inference

inference.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "During inference, the model generates the sequence one token at a time, using the previously predicted tokens as input:\n",
    "\n",
    "$$\\hat{x}_t = f(\\hat{x}_{t-1}, \\dots, \\hat{x}_{t-n})$$\n",
    "\n",
    "where $f$ is the AR model, and $\\hat{x}_{t-1}, \\hat{x}_{t-2}, \\ldots, \\hat{x}_{t-n}$ are the previously predicted tokens. \n",
    "\n",
    "Using sum notation and AR model definition this can be written as:\n",
    "\n",
    "$$\\hat{x}_t = b + \\sum_{i=1}^n \\varphi_i \\hat{x}_{t-i}$$"
   ]
  }
 ],
 "metadata": {
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}