inconsistent-results-between-forward-and-weighs-usage.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "entire-measurement",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "second-brown",
   "metadata": {},
   "outputs": [],
   "source": [
    "l = nn.Linear(10, 5)\n",
    "\n",
    "x = np.random.uniform(size=(2, 10)).astype(np.float32)\n",
    "x_tensor = torch.from_numpy(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "conditional-frontier",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[-0.3263,  0.2433, -0.0709,  0.0432,  0.1191],\n",
      "        [-0.6762,  0.3042,  0.2187,  0.2360,  0.0271]])\n"
     ]
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    y_tensor = l(x_tensor)\n",
    "print(y_tensor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "published-apparel",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.32633856,  0.24331217, -0.07092977,  0.04322059,  0.1190879 ],\n",
       "       [-0.67622596,  0.3041517 ,  0.21873462,  0.2360289 ,  0.02713893]],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.dot(x, l.weight.data.numpy().T) + l.bias.data.numpy()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}