Files
Python/d2l/d2l-zh/pytorch/chapter_optimization/adam.ipynb
T
2025-12-16 09:23:53 +08:00

2325 lines
90 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
{
"cells": [
{
"cell_type": "markdown",
"id": "efd11d18",
"metadata": {
"origin_pos": 0
},
"source": [
"# Adam算法\n",
":label:`sec_adam`\n",
"\n",
"本章我们已经学习了许多有效优化的技术。\n",
"在本节讨论之前,我们先详细回顾一下这些技术:\n",
"\n",
"* 在 :numref:`sec_sgd`中,我们学习了:随机梯度下降在解决优化问题时比梯度下降更有效。\n",
"* 在 :numref:`sec_minibatch_sgd`中,我们学习了:在一个小批量中使用更大的观测值集,可以通过向量化提供额外效率。这是高效的多机、多GPU和整体并行处理的关键。\n",
"* 在 :numref:`sec_momentum`中我们添加了一种机制,用于汇总过去梯度的历史以加速收敛。\n",
"* 在 :numref:`sec_adagrad`中,我们通过对每个坐标缩放来实现高效计算的预处理器。\n",
"* 在 :numref:`sec_rmsprop`中,我们通过学习率的调整来分离每个坐标的缩放。\n",
"\n",
"Adam算法 :cite:`Kingma.Ba.2014`将所有这些技术汇总到一个高效的学习算法中。\n",
"不出预料,作为深度学习中使用的更强大和有效的优化算法之一,它非常受欢迎。\n",
"但是它并非没有问题,尤其是 :cite:`Reddi.Kale.Kumar.2019`表明,有时Adam算法可能由于方差控制不良而发散。\n",
"在完善工作中, :cite:`Zaheer.Reddi.Sachan.ea.2018`给Adam算法提供了一个称为Yogi的热补丁来解决这些问题。\n",
"下面我们了解一下Adam算法。\n",
"\n",
"## 算法\n",
"\n",
"Adam算法的关键组成部分之一是:它使用指数加权移动平均值来估算梯度的动量和二次矩,即它使用状态变量\n",
"\n",
"$$\\begin{aligned}\n",
" \\mathbf{v}_t & \\leftarrow \\beta_1 \\mathbf{v}_{t-1} + (1 - \\beta_1) \\mathbf{g}_t, \\\\\n",
" \\mathbf{s}_t & \\leftarrow \\beta_2 \\mathbf{s}_{t-1} + (1 - \\beta_2) \\mathbf{g}_t^2.\n",
"\\end{aligned}$$\n",
"\n",
"这里$\\beta_1$和$\\beta_2$是非负加权参数。\n",
"常将它们设置为$\\beta_1 = 0.9$和$\\beta_2 = 0.999$。\n",
"也就是说,方差估计的移动远远慢于动量估计的移动。\n",
"注意,如果我们初始化$\\mathbf{v}_0 = \\mathbf{s}_0 = 0$,就会获得一个相当大的初始偏差。\n",
"我们可以通过使用$\\sum_{i=0}^t \\beta^i = \\frac{1 - \\beta^t}{1 - \\beta}$来解决这个问题。\n",
"相应地,标准化状态变量由下式获得\n",
"\n",
"$$\\hat{\\mathbf{v}}_t = \\frac{\\mathbf{v}_t}{1 - \\beta_1^t} \\text{ and } \\hat{\\mathbf{s}}_t = \\frac{\\mathbf{s}_t}{1 - \\beta_2^t}.$$\n",
"\n",
"有了正确的估计,我们现在可以写出更新方程。\n",
"首先,我们以非常类似于RMSProp算法的方式重新缩放梯度以获得\n",
"\n",
"$$\\mathbf{g}_t' = \\frac{\\eta \\hat{\\mathbf{v}}_t}{\\sqrt{\\hat{\\mathbf{s}}_t} + \\epsilon}.$$\n",
"\n",
"与RMSProp不同,我们的更新使用动量$\\hat{\\mathbf{v}}_t$而不是梯度本身。\n",
"此外,由于使用$\\frac{1}{\\sqrt{\\hat{\\mathbf{s}}_t} + \\epsilon}$而不是$\\frac{1}{\\sqrt{\\hat{\\mathbf{s}}_t + \\epsilon}}$进行缩放,两者会略有差异。\n",
"前者在实践中效果略好一些,因此与RMSProp算法有所区分。\n",
"通常,我们选择$\\epsilon = 10^{-6}$,这是为了在数值稳定性和逼真度之间取得良好的平衡。\n",
"\n",
"最后,我们简单更新:\n",
"\n",
"$$\\mathbf{x}_t \\leftarrow \\mathbf{x}_{t-1} - \\mathbf{g}_t'.$$\n",
"\n",
"回顾Adam算法,它的设计灵感很清楚:\n",
"首先,动量和规模在状态变量中清晰可见,\n",
"它们相当独特的定义使我们移除偏项(这可以通过稍微不同的初始化和更新条件来修正)。\n",
"其次,RMSProp算法中两项的组合都非常简单。\n",
"最后,明确的学习率$\\eta$使我们能够控制步长来解决收敛问题。\n",
"\n",
"## 实现\n",
"\n",
"从头开始实现Adam算法并不难。\n",
"为方便起见,我们将时间步$t$存储在`hyperparams`字典中。\n",
"除此之外,一切都很简单。\n"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "5a412a5c",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T07:06:58.408547Z",
"iopub.status.busy": "2023-08-18T07:06:58.407890Z",
"iopub.status.idle": "2023-08-18T07:07:00.371669Z",
"shell.execute_reply": "2023-08-18T07:07:00.370790Z"
},
"origin_pos": 2,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"%matplotlib inline\n",
"import torch\n",
"from d2l import torch as d2l\n",
"\n",
"\n",
"def init_adam_states(feature_dim):\n",
" v_w, v_b = torch.zeros((feature_dim, 1)), torch.zeros(1)\n",
" s_w, s_b = torch.zeros((feature_dim, 1)), torch.zeros(1)\n",
" return ((v_w, s_w), (v_b, s_b))\n",
"\n",
"def adam(params, states, hyperparams):\n",
" beta1, beta2, eps = 0.9, 0.999, 1e-6\n",
" for p, (v, s) in zip(params, states):\n",
" with torch.no_grad():\n",
" v[:] = beta1 * v + (1 - beta1) * p.grad\n",
" s[:] = beta2 * s + (1 - beta2) * torch.square(p.grad)\n",
" v_bias_corr = v / (1 - beta1 ** hyperparams['t'])\n",
" s_bias_corr = s / (1 - beta2 ** hyperparams['t'])\n",
" p[:] -= hyperparams['lr'] * v_bias_corr / (torch.sqrt(s_bias_corr)\n",
" + eps)\n",
" p.grad.data.zero_()\n",
" hyperparams['t'] += 1"
]
},
{
"cell_type": "markdown",
"id": "4c28a0af",
"metadata": {
"origin_pos": 5
},
"source": [
"现在,我们用以上Adam算法来训练模型,这里我们使用$\\eta = 0.01$的学习率。\n"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "959e8f03",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T07:07:00.376075Z",
"iopub.status.busy": "2023-08-18T07:07:00.375389Z",
"iopub.status.idle": "2023-08-18T07:07:03.158768Z",
"shell.execute_reply": "2023-08-18T07:07:03.157913Z"
},
"origin_pos": 6,
"tab": [
"pytorch"
]
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"loss: 0.244, 0.015 sec/epoch\n"
]
},
{
"data": {
"image/svg+xml": [
"<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n",
"<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
" \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
"<svg xmlns:xlink=\"http://www.w3.org/1999/xlink\" width=\"266.957813pt\" height=\"184.455469pt\" viewBox=\"0 0 266.957813 184.455469\" xmlns=\"http://www.w3.org/2000/svg\" version=\"1.1\">\n",
" <metadata>\n",
" <rdf:RDF xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:cc=\"http://creativecommons.org/ns#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">\n",
" <cc:Work>\n",
" <dc:type rdf:resource=\"http://purl.org/dc/dcmitype/StillImage\"/>\n",
" <dc:date>2023-08-18T07:07:03.123695</dc:date>\n",
" <dc:format>image/svg+xml</dc:format>\n",
