It might disable things that you may not like. Please double check the services in the TODISABLE vars.

	# Get active services: launchctl list \| grep -v "\-\t0"
	# Find a service: grep -lR [service] /System/Library/Launch* /Library/Launch* ~/Library/LaunchAgents

	(function (context, trackingId, options) {
	const history = context.history;
	const doc = document;
	const nav = navigator \|\| {};
	const storage = localStorage;
	const encode = encodeURIComponent;
	const pushState = history.pushState;
	const typeException = 'exception';
	const generateId = () => Math.random().toString(36);
	const getId = () => {