verify_nvidia.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "GPU 서버에 옮겨서 돌리면 될듯"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import numpy as np\n",
    "import os\n",
    "from SAM import SAM\n",
    "from tqdm import tqdm\n",
    "\n",
    "\n",
    "LR = 0.01\n",
    "RHO=0.1\n",
    "WEIGHT_DECAY = 0.0\n",
    "MOMENTUM = 0.9\n",
    "IS_LAZY = True\n",
    "\n",
    "SEED = 0\n",
    "\n",
    "if IS_LAZY:\n",
    "    os.environ['PT_HPU_LAZY_MODE'] = '1'\n",
    "else:\n",
    "    os.environ['PT_HPU_LAZY_MODE'] = '0'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 50000/50000 [02:15<00:00, 368.62it/s, loss=1.11e-6] \n"
     ]
    }
   ],
   "source": [
    "torch.manual_seed(SEED)          # CPU 시드 고정\n",
    "torch.cuda.manual_seed(SEED)     # 현재 GPU의 시드 고정\n",
    "torch.cuda.manual_seed_all(SEED) # 모든 GPU의 시드 고정\n",
    "np.random.seed(SEED)\n",
    "device = torch.device(\"cuda:7\")\n",
    "\n",
    "class MinimalModel(nn.Module):\n",
    "    linear1: nn.Module\n",
    "    linear2: nn.Module\n",
    "\n",
    "    def __init__(self):\n",
    "        super(MinimalModel, self).__init__()\n",
    "        self.linear1: nn.Module = nn.Linear(in_features=2, out_features=1, bias=False)\n",
    "        self.linear2: nn.Module = nn.Linear(in_features=1, out_features=1, bias=False)\n",
    "        self.activation = nn.Tanh()\n",
    "    \n",
    "    def forward(self, x):\n",
    "        out1 = self.linear1(x)\n",
    "        out1 = self.activation(out1)\n",
    "        out2 = self.linear2(out1)\n",
    "        out2 = out2.flatten()\n",
    "        return out2\n",
    "    \n",
    "\n",
    "model = MinimalModel().to(device)\n",
    "if not IS_LAZY:\n",
    "    model = torch.compile(model)\n",
    "\n",
    "loss_func = nn.MSELoss()\n",
    "\n",
    "X = np.array([[1.05, 0.95], [0.95, 1.05]])\n",
    "y = np.array([1.0, -1.0])\n",
    "epoch_nums = 50000\n",
    "\n",
    "X = torch.tensor(X).to(device, dtype=torch.float32)\n",
    "y = torch.tensor(y).to(device, dtype=torch.float32)\n",
    "\n",
    "optimizer = SAM(model.parameters(), optim.SGD, lr=LR, rho=RHO, \n",
    "                weight_decay=WEIGHT_DECAY, momentum=MOMENTUM)\n",
    "\n",
    "loss_list = []\n",
    "state_dict_list = []\n",
    "pbar = tqdm(range(epoch_nums))\n",
    "\n",
    "# Full batch SAM\n",
    "for epoch in pbar:\n",
    "    optimizer.zero_grad()\n",
    "    out = model(X)\n",
    "    loss = loss_func(out, y)\n",
    "    loss_list.append(loss.item())\n",
    "    pbar.set_postfix({'loss': loss.item()})\n",
    "    detached_state_dict = {}\n",
    "    for name, param in model.state_dict().items():\n",
    "        detached_state_dict[name] = param.detach().cpu().numpy()\n",
    "    state_dict_list.append(detached_state_dict)\n",
    "    loss.backward()\n",
    "    #htcore.mark_step()\n",
    "    optimizer.first_step(zero_grad=True)\n",
    "    #htcore.mark_step()\n",
    "    out2 = model(X)\n",
    "    second_loss = loss_func(out2, y)\n",
    "    second_loss.backward()\n",
    "    #htcore.mark_step()\n",