" <dc:creator>\n",
" <cc:Agent>\n",
" <dc:title>Matplotlib v3.5.1, https://matplotlib.org/</dc:title>\n",
" </cc:Agent>\n",
" </dc:creator>\n",
" </cc:Work>\n",
" </rdf:RDF>\n",
" </metadata>\n",
" <defs>\n",
" <style type=\"text/css\">*{stroke-linejoin: round; stroke-linecap: butt}</style>\n",
" </defs>\n",
" <g id=\"figure_1\">\n",
" <g id=\"patch_1\">\n",
" <path d=\"M -0 184.455469 \n",
"L 266.957813 184.455469 \n",
"L 266.957813 0 \n",
"L -0 0 \n",
"L -0 184.455469 \n",
"z\n",
"\" style=\"fill: none\"/>\n",
" </g>\n",
" <g id=\"axes_1\">\n",
" <g id=\"patch_2\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"L 56.50625 10.999219 \n",
"z\n",
"\" style=\"fill: #ffffff\"/>\n",
" </g>\n",
" <g id=\"matplotlib.axis_1\">\n",
" <g id=\"xtick_1\">\n",
" <g id=\"line2d_1\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_2\">\n",
" <defs>\n",
" <path id=\"m7dc2fc49b2\" d=\"M 0 0 \n",
"L 0 3.5 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#m7dc2fc49b2\" x=\"56.50625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_1\">\n",
" <!-- 0.0 -->\n",
" <g transform=\"translate(48.554688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-30\" d=\"M 2034 4250 \n",
"Q 1547 4250 1301 3770 \n",
"Q 1056 3291 1056 2328 \n",
"Q 1056 1369 1301 889 \n",
"Q 1547 409 2034 409 \n",
"Q 2525 409 2770 889 \n",
"Q 3016 1369 3016 2328 \n",
"Q 3016 3291 2770 3770 \n",
"Q 2525 4250 2034 4250 \n",
"z\n",
"M 2034 4750 \n",
"Q 2819 4750 3233 4129 \n",
"Q 3647 3509 3647 2328 \n",
"Q 3647 1150 3233 529 \n",
"Q 2819 -91 2034 -91 \n",
"Q 1250 -91 836 529 \n",
"Q 422 1150 422 2328 \n",
"Q 422 3509 836 4129 \n",
"Q 1250 4750 2034 4750 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-2e\" d=\"M 684 794 \n",
"L 1344 794 \n",
"L 1344 0 \n",
"L 684 0 \n",
"L 684 794 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_2\">\n",
" <g id=\"line2d_3\">\n",
" <path d=\"M 105.33125 146.899219 \n",
"L 105.33125 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_4\">\n",
" <g>\n",
" <use xlink:href=\"#m7dc2fc49b2\" x=\"105.33125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_2\">\n",
" <!-- 0.5 -->\n",
" <g transform=\"translate(97.379688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-35\" d=\"M 691 4666 \n",
"L 3169 4666 \n",
"L 3169 4134 \n",
"L 1269 4134 \n",
"L 1269 2991 \n",
"Q 1406 3038 1543 3061 \n",
"Q 1681 3084 1819 3084 \n",
"Q 2600 3084 3056 2656 \n",
"Q 3513 2228 3513 1497 \n",
"Q 3513 744 3044 326 \n",
"Q 2575 -91 1722 -91 \n",
"Q 1428 -91 1123 -41 \n",
"Q 819 9 494 109 \n",
"L 494 744 \n",
"Q 775 591 1075 516 \n",
"Q 1375 441 1709 441 \n",
"Q 2250 441 2565 725 \n",
"Q 2881 1009 2881 1497 \n",
"Q 2881 1984 2565 2268 \n",
"Q 2250 2553 1709 2553 \n",
"Q 1456 2553 1204 2497 \n",
"Q 953 2441 691 2322 \n",
"L 691 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_3\">\n",
" <g id=\"line2d_5\">\n",
" <path d=\"M 154.15625 146.899219 \n",
"L 154.15625 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_6\">\n",
" <g>\n",
" <use xlink:href=\"#m7dc2fc49b2\" x=\"154.15625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_3\">\n",
" <!-- 1.0 -->\n",
" <g transform=\"translate(146.204688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-31\" d=\"M 794 531 \n",
"L 1825 531 \n",
"L 1825 4091 \n",
"L 703 3866 \n",
"L 703 4441 \n",
"L 1819 4666 \n",
"L 2450 4666 \n",
"L 2450 531 \n",
"L 3481 531 \n",
"L 3481 0 \n",
"L 794 0 \n",
"L 794 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-31\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_4\">\n",
" <g id=\"line2d_7\">\n",
" <path d=\"M 202.98125 146.899219 \n",
"L 202.98125 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_8\">\n",
" <g>\n",
" <use xlink:href=\"#m7dc2fc49b2\" x=\"202.98125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_4\">\n",
" <!-- 1.5 -->\n",
" <g transform=\"translate(195.029688 161.497656)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-31\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_5\">\n",
" <g id=\"line2d_9\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_10\">\n",
" <g>\n",
" <use xlink:href=\"#m7dc2fc49b2\" x=\"251.80625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_5\">\n",
" <!-- 2.0 -->\n",
" <g transform=\"translate(243.854688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-32\" d=\"M 1228 531 \n",
"L 3431 531 \n",
"L 3431 0 \n",
"L 469 0 \n",
"L 469 531 \n",
"Q 828 903 1448 1529 \n",
"Q 2069 2156 2228 2338 \n",
"Q 2531 2678 2651 2914 \n",
"Q 2772 3150 2772 3378 \n",
"Q 2772 3750 2511 3984 \n",
"Q 2250 4219 1831 4219 \n",
"Q 1534 4219 1204 4116 \n",
"Q 875 4013 500 3803 \n",
"L 500 4441 \n",
"Q 881 4594 1212 4672 \n",
"Q 1544 4750 1819 4750 \n",
"Q 2544 4750 2975 4387 \n",
"Q 3406 4025 3406 3419 \n",
"Q 3406 3131 3298 2873 \n",
"Q 3191 2616 2906 2266 \n",
"Q 2828 2175 2409 1742 \n",
"Q 1991 1309 1228 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-32\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_6\">\n",
" <!-- epoch -->\n",
" <g transform=\"translate(138.928125 175.175781)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-65\" d=\"M 3597 1894 \n",
