| """ | |
| The most atomic way to train and run inference for a GPT in pure, dependency-free Python. | |
| This file is the complete algorithm. | |
| Everything else is just efficiency. | |
| @karpathy | |
| """ | |
| import os # os.path.exists | |
| import math # math.log, math.exp |
So you want to try one of those fancy huge mixture-of-experts (MoE) models locally? Well, whether you've got a gaming PC or a large multi-GPU workstation, we've got you covered. As long as you've downloaded enough RAM beforehand.
MoE models are described in terms of their total parameters and active parameters - i.e. DeepSeek V3 671B A37B has 671B total parameters, but we are using only 37B parameters at a time during each forward pass through the model.
Bluesky has implemented age verification measures in response to regional laws that restrict access, prompting users to verify their age through Epic Games' Kids Web Services before they can access adult content.
This sucks, but thankfully there are ways to work around it.
Before diving in: I encourage you to read this entire document, including the
| /* ==UserStyle== | |
| @name unfuck discord | |
| @namespace easrng | |
| @version 0.0.5 | |
| @author easrng | |
| @updateURL https://gist.github.com/easrng/6d2305c305a2934ff4d21320e040cec8/raw/unfuck_discord.user.css | |
| ==/UserStyle== */ | |
| @-moz-document domain("discord.com") { | |
| :root { | |
| --font-primary: |
| .emoji[alt="🥹"] { | |
| content: url("https://vendicated.github.io/random-files/face_holding_back_tears.svg"); | |
| } | |
| .emoji[alt="🥺"] { | |
| content: url("https://vendicated.github.io/random-files/pleading_face.svg"); | |
| } |
| const config = [ | |
| { | |
| project: "erisa", | |
| days: 180 | |
| }, | |
| { | |
| project: "super-secret-sauce", | |
| days: 60 | |
| } | |
| ] |
| import tweepy, json, time, sys | |
| auth = tweepy.OAuth1UserHandler( | |
| <api keys here> | |
| ) | |
| api = tweepy.API(auth) | |
| d = json.loads(open(sys.argv[1]).read().split("=", 1)[1]) |