    "    optimizer.second_step(zero_grad=True)\n",
    "    #htcore.mark_step()\n",
    "\n",
    "import pickle as pkl\n",
    "to_save = [loss_list, state_dict_list]\n",
    "if IS_LAZY:\n",
    "    with open(\"verify_lazy_nvidia.pkl\", 'wb') as f:\n",
    "        pkl.dump(to_save, f, protocol=pkl.HIGHEST_PROTOCOL)\n",
    "else:\n",
    "    with open(\"verify_eager_nvidia.pkl\", 'wb') as f:\n",
    "        pkl.dump(to_save, f, protocol=pkl.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([ 1., -1.], device='cuda:7', dtype=torch.float64)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle as pkl\n",
    "\n",
    "data_eager = pkl.load(open(\"verify_eager.pkl\", 'rb'))\n",
    "data_lazy = pkl.load(open(\"verify_lazy.pkl\", 'rb'))\n",
    "nvidia_eager = pkl.load(open(\"verify_eager_nvidia.pkl\", 'rb'))\n",
    "nvidia_lazy = pkl.load(open(\"verify_lazy_nvidia.pkl\", 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "time_step = np.arange(13000, 18000)\n",
    "\n",
    "plt.plot(time_step, np.array(nvidia_eager[0])[time_step], label='nvidia eager')\n",
    "plt.plot(time_step, np.array(nvidia_lazy[0])[time_step], label='nvidia lazy')\n",
    "plt.plot(time_step, np.array(data_eager[0])[time_step], label='eager')\n",
    "#plt.plot(time_step, np.array(data_lazy[0])[time_step], label='lazy')\n",
    "plt.legend()\n",
    "plt.yscale('log')\n",
    "plt.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7.131077950361941e-10\n",
      "6.920863881987316e-11\n",
      "1.3509016127954965e-10\n",
      "1.3509016127954965e-10\n"
     ]
    }
   ],
   "source": [
    "print(min(data_eager[0]))\n",
    "print(min(data_lazy[0]))\n",
    "print(min(nvidia_eager[0]))\n",
    "print(min(nvidia_lazy[0]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3.85493376e-01, 4.83340210e+00, 3.45869658e-01, 3.35918125e+00,\n",
       "       5.99798969e-01, 9.37779659e-01, 2.61955645e+00, 3.22633277e-01,\n",
       "       5.79039033e+01, 1.15897366e-01, 1.01129601e+00, 4.84232795e-02,\n",
       "       8.11957356e-02, 5.91891353e-01, 3.69462902e-02, 2.55560854e-02,\n",
       "       1.57984094e-02, 9.87660679e+01, 3.25781478e-03, 1.70455229e-02,\n",
       "       2.54035617e-03, 2.83490942e-02, 1.13061313e+01, 2.32980568e-02,\n",
       "       2.75327718e-01, 1.97476124e-02, 1.57364311e+01, 1.77988671e+00,\n",
       "       1.78397494e-01, 3.26085448e-01, 3.60809434e-01, 8.31429790e-01,\n",
       "       1.32991726e+02, 9.66142984e-01, 7.87858953e-03, 5.35809913e+01,\n",
       "       2.18136705e+01, 5.09963854e-01, 1.29146388e+00, 1.87545734e-01,\n",
       "       1.94696954e-02, 1.01366057e+01, 2.55392834e+01, 6.65240943e-02,\n",
       "       9.50214123e-02, 3.02428221e+01, 1.08434307e+00, 7.13790593e+00,\n",
       "       3.39168784e+01, 6.18041313e-02, 2.86136605e+01, 3.54388196e+00,\n",
       "       1.12288886e+00, 4.14131700e-01, 5.21448624e-01, 9.26148125e-02,\n",
       "       5.61698836e-01, 3.58184223e+00, 4.31392535e-02, 1.00911515e+01,\n",
       "       3.25201063e+00, 8.84699705e-01, 3.58820913e-01, 4.69080730e-01,\n",