"L 3597 1613 \n",
"L 953 1613 \n",
"Q 991 1019 1311 708 \n",
"Q 1631 397 2203 397 \n",
"Q 2534 397 2845 478 \n",
"Q 3156 559 3463 722 \n",
"L 3463 178 \n",
"Q 3153 47 2828 -22 \n",
"Q 2503 -91 2169 -91 \n",
"Q 1331 -91 842 396 \n",
"Q 353 884 353 1716 \n",
"Q 353 2575 817 3079 \n",
"Q 1281 3584 2069 3584 \n",
"Q 2775 3584 3186 3129 \n",
"Q 3597 2675 3597 1894 \n",
"z\n",
"M 3022 2063 \n",
"Q 3016 2534 2758 2815 \n",
"Q 2500 3097 2075 3097 \n",
"Q 1594 3097 1305 2825 \n",
"Q 1016 2553 972 2059 \n",
"L 3022 2063 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-70\" d=\"M 1159 525 \n",
"L 1159 -1331 \n",
"L 581 -1331 \n",
"L 581 3500 \n",
"L 1159 3500 \n",
"L 1159 2969 \n",
"Q 1341 3281 1617 3432 \n",
"Q 1894 3584 2278 3584 \n",
"Q 2916 3584 3314 3078 \n",
"Q 3713 2572 3713 1747 \n",
"Q 3713 922 3314 415 \n",
"Q 2916 -91 2278 -91 \n",
"Q 1894 -91 1617 61 \n",
"Q 1341 213 1159 525 \n",
"z\n",
"M 3116 1747 \n",
"Q 3116 2381 2855 2742 \n",
"Q 2594 3103 2138 3103 \n",
"Q 1681 3103 1420 2742 \n",
"Q 1159 2381 1159 1747 \n",
"Q 1159 1113 1420 752 \n",
"Q 1681 391 2138 391 \n",
"Q 2594 391 2855 752 \n",
"Q 3116 1113 3116 1747 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-6f\" d=\"M 1959 3097 \n",
"Q 1497 3097 1228 2736 \n",
"Q 959 2375 959 1747 \n",
"Q 959 1119 1226 758 \n",
"Q 1494 397 1959 397 \n",
"Q 2419 397 2687 759 \n",
"Q 2956 1122 2956 1747 \n",
"Q 2956 2369 2687 2733 \n",
"Q 2419 3097 1959 3097 \n",
"z\n",
"M 1959 3584 \n",
"Q 2709 3584 3137 3096 \n",
"Q 3566 2609 3566 1747 \n",
"Q 3566 888 3137 398 \n",
"Q 2709 -91 1959 -91 \n",
"Q 1206 -91 779 398 \n",
"Q 353 888 353 1747 \n",
"Q 353 2609 779 3096 \n",
"Q 1206 3584 1959 3584 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-63\" d=\"M 3122 3366 \n",
"L 3122 2828 \n",
"Q 2878 2963 2633 3030 \n",
"Q 2388 3097 2138 3097 \n",
"Q 1578 3097 1268 2742 \n",
"Q 959 2388 959 1747 \n",
"Q 959 1106 1268 751 \n",
"Q 1578 397 2138 397 \n",
"Q 2388 397 2633 464 \n",
"Q 2878 531 3122 666 \n",
"L 3122 134 \n",
"Q 2881 22 2623 -34 \n",
"Q 2366 -91 2075 -91 \n",
"Q 1284 -91 818 406 \n",
"Q 353 903 353 1747 \n",
"Q 353 2603 823 3093 \n",
"Q 1294 3584 2113 3584 \n",
"Q 2378 3584 2631 3529 \n",
"Q 2884 3475 3122 3366 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-68\" d=\"M 3513 2113 \n",
"L 3513 0 \n",
"L 2938 0 \n",
"L 2938 2094 \n",
"Q 2938 2591 2744 2837 \n",
"Q 2550 3084 2163 3084 \n",
"Q 1697 3084 1428 2787 \n",
"Q 1159 2491 1159 1978 \n",
"L 1159 0 \n",
"L 581 0 \n",
"L 581 4863 \n",
"L 1159 4863 \n",
"L 1159 2956 \n",
"Q 1366 3272 1645 3428 \n",
"Q 1925 3584 2291 3584 \n",
"Q 2894 3584 3203 3211 \n",
"Q 3513 2838 3513 2113 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-65\"/>\n",
" <use xlink:href=\"#DejaVuSans-70\" x=\"61.523438\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"125\"/>\n",
" <use xlink:href=\"#DejaVuSans-63\" x=\"186.181641\"/>\n",
" <use xlink:href=\"#DejaVuSans-68\" x=\"241.162109\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"matplotlib.axis_2\">\n",
" <g id=\"ytick_1\">\n",
" <g id=\"line2d_11\">\n",
" <path d=\"M 56.50625 141.672296 \n",
"L 251.80625 141.672296 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_12\">\n",
" <defs>\n",
" <path id=\"mf8fba4ad52\" d=\"M 0 0 \n",
"L -3.5 0 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"141.672296\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_7\">\n",
" <!-- 0.225 -->\n",
" <g transform=\"translate(20.878125 145.471514)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_2\">\n",
" <g id=\"line2d_13\">\n",
" <path d=\"M 56.50625 115.53768 \n",
"L 251.80625 115.53768 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_14\">\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"115.53768\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_8\">\n",
" <!-- 0.250 -->\n",
" <g transform=\"translate(20.878125 119.336899)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_3\">\n",
" <g id=\"line2d_15\">\n",
" <path d=\"M 56.50625 89.403065 \n",
"L 251.80625 89.403065 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_16\">\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"89.403065\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_9\">\n",
" <!-- 0.275 -->\n",
" <g transform=\"translate(20.878125 93.202284)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-37\" d=\"M 525 4666 \n",
"L 3525 4666 \n",
"L 3525 4397 \n",
"L 1831 0 \n",
"L 1172 0 \n",
"L 2766 4134 \n",
"L 525 4134 \n",
"L 525 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-37\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_4\">\n",
" <g id=\"line2d_17\">\n",
" <path d=\"M 56.50625 63.26845 \n",
"L 251.80625 63.26845 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_18\">\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"63.26845\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_10\">\n",
" <!-- 0.300 -->\n",
" <g transform=\"translate(20.878125 67.067668)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-33\" d=\"M 2597 2516 \n",
"Q 3050 2419 3304 2112 \n",
"Q 3559 1806 3559 1356 \n",
"Q 3559 666 3084 287 \n",
"Q 2609 -91 1734 -91 \n",
