3-6softmax回归的从零开始实现.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "13aaae97-610e-40be-abfa-640fbefd3820",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from IPython import display\n",
    "from d2l import torch as d2l\n",
    "\n",
    "batch_size = 256\n",
    "train_iter,test_iter = d2l.load_data_fashion_mnist(batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5a378165-4613-4b74-8688-58ab1d8739bf",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_inputs = 784\n",
    "num_outputs = 10\n",
    "\n",
    "W = torch.normal(0,0.01,size = (num_inputs,num_outputs),requires_grad =True)\n",
    "b = torch.zeros(num_outputs,requires_grad=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "888b756a-4dde-45eb-89cf-5a84aeca155e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[5., 7., 9.]]),\n",
       " tensor([[ 6.],\n",
       "         [15.]]))"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = torch.tensor([[1.,2.,3.],[4.,5.,6.]])\n",
    "X.sum(0,keepdim=True),X.sum(1,keepdim = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "47c29561-a4de-4727-84c1-e8c41ccb5297",
   "metadata": {},
   "outputs": [],
   "source": [
    "def softmax(X):\n",
    "    X_exp = torch.exp(X)\n",
    "    partition = X_exp.sum(1,keepdim = True)\n",
    "    return X_exp/partition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "57a5402e-be53-4e45-b34b-81c226806126",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[0.0471, 0.0498, 0.0141, 0.2353, 0.6537],\n",
       "         [0.5057, 0.0568, 0.2002, 0.1119, 0.1254]]),\n",
       " tensor([1., 1.]))"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = torch.normal(0, 1, (2, 5))\n",
    "X_prob = softmax(X)\n",
    "X_prob, X_prob.sum(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d20cd427-d180-40f1-b790-5c093995ff2f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def net(X):\n",
    "    return softmax(torch.matmul(X.reshape((-1,W.shape[0])),W)+b)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "8458c23d-889e-434b-a54d-5d3e65c43c1c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([0.1000, 0.5000])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y = torch.tensor([0,2])\n",
    "y_hat = torch.tensor([[0.1,0.3,0.6],[0.3,0.2,0.5]])\n",
    "y_hat[[0,1],y]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2ee39135-a3ce-4c53-944a-b48ee606e7d8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([2.3026, 0.6931])"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def cross_entorpy(y_hat,y):\n",
    "    return -torch.log(y_hat[range(len(y_hat)),y])\n",
    "\n",
    "cross_entorpy(y_hat,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "0d5c49f7-c1df-4d4f-a104-78955f002a6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def accuracy(y_hat,y):\n",
    "    if len(y_hat.shape) >1 and y_hat.shape[1]>1:\n",
    "        y_hat = y_hat.argmax(axis=1)\n",
    "    cmp = y_hat.type(y.dtype) == y\n",
    "    return float(cmp.type(y.dtype).sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "5fad3eb8-b62b-4f04-aa6f-c2a90d05232a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[0.1000, 0.3000, 0.6000],\n",
       "         [0.3000, 0.2000, 0.5000]]),\n",
       " tensor([2, 2]))"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_hat,y_hat.argmax(axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "e5c5e3e0-cfce-47cf-9a3a-a9a4810adfef",
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_accuracy(net,data_iter):\n",
    "    if isinstance(net,torch.nn.Module):\n",
    "        net.eval()\n",
    "    metric = Accummulator(2)\n",
    "    for X,y in data_iter:\n",
    "        metric.add(accuracy(net(X),y),y.numel()\n",
    "    return metric[0] / metric[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "b0fb0e70-8fc2-4602-9ea7-09c05e8988a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Accumulator:\n",
    "    def __init__(self,n):\n",
    "        self.data = [0.0] * n\n",
    "    def add(self,*args):\n",
    "        self.data = [a + float(b) for a,b in zip[self.data,args]]\n",
    "    def reset(self):\n",
    "        self.data = [0.0] * len(self.data)\n",
    "    def __getitem__(self, idx):\n",
    "        return self.data[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "f75b0713-db03-48c3-95ba-7e105cb2e20d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'None'"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f'{evaluate_accuracy(net, test_iter)}'"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "train",
   "language": "python",
   "name": "train"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}