       "       1.84127069e-02, 1.26237043e+00, 2.47968983e+01, 2.27934383e+00,\n",
       "       2.42353374e+01, 2.03887790e+00, 5.20478815e-01, 2.54663314e+00,\n",
       "       4.12101880e+00, 2.48375184e+02, 1.49067274e+00, 4.26179655e-01,\n",
       "       6.56498141e-02, 5.71567479e-01, 4.51159554e-03, 4.63257821e+01,\n",
       "       8.23753611e+01, 1.89317353e-01, 6.00224589e-01, 5.29351708e-01,\n",
       "       2.79737695e-01, 1.39223870e+00, 8.15337780e-02, 1.36871533e-01,\n",
       "       2.99769174e-01, 5.68071727e-01, 9.73986821e-02, 2.02574601e+00,\n",
       "       6.52166008e-01, 6.14694622e-02, 3.64814082e-01, 4.14812079e-01,\n",
       "       2.13276639e-01, 4.97633477e-03, 8.01060683e+00, 1.74653707e-01,\n",
       "       2.60190644e-02, 8.50482635e-01, 6.38392543e+00, 1.27157445e-01,\n",
       "       1.50879286e+01, 3.12693343e+00, 8.90170958e-01, 7.91557600e-02,\n",
       "       6.21857640e-01, 4.14907096e-03, 2.98015068e+00, 2.73068774e+00,\n",
       "       1.47829459e-01, 6.64760855e+00, 3.45025814e+00, 1.08537374e+00,\n",
       "       5.20749123e-02, 9.47201173e+00, 7.99823322e+00, 4.79762281e+01,\n",
       "       1.83990183e+00, 1.06140510e-01, 2.25357410e+00, 4.62169707e+00,\n",
       "       4.06244429e-02, 2.00241660e+01, 4.31782247e+00, 1.71929942e+00,\n",
       "       6.02441067e-01, 5.52293544e+00, 5.51281097e+00, 3.08106465e+01,\n",
       "       1.05573052e+00, 3.03882726e-01, 3.40400610e+00, 4.61705971e+00,\n",
       "       2.74615015e-02, 1.87375501e+01, 6.22889957e+00, 2.80366213e+00,\n",
       "       2.20629859e+00, 1.36247173e+00, 2.44247799e+00, 1.14285651e+01,\n",
       "       7.28829119e-02, 1.28153085e+00, 7.94891106e+00, 1.32245863e+01,\n",
       "       1.06197857e+00, 6.83233592e+00, 2.81616307e+00, 1.24643755e+00,\n",
       "       3.80182231e-01, 6.22042753e+00, 5.47468053e+00, 3.61089273e+01,\n",
       "       4.32098659e-01, 6.14424389e-01, 4.46186735e+00, 1.52624695e+01,\n",
       "       7.67199625e-01, 2.41746447e+01, 2.64233894e+00, 1.19455708e+00,\n",
       "       3.69301731e-01, 1.19687688e+01, 8.19583978e+00, 1.98710277e+02,\n",
       "       3.41160673e-01, 4.35056249e-01, 2.68553185e+00, 7.18102112e+01,\n",
       "       1.31147312e+00, 3.64692380e+02, 2.58061966e+01, 2.43478512e+02,\n",
       "       3.16146443e+00, 1.61572209e+02, 1.86860766e+01, 1.20718790e+03,\n",
       "       3.21900307e-01, 7.85495184e-01, 1.36828824e+01, 2.39394605e+00,\n",
       "       1.46743312e-01, 7.29631317e-01, 7.17838454e+01, 1.16268938e+00,\n",
       "       8.68185933e-02, 2.86826403e-01, 3.96072063e+00, 2.81469277e+00,\n",
       "       1.09839365e-01, 4.27155607e-01, 4.49957053e+00, 5.43365073e+00,\n",
       "       2.73187778e-01, 1.21615998e+00, 4.84351835e+01, 5.10532118e+00,\n",
       "       1.44197278e+00, 4.53731124e+00, 9.17304900e+00, 6.45240148e+01,\n",
       "       1.65404998e+00, 1.58150378e+01, 5.21730507e+00, 3.69204714e+00,\n",
       "       5.21299241e+01, 1.24528926e+00, 2.90960995e+00, 3.39320319e+01,\n",