"Q 1441 -91 1130 -33 \n",
"Q 819 25 488 141 \n",
"L 488 750 \n",
"Q 750 597 1062 519 \n",
"Q 1375 441 1716 441 \n",
"Q 2309 441 2620 675 \n",
"Q 2931 909 2931 1356 \n",
"Q 2931 1769 2642 2001 \n",
"Q 2353 2234 1838 2234 \n",
"L 1294 2234 \n",
"L 1294 2753 \n",
"L 1863 2753 \n",
"Q 2328 2753 2575 2939 \n",
"Q 2822 3125 2822 3475 \n",
"Q 2822 3834 2567 4026 \n",
"Q 2313 4219 1838 4219 \n",
"Q 1578 4219 1281 4162 \n",
"Q 984 4106 628 3988 \n",
"L 628 4550 \n",
"Q 988 4650 1302 4700 \n",
"Q 1616 4750 1894 4750 \n",
"Q 2613 4750 3031 4423 \n",
"Q 3450 4097 3450 3541 \n",
"Q 3450 3153 3228 2886 \n",
"Q 3006 2619 2597 2516 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_5\">\n",
" <g id=\"line2d_19\">\n",
" <path d=\"M 56.50625 37.133834 \n",
"L 251.80625 37.133834 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_20\">\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"37.133834\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_11\">\n",
" <!-- 0.325 -->\n",
" <g transform=\"translate(20.878125 40.933053)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_6\">\n",
" <g id=\"line2d_21\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_22\">\n",
" <g>\n",
" <use xlink:href=\"#mf8fba4ad52\" x=\"56.50625\" y=\"10.999219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_12\">\n",
" <!-- 0.350 -->\n",
" <g transform=\"translate(20.878125 14.798437)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_13\">\n",
" <!-- loss -->\n",
" <g transform=\"translate(14.798438 88.607031)rotate(-90)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-6c\" d=\"M 603 4863 \n",
"L 1178 4863 \n",
"L 1178 0 \n",
"L 603 0 \n",
"L 603 4863 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-73\" d=\"M 2834 3397 \n",
"L 2834 2853 \n",
"Q 2591 2978 2328 3040 \n",
"Q 2066 3103 1784 3103 \n",
"Q 1356 3103 1142 2972 \n",
"Q 928 2841 928 2578 \n",
"Q 928 2378 1081 2264 \n",
"Q 1234 2150 1697 2047 \n",
"L 1894 2003 \n",
"Q 2506 1872 2764 1633 \n",
"Q 3022 1394 3022 966 \n",
"Q 3022 478 2636 193 \n",
"Q 2250 -91 1575 -91 \n",
"Q 1294 -91 989 -36 \n",
"Q 684 19 347 128 \n",
"L 347 722 \n",
"Q 666 556 975 473 \n",
"Q 1284 391 1588 391 \n",
"Q 1994 391 2212 530 \n",
"Q 2431 669 2431 922 \n",
"Q 2431 1156 2273 1281 \n",
"Q 2116 1406 1581 1522 \n",
"L 1381 1569 \n",
"Q 847 1681 609 1914 \n",
"Q 372 2147 372 2553 \n",
"Q 372 3047 722 3315 \n",
"Q 1072 3584 1716 3584 \n",
"Q 2034 3584 2315 3537 \n",
"Q 2597 3491 2834 3397 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-6c\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"27.783203\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"88.964844\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"141.064453\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"line2d_23\">\n",
" <path d=\"M 75.933442 -1 \n",
"L 82.54625 37.194192 \n",
"L 95.56625 81.765788 \n",
"L 108.58625 108.961006 \n",
"L 121.60625 115.915567 \n",
"L 134.62625 118.953393 \n",
"L 147.64625 121.015797 \n",
"L 160.66625 122.422801 \n",
"L 173.68625 121.738152 \n",
"L 186.70625 119.471584 \n",
"L 199.72625 122.077453 \n",
"L 212.74625 122.722963 \n",
"L 225.76625 122.039335 \n",
"L 238.78625 122.261705 \n",
"L 251.80625 122.052467 \n",
"\" clip-path=\"url(#p1d294cb0bf)\" style=\"fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_3\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_4\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_5\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_6\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <defs>\n",
" <clipPath id=\"p1d294cb0bf\">\n",
" <rect x=\"56.50625\" y=\"10.999219\" width=\"195.3\" height=\"135.9\"/>\n",
" </clipPath>\n",
" </defs>\n",
"</svg>\n"
],
"text/plain": [
"<Figure size 252x180 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"data_iter, feature_dim = d2l.get_data_ch11(batch_size=10)\n",
"d2l.train_ch11(adam, init_adam_states(feature_dim),\n",
" {'lr': 0.01, 't': 1}, data_iter, feature_dim);"
]
},
{
"cell_type": "markdown",
"id": "1f3ac5ff",
"metadata": {
"origin_pos": 7
},
"source": [
"此外,我们可以用深度学习框架自带算法应用Adam算法,这里我们只需要传递配置参数。\n"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "bba60ced",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T07:07:03.162734Z",
"iopub.status.busy": "2023-08-18T07:07:03.162173Z",
"iopub.status.idle": "2023-08-18T07:07:08.522679Z",
"shell.execute_reply": "2023-08-18T07:07:08.521534Z"
},
"origin_pos": 9,
"tab": [
"pytorch"
]
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"loss: 0.254, 0.015 sec/epoch\n"
]
},
{
"data": {
"image/svg+xml": [
"<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n",
"<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
" \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
"<svg xmlns:xlink=\"http://www.w3.org/1999/xlink\" width=\"262.1875pt\" height=\"184.455469pt\" viewBox=\"0 0 262.1875 184.455469\" xmlns=\"http://www.w3.org/2000/svg\" version=\"1.1\">\n",
" <metadata>\n",
" <rdf:RDF xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:cc=\"http://creativecommons.org/ns#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">\n",
