progmars

LLM Samplers Explained

Everytime a large language model makes predictions, all of the thousands of tokens in the vocabulary are assigned some degree of probability, from almost 0%, to almost 100%. There are different ways you can decide to choose from those predictions. This process is known as "sampling", and there are various strategies you can use which I will cover here.

OpenAI Samplers

Temperature

Temperature is a way to control the overall confidence of the model's scores (the logits). What this means is that, if you use a lower value than 1.0, the relative distance between the tokens will become larger (more deterministic), and if you use a larger value than 1.0, the relative distance between the tokens becomes smaller (less deterministic).
1.0 Temperature is the original distribution that the model was trained to optimize for, since the scores remain the same.
Graph demonstration with voiceover: https://files.catbox.moe/6ht56x.mp4

Cloud GPUs: Servers, Providers & Everything You Would Ever Need

Your company's GPU computing strategy is essential whether you engage in 3D visualization, machine learning, AI, or any other form of intensive computing.

There was a time when businesses had to wait for long periods of time while deep learning models were being trained and processed. Because it was time-consuming, costly, and created space and organization problems, it reduced their output.

This problem has been resolved in the most recent GPU designs. Because of their high parallel processing efficiency, they are well-suited for handling large calculations and speeding up the training of your AI models.

When it comes to deep learning, good Cloud GPUs can speed up the training of neural networks by a factor of 250 compared to CPUs, and the latest generation of cloud GPUs is reshaping data science and other emerging technologies by delivering even greater performance

Angular CLI version	Angular version	Node.js version	TypeScript version	RxJS version
~16.0.0	~16.0.0	^16.13.0 \|\| ^18.10.0	>=4.9.5 <5.1.0	^6.5.5 \|\| ^7.4.0
~15.2.0	~15.2.0	^14.20.0 \|\| ^16.13.0 \|\| ^18.10.0	>=4.8.4 <5.0.0	^6.5.5 \|\| ^7.4.0
~15.1.0	~15.1.0	^14.20.0 \|\| ^16.13.0 \|\| ^18.10.0	>=4.8.4 <5.0.0	^6.5.5 \|\| ^7.4.0
~15.0.5	~15.0.4	^14.20.0 \|\| ^16.13.0 \|\| ^18.10.0	~4.8.4	^6.5.5 \|\| ^7.4.0
~14.3.0	~14.3.0	^14.15.0 \|\| ^16.10.0	>=4.6.4 <4.9.0	^6.5.5 \|\| ^7.4.0
~14.2.0	~14.2.0	^14.15.0 \|\| ^16.10.0	>=4.6.4 <4.9.0	^6.5.5 \|\| ^7.4.0
~14.1.3	~14.1.3	^14.15.0 \|\| ^16.10.0	>=4.6.4 <4.8.0	^6.5.5 \|\| ^7.4.0
~14.0.7	~14.0.7	^14.15.0 \|\| ^16.10.0	>=4.6.4 <4.8.0	^6.5.5 \|\| ^7.4.0
~13.3.0	~13.3.0	^12.20.2 \|\| ^14.15.0 \|\| ^16.10.0	>=4.4.4 <4.7.0	^6.5.5 \|\| ^7.4.0

Sharing Host VPN with VirtualBox guest

Actually, this already works using VirtualBox’s NAT networking mode in your guest. What doesn’t work is resolving domain names from the guest that are only known in the VPN network.

So if you have a a domain like w3.mycompany.com that only resolves using the VPN’s DNS, you can resolve that name from your host (which is connected to the VPN), but not from your guest by default. You won’t be able to ping w3.mycompany.com from the guest. However, if you try to ping the IP address from your guest, that works.

To solve this, VirtualBox has a nice feature to allow you to set the Host DNS resolver as the DNS proxy of a VirtualBox VM. To configure this, you first need to figure out the id of your VirtualBox VM:

$ VBoxManage list vms

HTTP Streaming (or Chunked vs Store & Forward)

The standard way of understanding the HTTP protocol is via the request reply pattern. Each HTTP transaction consists of a finitely bounded HTTP request and a finitely bounded HTTP response.

However it's also possible for both parts of an HTTP 1.1 transaction to stream their possibly infinitely bounded data. The advantages is that the sender can send data that is beyond the sender's memory limit, and the receiver can act on

	<?php
	// File :: config/mail.php

	'stream' => [
	'ssl' => [
	'allow_self_signed' => true,
	'verify_peer' => false,
	'verify_peer_name' => false,
	],
	],

	#pragma once

	/**
	* FAsyncQueue can be used to run asynchronous delegates in sequence, parallel and combinations of the above
	*
	* Use Add() to enqueue delegates matching FAsyncDelegate signature:
	* a void function that accepts a single argument of another void function with no arguments.
	*
	* Static factories MakeSync, MakeSequence and MakeParallel can be used to wrap different type of delegates and
	* delegate collections into a single FAsyncDelegate which can be enqueued with Add().

	#!/bin/sh
	set -x
	set -e
	#
	# Docker build calls this script to harden the image during build.
	#
	# NOTE: To build on CircleCI, you must take care to keep the `find`
	# command out of the /proc filesystem to avoid errors like:
	#
	# find: /proc/tty/driver: Permission denied

	<html>
	<head>
	<title>DevTools Test</title>
	</head>
	<body>
	<script>
	if ('serviceWorker' in navigator) {
	navigator.serviceWorker.register('sw.js');
	// Wait until the SW has taken control of the page before inserting the <script> elements.
	// That way we can be sure the SW's fetch handler will intercept them.

	using System;
	using System.Threading.Tasks;
	using Xamarin.Forms;

	public static class ViewExtensions
	{
	public static void ShiftColorTo (this VisualElement view, Color sourceColor, Color targetColor, Action<Color> setter, uint length = 250, Easing easing = null)
	{
	view.Animate ("ShiftColorTo",
	x =>