       "       2.60730304e-01, 3.35653210e-01, 1.33727069e+00, 1.14486912e+01,\n",
       "       2.24408844e-01, 3.00242686e-01, 1.03708511e+00, 5.01935330e+00,\n",
       "       3.47410058e-01, 3.26782270e-01, 9.93307262e-01, 3.04964941e+00,\n",
       "       8.27294891e-01, 7.45105438e-01, 3.59267055e+00, 4.76603799e+01,\n",
       "       8.01086439e-01, 4.49457586e+00, 1.33341435e+01, 8.07739312e+00,\n",
       "       7.33589540e+00, 6.82187276e+00, 9.91198581e+01, 6.29443547e+00,\n",
       "       1.35576680e+00, 7.78845134e+00, 2.46520815e+01, 1.01691915e+01,\n",
       "       1.87655105e+00, 7.11077267e+00, 3.15893229e+01, 7.41415435e+00,\n",
       "       1.14026091e+00, 1.71375082e+01, 4.05927032e+01, 1.49293672e+01,\n",
       "       5.79925211e+00, 2.77999205e+00, 8.30494752e+00, 4.25548832e+01,\n",
       "       3.53453662e+00, 9.39375408e+00, 7.23273867e+01, 6.59864232e+00,\n",
       "       1.22737492e+00, 1.25643499e+01, 3.59597287e+01, 9.88406881e+00,\n",
       "       1.99501309e+00, 6.47712057e+00, 2.23160391e+01, 1.11603171e+01,\n",
       "       1.83430773e+00, 9.41777196e+00, 4.69603142e+01, 6.19806781e+00,\n",
       "       8.65025931e-01, 7.33818079e+01, 3.84636161e+02, 5.64135055e+00,\n",
       "       1.50208795e+00, 4.73111193e+00, 1.06110387e+01, 3.91156444e+01,\n",
       "       2.29806684e+01, 1.67635237e+00, 4.31091783e+00, 3.93249306e+02,\n",
       "       2.50882722e-01, 4.34497790e-01, 2.85481522e+00, 2.35910276e+01,\n",
       "       6.09473195e-01, 6.64203801e+00, 7.02900038e+00, 3.86532800e+00,\n",
       "       1.27167948e+01, 8.86516258e-01, 1.44981664e+00, 2.69308658e+00,\n",
       "       4.05503314e+00, 3.06802670e+00, 2.03865133e+02, 2.18233162e+00,\n",
       "       2.13857382e-01, 1.61599636e+00, 1.30723971e+01, 3.34973361e+00,\n",
       "       2.32081581e+00, 1.70281262e+00, 2.51850675e+00, 7.46148284e+00,\n",
       "       8.43089055e-01, 9.85882423e-01, 9.80606920e+00, 5.17670041e+00,\n",
       "       3.41814915e-01, 4.98441670e+00, 5.36743953e+00, 2.13004592e+00,\n",
       "       1.15428120e+00, 2.42225323e+00, 3.01616669e+00, 9.68035502e+00,\n",
       "       6.14801020e-01, 1.04884643e+00, 1.08573619e+01, 4.95049914e+00,\n",
       "       3.59518641e-01, 3.24810915e+00, 6.75196541e+00, 2.50201649e+00,\n",
       "       1.59530847e+00, 1.82611400e+00, 2.43412440e+00, 6.08508114e+00,\n",
       "       1.18431899e+00, 1.29065942e+00, 1.97467218e+01, 3.15563039e+00,\n",
       "       2.29850615e-01, 2.35447771e+00, 1.11880162e+01, 3.52550108e+00,\n",
       "       3.42864794e+00, 1.27496591e+00, 1.93386596e+00, 4.38293368e+00,\n",
       "       1.28657622e+00, 1.70334888e+00, 2.98646600e+01, 3.03054507e+00,\n",
       "       2.61659859e-01, 1.86118344e+00, 1.20323978e+01, 3.51975277e+00,\n",
       "       3.32452947e+00, 1.35605142e+00, 1.99433374e+00, 4.42158882e+00,\n",
       "       1.79818538e+00, 1.62404854e+00, 3.45309807e+01, 2.56906188e+00,\n",
       "       1.96268902e-01, 1.85757727e+00, 1.65946625e+01, 4.51671196e+00,\n",