" <cc:Work>\n",
" <dc:type rdf:resource=\"http://purl.org/dc/dcmitype/StillImage\"/>\n",
" <dc:date>2023-08-18T07:07:08.488032</dc:date>\n",
" <dc:format>image/svg+xml</dc:format>\n",
" <dc:creator>\n",
" <cc:Agent>\n",
" <dc:title>Matplotlib v3.5.1, https://matplotlib.org/</dc:title>\n",
" </cc:Agent>\n",
" </dc:creator>\n",
" </cc:Work>\n",
" </rdf:RDF>\n",
" </metadata>\n",
" <defs>\n",
" <style type=\"text/css\">*{stroke-linejoin: round; stroke-linecap: butt}</style>\n",
" </defs>\n",
" <g id=\"figure_1\">\n",
" <g id=\"patch_1\">\n",
" <path d=\"M -0 184.455469 \n",
"L 262.1875 184.455469 \n",
"L 262.1875 0 \n",
"L -0 0 \n",
"L -0 184.455469 \n",
"z\n",
"\" style=\"fill: none\"/>\n",
" </g>\n",
" <g id=\"axes_1\">\n",
" <g id=\"patch_2\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"L 56.50625 10.999219 \n",
"z\n",
"\" style=\"fill: #ffffff\"/>\n",
" </g>\n",
" <g id=\"matplotlib.axis_1\">\n",
" <g id=\"xtick_1\">\n",
" <g id=\"line2d_1\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_2\">\n",
" <defs>\n",
" <path id=\"m12a7fbcaf4\" d=\"M 0 0 \n",
"L 0 3.5 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#m12a7fbcaf4\" x=\"56.50625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_1\">\n",
" <!-- 0 -->\n",
" <g transform=\"translate(53.325 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-30\" d=\"M 2034 4250 \n",
"Q 1547 4250 1301 3770 \n",
"Q 1056 3291 1056 2328 \n",
"Q 1056 1369 1301 889 \n",
"Q 1547 409 2034 409 \n",
"Q 2525 409 2770 889 \n",
"Q 3016 1369 3016 2328 \n",
"Q 3016 3291 2770 3770 \n",
"Q 2525 4250 2034 4250 \n",
"z\n",
"M 2034 4750 \n",
"Q 2819 4750 3233 4129 \n",
"Q 3647 3509 3647 2328 \n",
"Q 3647 1150 3233 529 \n",
"Q 2819 -91 2034 -91 \n",
"Q 1250 -91 836 529 \n",
"Q 422 1150 422 2328 \n",
"Q 422 3509 836 4129 \n",
"Q 1250 4750 2034 4750 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_2\">\n",
" <g id=\"line2d_3\">\n",
" <path d=\"M 105.33125 146.899219 \n",
"L 105.33125 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_4\">\n",
" <g>\n",
" <use xlink:href=\"#m12a7fbcaf4\" x=\"105.33125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_2\">\n",
" <!-- 1 -->\n",
" <g transform=\"translate(102.15 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-31\" d=\"M 794 531 \n",
"L 1825 531 \n",
"L 1825 4091 \n",
"L 703 3866 \n",
"L 703 4441 \n",
"L 1819 4666 \n",
"L 2450 4666 \n",
"L 2450 531 \n",
"L 3481 531 \n",
"L 3481 0 \n",
"L 794 0 \n",
"L 794 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-31\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_3\">\n",
" <g id=\"line2d_5\">\n",
" <path d=\"M 154.15625 146.899219 \n",
"L 154.15625 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_6\">\n",
" <g>\n",
" <use xlink:href=\"#m12a7fbcaf4\" x=\"154.15625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_3\">\n",
" <!-- 2 -->\n",
" <g transform=\"translate(150.975 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-32\" d=\"M 1228 531 \n",
"L 3431 531 \n",
"L 3431 0 \n",
"L 469 0 \n",
"L 469 531 \n",
"Q 828 903 1448 1529 \n",
"Q 2069 2156 2228 2338 \n",
"Q 2531 2678 2651 2914 \n",
"Q 2772 3150 2772 3378 \n",
"Q 2772 3750 2511 3984 \n",
"Q 2250 4219 1831 4219 \n",
"Q 1534 4219 1204 4116 \n",
"Q 875 4013 500 3803 \n",
"L 500 4441 \n",
"Q 881 4594 1212 4672 \n",
"Q 1544 4750 1819 4750 \n",
"Q 2544 4750 2975 4387 \n",
"Q 3406 4025 3406 3419 \n",
"Q 3406 3131 3298 2873 \n",
"Q 3191 2616 2906 2266 \n",
"Q 2828 2175 2409 1742 \n",
"Q 1991 1309 1228 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-32\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_4\">\n",
" <g id=\"line2d_7\">\n",
" <path d=\"M 202.98125 146.899219 \n",
"L 202.98125 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_8\">\n",
" <g>\n",
" <use xlink:href=\"#m12a7fbcaf4\" x=\"202.98125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_4\">\n",
" <!-- 3 -->\n",
" <g transform=\"translate(199.8 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-33\" d=\"M 2597 2516 \n",
"Q 3050 2419 3304 2112 \n",
"Q 3559 1806 3559 1356 \n",
"Q 3559 666 3084 287 \n",
"Q 2609 -91 1734 -91 \n",
"Q 1441 -91 1130 -33 \n",
"Q 819 25 488 141 \n",
"L 488 750 \n",
"Q 750 597 1062 519 \n",
"Q 1375 441 1716 441 \n",
"Q 2309 441 2620 675 \n",
"Q 2931 909 2931 1356 \n",
"Q 2931 1769 2642 2001 \n",
"Q 2353 2234 1838 2234 \n",
"L 1294 2234 \n",
"L 1294 2753 \n",
"L 1863 2753 \n",
"Q 2328 2753 2575 2939 \n",
"Q 2822 3125 2822 3475 \n",
"Q 2822 3834 2567 4026 \n",
"Q 2313 4219 1838 4219 \n",
"Q 1578 4219 1281 4162 \n",
"Q 984 4106 628 3988 \n",
"L 628 4550 \n",
"Q 988 4650 1302 4700 \n",
"Q 1616 4750 1894 4750 \n",
"Q 2613 4750 3031 4423 \n",
"Q 3450 4097 3450 3541 \n",
"Q 3450 3153 3228 2886 \n",
"Q 3006 2619 2597 2516 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-33\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_5\">\n",
" <g id=\"line2d_9\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_10\">\n",
" <g>\n",
" <use xlink:href=\"#m12a7fbcaf4\" x=\"251.80625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_5\">\n",
" <!-- 4 -->\n",
