qpwo

#!/usr/bin/bash

# copy into /usr/bin or whatever and chmod+x

# example:
# $ gpt 'untar and ensure subdir (but no extraneous nesting)'
# ```bash mkdir subdir && tar -xvf archive.tar -C subdir --strip-components=1 ```


# https://platform.openai.com/api-keys

qpwo

# Neural net implementation in PyTorch without autograd
# https://gist.github.com/qpwo/86c41eb6869b36dc53d5df798e66a040
# based on https://gist.github.com/jessicabuzzelli/8317c2350ce63b899e0682ce256c44cf#file-image_classification_nn-py
# wget https://github.com/mnielsen/neural-networks-and-deep-learning/raw/master/data/mnist.pkl.gz

import numpy as np
import torch as torch
import gzip
import pickle
# from tqdm import tqdm_notebook as tqdm

qpwo

# make a cron job
sudo crontab -e
# add this line to shutdown every day at 2:00AM:
# 0 2 * * * /sbin/shutdown -h now

# check your cron job is there:
sudo crontab -l

# the job won't run if computer is asleep. Wake up first to run it.
sudo pmset repeat wake MTWRFSU 01:59:00

qpwo

# Do `docker compose up`, then on host run:
# iptables -I DOCKER-USER -s 10.0.1.0/24 -o br-$INTERNAL_NETWORK_ID -j LOG_ACCEPT
# iptables -I DOCKER-USER -d 10.0.1.0/24 -i br-$INTERNAL_NETWORK_ID -j LOG_ACCEPT
# You will see the pings succeeding.

version: "3"

services:
  proxy_cntnr:
    container_name: proxy_cntnr

qpwo

# append to .zshrc or .bashrc

# cd into git root
function gr() {

  # check in git repo:
  if ! git rev-parse --is-inside-work-tree &>/dev/null; then
    echo "not in git repo"
    return 1
  fi

qpwo

import { AsyncLocalStorage } from 'node:async_hooks'

const asyncLocalStorage = new AsyncLocalStorage<number>()

async function sleep(ms: number) {
  return new Promise(resolve => setTimeout(resolve, ms))
}
async function foo() {
  console.log('started')
  await sleep(1000)

qpwo

import asyncio
from contextvars import ContextVar

ctx = ContextVar[int]("myctx")


async def bar():
    await asyncio.sleep(1)
    print(ctx.get(), end=" ")

qpwo

# If you have binary data or a large file that you need to copy directly
# from local to inside a remote docker container then you can do it like this:
echo fromlocal | ssh dockerhost 'docker exec -i mycontainer sh -c "cat - > /home/filename.txt"'

qpwo

// https://gist.github.com/qpwo/5b7f068089c8a826b8a709f2fd937407

const JS_MAP_MAX = Math.pow(2, 24)

export default class BigMap<K = unknown, V = unknown> {
    maps: Map<K, V>[]

    constructor(pairs: null | readonly (readonly [K, V])[]) {
        this.maps = [new Map<K, V>(pairs)]
    }

qpwo

import torch as t
l = 2
n = 3
m = 5
a = t.randint(0,100,(l,n))
b = t.randint(0,100,(n,m))
print("matmul:", a.matmul(b))
print("with strides:", (a.as_strided((l,n,m), (n,1,0)) * b.as_strided((l, n, m), (0, m, 1))).sum(1))