       "       6.96599745e+00, 1.00965821e+00, 1.60396947e+00, 3.14017849e+00,\n",
       "       2.50432392e+00, 2.61955711e+00, 1.33820244e+02, 2.08964654e+00,\n",
       "       1.92685447e-01, 1.05309285e+00, 5.99738092e+01, 8.52463432e+00,\n",
       "       2.64188272e+02, 1.88202084e+00, 6.74192281e+00, 4.04300479e+01,\n",
       "       5.30392189e+00, 3.81415594e+00, 1.84780118e+01, 7.76640321e+00,\n",
       "       9.75648642e-01, 2.26827186e+01, 1.85539962e+02, 4.24777629e+00,\n",
       "       6.65432540e-01, 1.91785314e+02, 3.42142382e+00, 2.38381660e+00,\n",
       "       4.72838662e+00, 8.60881460e-01, 1.14567356e+00, 1.49707350e+00,\n",
       "       1.04711447e+01, 4.07549585e-01, 5.12084949e-01, 4.45925559e-01,\n",
       "       1.66555378e-01, 2.20504914e+01, 3.67443952e+00, 4.22238477e+00,\n",
       "       1.60419149e+01, 7.75011974e+00, 6.54700307e+01, 5.35045217e+01,\n",
       "       2.21889327e+00, 5.97843019e+01, 5.80288248e+00, 5.22285493e+00,\n",
       "       1.39840634e+01, 7.16633974e+00, 8.35548902e+01, 1.23360973e+01,\n",
       "       1.04541860e+01, 2.16293131e+00, 6.37357131e+00, 7.79899471e+01,\n",
       "       4.17062273e+00, 1.35601755e+01, 8.62485529e+01, 1.08252604e+01,\n",
       "       4.45101887e+00, 2.62844937e+00, 6.67581628e+00, 1.83289824e+02,\n",
       "       5.50098809e+01, 2.04064832e+00, 7.54294901e+00, 3.47309723e+01,\n",
       "       4.01363908e+00, 5.27181127e+00, 2.45492145e+01, 9.02555922e+00,\n",
       "       1.34741712e+00, 2.29473593e+01, 2.15345590e+02, 4.25001837e+00,\n",
       "       6.20091681e-01, 1.29601548e+02, 5.24890271e+02, 4.88041564e+00,\n",
       "       9.15983671e-01, 4.43352106e+01, 2.82153099e+02, 4.99158257e+00,\n",
       "       8.90324708e-01, 1.68647792e+01, 4.92066373e+01, 8.71106565e+00,\n",
       "       1.84637429e+00, 7.24092469e+00, 2.36818450e+01, 1.09054757e+01,\n",
       "       1.93571532e+00, 6.40667758e+00, 2.06326275e+01, 1.38481128e+01,\n",
       "       2.78151298e+00, 5.26990715e+00, 1.88122317e+01, 1.20461785e+01,\n",
       "       1.97601419e+00, 6.79451410e+00, 2.38044738e+01, 1.15962974e+01,\n",
       "       2.16983926e+00, 7.08132520e+00, 2.75075235e+01, 8.83278038e+00,\n",
       "       1.38413499e+00, 1.04652663e+01, 3.98042415e+01, 8.37233888e+00,\n",
       "       1.53141227e+00, 1.15242992e+01, 4.81498542e+01, 6.68872126e+00,\n",
       "       1.03845447e+00, 1.69143354e+01, 6.86935718e+01, 6.72751697e+00,\n",
       "       1.21763571e+00, 1.88038810e+01, 8.52632329e+01, 5.48560107e+00,\n",
       "       8.20112022e-01, 3.58219900e+01, 1.70877787e+02, 5.19498951e+00,\n",
       "       8.87996750e-01, 8.37297192e+01, 1.32502193e+03, 4.40772779e+00,\n",
       "       7.61015743e-01, 2.99810384e+01, 1.30319434e+02, 5.36796761e+00,\n",
       "       8.90341814e-01, 1.03396872e+02, 5.49480171e+04, 4.63328215e+00])"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(data_eager[0][17500:18000]) / np.array(data_lazy[0][17500:18000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}