" <g transform=\"translate(248.625 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-34\" d=\"M 2419 4116 \n",
"L 825 1625 \n",
"L 2419 1625 \n",
"L 2419 4116 \n",
"z\n",
"M 2253 4666 \n",
"L 3047 4666 \n",
"L 3047 1625 \n",
"L 3713 1625 \n",
"L 3713 1100 \n",
"L 3047 1100 \n",
"L 3047 0 \n",
"L 2419 0 \n",
"L 2419 1100 \n",
"L 313 1100 \n",
"L 313 1709 \n",
"L 2253 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-34\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_6\">\n",
" <!-- epoch -->\n",
" <g transform=\"translate(138.928125 175.175781)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-65\" d=\"M 3597 1894 \n",
"L 3597 1613 \n",
"L 953 1613 \n",
"Q 991 1019 1311 708 \n",
"Q 1631 397 2203 397 \n",
"Q 2534 397 2845 478 \n",
"Q 3156 559 3463 722 \n",
"L 3463 178 \n",
"Q 3153 47 2828 -22 \n",
"Q 2503 -91 2169 -91 \n",
"Q 1331 -91 842 396 \n",
"Q 353 884 353 1716 \n",
"Q 353 2575 817 3079 \n",
"Q 1281 3584 2069 3584 \n",
"Q 2775 3584 3186 3129 \n",
"Q 3597 2675 3597 1894 \n",
"z\n",
"M 3022 2063 \n",
"Q 3016 2534 2758 2815 \n",
"Q 2500 3097 2075 3097 \n",
"Q 1594 3097 1305 2825 \n",
"Q 1016 2553 972 2059 \n",
"L 3022 2063 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-70\" d=\"M 1159 525 \n",
"L 1159 -1331 \n",
"L 581 -1331 \n",
"L 581 3500 \n",
"L 1159 3500 \n",
"L 1159 2969 \n",
"Q 1341 3281 1617 3432 \n",
"Q 1894 3584 2278 3584 \n",
"Q 2916 3584 3314 3078 \n",
"Q 3713 2572 3713 1747 \n",
"Q 3713 922 3314 415 \n",
"Q 2916 -91 2278 -91 \n",
"Q 1894 -91 1617 61 \n",
"Q 1341 213 1159 525 \n",
"z\n",
"M 3116 1747 \n",
"Q 3116 2381 2855 2742 \n",
"Q 2594 3103 2138 3103 \n",
"Q 1681 3103 1420 2742 \n",
"Q 1159 2381 1159 1747 \n",
"Q 1159 1113 1420 752 \n",
"Q 1681 391 2138 391 \n",
"Q 2594 391 2855 752 \n",
"Q 3116 1113 3116 1747 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-6f\" d=\"M 1959 3097 \n",
"Q 1497 3097 1228 2736 \n",
"Q 959 2375 959 1747 \n",
"Q 959 1119 1226 758 \n",
"Q 1494 397 1959 397 \n",
"Q 2419 397 2687 759 \n",
"Q 2956 1122 2956 1747 \n",
"Q 2956 2369 2687 2733 \n",
"Q 2419 3097 1959 3097 \n",
"z\n",
"M 1959 3584 \n",
"Q 2709 3584 3137 3096 \n",
"Q 3566 2609 3566 1747 \n",
"Q 3566 888 3137 398 \n",
"Q 2709 -91 1959 -91 \n",
"Q 1206 -91 779 398 \n",
"Q 353 888 353 1747 \n",
"Q 353 2609 779 3096 \n",
"Q 1206 3584 1959 3584 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-63\" d=\"M 3122 3366 \n",
"L 3122 2828 \n",
"Q 2878 2963 2633 3030 \n",
"Q 2388 3097 2138 3097 \n",
"Q 1578 3097 1268 2742 \n",
"Q 959 2388 959 1747 \n",
"Q 959 1106 1268 751 \n",
"Q 1578 397 2138 397 \n",
"Q 2388 397 2633 464 \n",
"Q 2878 531 3122 666 \n",
"L 3122 134 \n",
"Q 2881 22 2623 -34 \n",
"Q 2366 -91 2075 -91 \n",
"Q 1284 -91 818 406 \n",
"Q 353 903 353 1747 \n",
"Q 353 2603 823 3093 \n",
"Q 1294 3584 2113 3584 \n",
"Q 2378 3584 2631 3529 \n",
"Q 2884 3475 3122 3366 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-68\" d=\"M 3513 2113 \n",
"L 3513 0 \n",
"L 2938 0 \n",
"L 2938 2094 \n",
"Q 2938 2591 2744 2837 \n",
"Q 2550 3084 2163 3084 \n",
"Q 1697 3084 1428 2787 \n",
"Q 1159 2491 1159 1978 \n",
"L 1159 0 \n",
"L 581 0 \n",
"L 581 4863 \n",
"L 1159 4863 \n",
"L 1159 2956 \n",
"Q 1366 3272 1645 3428 \n",
"Q 1925 3584 2291 3584 \n",
"Q 2894 3584 3203 3211 \n",
"Q 3513 2838 3513 2113 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-65\"/>\n",
" <use xlink:href=\"#DejaVuSans-70\" x=\"61.523438\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"125\"/>\n",
" <use xlink:href=\"#DejaVuSans-63\" x=\"186.181641\"/>\n",
" <use xlink:href=\"#DejaVuSans-68\" x=\"241.162109\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"matplotlib.axis_2\">\n",
" <g id=\"ytick_1\">\n",
" <g id=\"line2d_11\">\n",
" <path d=\"M 56.50625 141.672296 \n",
"L 251.80625 141.672296 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_12\">\n",
" <defs>\n",
" <path id=\"m27739734b3\" d=\"M 0 0 \n",
"L -3.5 0 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"141.672296\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_7\">\n",
" <!-- 0.225 -->\n",
" <g transform=\"translate(20.878125 145.471514)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-2e\" d=\"M 684 794 \n",
"L 1344 794 \n",
"L 1344 0 \n",
"L 684 0 \n",
"L 684 794 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-35\" d=\"M 691 4666 \n",
"L 3169 4666 \n",
"L 3169 4134 \n",
"L 1269 4134 \n",
"L 1269 2991 \n",
"Q 1406 3038 1543 3061 \n",
"Q 1681 3084 1819 3084 \n",
"Q 2600 3084 3056 2656 \n",
"Q 3513 2228 3513 1497 \n",
"Q 3513 744 3044 326 \n",
"Q 2575 -91 1722 -91 \n",
"Q 1428 -91 1123 -41 \n",
"Q 819 9 494 109 \n",
"L 494 744 \n",
"Q 775 591 1075 516 \n",
"Q 1375 441 1709 441 \n",
"Q 2250 441 2565 725 \n",
"Q 2881 1009 2881 1497 \n",
"Q 2881 1984 2565 2268 \n",
"Q 2250 2553 1709 2553 \n",
"Q 1456 2553 1204 2497 \n",
"Q 953 2441 691 2322 \n",
"L 691 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_2\">\n",
" <g id=\"line2d_13\">\n",
" <path d=\"M 56.50625 115.53768 \n",
"L 251.80625 115.53768 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_14\">\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"115.53768\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_8\">\n",
" <!-- 0.250 -->\n",
" <g transform=\"translate(20.878125 119.336899)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_3\">\n",
" <g id=\"line2d_15\">\n",
" <path d=\"M 56.50625 89.403065 \n",
"L 251.80625 89.403065 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_16\">\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"89.403065\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_9\">\n",
" <!-- 0.275 -->\n",
" <g transform=\"translate(20.878125 93.202284)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-37\" d=\"M 525 4666 \n",
"L 3525 4666 \n",
"L 3525 4397 \n",
"L 1831 0 \n",
"L 1172 0 \n",
"L 2766 4134 \n",
"L 525 4134 \n",
"L 525 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-37\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_4\">\n",
" <g id=\"line2d_17\">\n",
" <path d=\"M 56.50625 63.26845 \n",
"L 251.80625 63.26845 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_18\">\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"63.26845\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_10\">\n",
" <!-- 0.300 -->\n",
" <g transform=\"translate(20.878125 67.067668)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_5\">\n",
" <g id=\"line2d_19\">\n",
" <path d=\"M 56.50625 37.133834 \n",
"L 251.80625 37.133834 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_20\">\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"37.133834\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_11\">\n",
" <!-- 0.325 -->\n",
" <g transform=\"translate(20.878125 40.933053)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_6\">\n",
" <g id=\"line2d_21\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_22\">\n",
" <g>\n",
" <use xlink:href=\"#m27739734b3\" x=\"56.50625\" y=\"10.999219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_12\">\n",
" <!-- 0.350 -->\n",
" <g transform=\"translate(20.878125 14.798437)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_13\">\n",
" <!-- loss -->\n",
" <g transform=\"translate(14.798438 88.607031)rotate(-90)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-6c\" d=\"M 603 4863 \n",
"L 1178 4863 \n",
"L 1178 0 \n",
"L 603 0 \n",
"L 603 4863 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-73\" d=\"M 2834 3397 \n",
"L 2834 2853 \n",
"Q 2591 2978 2328 3040 \n",
"Q 2066 3103 1784 3103 \n",
"Q 1356 3103 1142 2972 \n",
"Q 928 2841 928 2578 \n",
"Q 928 2378 1081 2264 \n",
"Q 1234 2150 1697 2047 \n",
"L 1894 2003 \n",
"Q 2506 1872 2764 1633 \n",
"Q 3022 1394 3022 966 \n",
"Q 3022 478 2636 193 \n",
"Q 2250 -91 1575 -91 \n",
"Q 1294 -91 989 -36 \n",
"Q 684 19 347 128 \n",
"L 347 722 \n",
"Q 666 556 975 473 \n",
"Q 1284 391 1588 391 \n",
"Q 1994 391 2212 530 \n",
"Q 2431 669 2431 922 \n",
"Q 2431 1156 2273 1281 \n",
"Q 2116 1406 1581 1522 \n",
"L 1381 1569 \n",
"Q 847 1681 609 1914 \n",
"Q 372 2147 372 2553 \n",
"Q 372 3047 722 3315 \n",
"Q 1072 3584 1716 3584 \n",
"Q 2034 3584 2315 3537 \n",
"Q 2597 3491 2834 3397 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-6c\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"27.783203\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"88.964844\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"141.064453\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"line2d_23\">\n",
" <path d=\"M 66.69056 -1 \n",
"L 69.52625 37.371177 \n",
"L 76.03625 81.541253 \n",
"L 82.54625 105.384386 \n",
"L 89.05625 112.227497 \n",
"L 95.56625 120.636156 \n",
"L 102.07625 120.715804 \n",
"L 108.58625 112.552121 \n",
"L 115.09625 121.254148 \n",
"L 121.60625 121.542661 \n",
"L 128.11625 121.798723 \n",
"L 134.62625 122.25835 \n",
"L 141.13625 121.317633 \n",
"L 147.64625 122.931881 \n",
"L 154.15625 123.027853 \n",
"L 160.66625 123.069686 \n",
"L 167.17625 122.553625 \n",
"L 173.68625 122.508729 \n",
"L 180.19625 123.807006 \n",
"L 186.70625 123.146866 \n",
"L 193.21625 122.755345 \n",
"L 199.72625 123.158251 \n",
"L 206.23625 120.2054 \n",
"L 212.74625 120.426721 \n",
"L 219.25625 122.305816 \n",
"L 225.76625 123.435641 \n",
"L 232.27625 118.664978 \n",
"L 238.78625 122.102243 \n",
"L 245.29625 121.385643 \n",
"L 251.80625 111.625322 \n",
"\" clip-path=\"url(#p26492ce462)\" style=\"fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_3\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_4\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_5\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_6\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <defs>\n",
" <clipPath id=\"p26492ce462\">\n",
" <rect x=\"56.50625\" y=\"10.999219\" width=\"195.3\" height=\"135.9\"/>\n",
" </clipPath>\n",
" </defs>\n",
"</svg>\n"
],
"text/plain": [
"<Figure size 252x180 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"trainer = torch.optim.Adam\n",
"d2l.train_concise_ch11(trainer, {'lr': 0.01}, data_iter)"
]
},
{
"cell_type": "markdown",
"id": "6234218c",
"metadata": {
"origin_pos": 12
},
"source": [
"## Yogi\n",
"\n",
"Adam算法也存在一些问题:\n",
"即使在凸环境下,当$\\mathbf{s}_t$的二次矩估计值爆炸时,它可能无法收敛。\n",
" :cite:`Zaheer.Reddi.Sachan.ea.2018`为$\\mathbf{s}_t$提出了的改进更新和参数初始化。\n",
"论文中建议我们重写Adam算法更新如下:\n",
"\n",
"$$\\mathbf{s}_t \\leftarrow \\mathbf{s}_{t-1} + (1 - \\beta_2) \\left(\\mathbf{g}_t^2 - \\mathbf{s}_{t-1}\\right).$$\n",
"\n",
"每当$\\mathbf{g}_t^2$具有值很大的变量或更新很稀疏时,$\\mathbf{s}_t$可能会太快地“忘记”过去的值。\n",
"一个有效的解决方法是将$\\mathbf{g}_t^2 - \\mathbf{s}_{t-1}$替换为$\\mathbf{g}_t^2 \\odot \\mathop{\\mathrm{sgn}}(\\mathbf{g}_t^2 - \\mathbf{s}_{t-1})$。\n",
"这就是Yogi更新,现在更新的规模不再取决于偏差的量。\n",
"\n",
"$$\\mathbf{s}_t \\leftarrow \\mathbf{s}_{t-1} + (1 - \\beta_2) \\mathbf{g}_t^2 \\odot \\mathop{\\mathrm{sgn}}(\\mathbf{g}_t^2 - \\mathbf{s}_{t-1}).$$\n",
"\n",
"论文中,作者还进一步建议用更大的初始批量来初始化动量,而不仅仅是初始的逐点估计。\n"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "14083265",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T07:07:08.526607Z",
"iopub.status.busy": "2023-08-18T07:07:08.526054Z",
"iopub.status.idle": "2023-08-18T07:07:11.129122Z",
"shell.execute_reply": "2023-08-18T07:07:11.128016Z"
},
"origin_pos": 14,
"tab": [
"pytorch"
]
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"loss: 0.245, 0.015 sec/epoch\n"
]
},
{
"data": {
"image/svg+xml": [
"<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n",
"<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
" \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
"<svg xmlns:xlink=\"http://www.w3.org/1999/xlink\" width=\"266.957813pt\" height=\"184.455469pt\" viewBox=\"0 0 266.957813 184.455469\" xmlns=\"http://www.w3.org/2000/svg\" version=\"1.1\">\n",
" <metadata>\n",
" <rdf:RDF xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:cc=\"http://creativecommons.org/ns#\" xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">\n",
" <cc:Work>\n",
" <dc:type rdf:resource=\"http://purl.org/dc/dcmitype/StillImage\"/>\n",
" <dc:date>2023-08-18T07:07:11.093925</dc:date>\n",
" <dc:format>image/svg+xml</dc:format>\n",
" <dc:creator>\n",
" <cc:Agent>\n",
" <dc:title>Matplotlib v3.5.1, https://matplotlib.org/</dc:title>\n",
" </cc:Agent>\n",
" </dc:creator>\n",
" </cc:Work>\n",
" </rdf:RDF>\n",
" </metadata>\n",
" <defs>\n",
" <style type=\"text/css\">*{stroke-linejoin: round; stroke-linecap: butt}</style>\n",
" </defs>\n",
" <g id=\"figure_1\">\n",
" <g id=\"patch_1\">\n",
" <path d=\"M -0 184.455469 \n",
"L 266.957813 184.455469 \n",
"L 266.957813 0 \n",
"L -0 0 \n",
"L -0 184.455469 \n",
"z\n",
"\" style=\"fill: none\"/>\n",
" </g>\n",
" <g id=\"axes_1\">\n",
" <g id=\"patch_2\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"L 56.50625 10.999219 \n",
"z\n",
"\" style=\"fill: #ffffff\"/>\n",
" </g>\n",
" <g id=\"matplotlib.axis_1\">\n",
" <g id=\"xtick_1\">\n",
" <g id=\"line2d_1\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_2\">\n",
" <defs>\n",
" <path id=\"m237606db71\" d=\"M 0 0 \n",
"L 0 3.5 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#m237606db71\" x=\"56.50625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_1\">\n",
" <!-- 0.0 -->\n",
" <g transform=\"translate(48.554688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-30\" d=\"M 2034 4250 \n",
"Q 1547 4250 1301 3770 \n",
"Q 1056 3291 1056 2328 \n",
"Q 1056 1369 1301 889 \n",
"Q 1547 409 2034 409 \n",
"Q 2525 409 2770 889 \n",
"Q 3016 1369 3016 2328 \n",
"Q 3016 3291 2770 3770 \n",
"Q 2525 4250 2034 4250 \n",
"z\n",
"M 2034 4750 \n",
"Q 2819 4750 3233 4129 \n",
"Q 3647 3509 3647 2328 \n",
"Q 3647 1150 3233 529 \n",
"Q 2819 -91 2034 -91 \n",
"Q 1250 -91 836 529 \n",
"Q 422 1150 422 2328 \n",
"Q 422 3509 836 4129 \n",
"Q 1250 4750 2034 4750 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-2e\" d=\"M 684 794 \n",
"L 1344 794 \n",
"L 1344 0 \n",
"L 684 0 \n",
"L 684 794 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_2\">\n",
" <g id=\"line2d_3\">\n",
" <path d=\"M 105.33125 146.899219 \n",
"L 105.33125 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_4\">\n",
" <g>\n",
" <use xlink:href=\"#m237606db71\" x=\"105.33125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_2\">\n",
" <!-- 0.5 -->\n",
" <g transform=\"translate(97.379688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-35\" d=\"M 691 4666 \n",
"L 3169 4666 \n",
"L 3169 4134 \n",
"L 1269 4134 \n",
"L 1269 2991 \n",
"Q 1406 3038 1543 3061 \n",
"Q 1681 3084 1819 3084 \n",
"Q 2600 3084 3056 2656 \n",
"Q 3513 2228 3513 1497 \n",
"Q 3513 744 3044 326 \n",
"Q 2575 -91 1722 -91 \n",
"Q 1428 -91 1123 -41 \n",
"Q 819 9 494 109 \n",
"L 494 744 \n",
"Q 775 591 1075 516 \n",
"Q 1375 441 1709 441 \n",
"Q 2250 441 2565 725 \n",
"Q 2881 1009 2881 1497 \n",
"Q 2881 1984 2565 2268 \n",
"Q 2250 2553 1709 2553 \n",
"Q 1456 2553 1204 2497 \n",
"Q 953 2441 691 2322 \n",
"L 691 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_3\">\n",
" <g id=\"line2d_5\">\n",
" <path d=\"M 154.15625 146.899219 \n",
"L 154.15625 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_6\">\n",
" <g>\n",
" <use xlink:href=\"#m237606db71\" x=\"154.15625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_3\">\n",
" <!-- 1.0 -->\n",
" <g transform=\"translate(146.204688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-31\" d=\"M 794 531 \n",
"L 1825 531 \n",
"L 1825 4091 \n",
"L 703 3866 \n",
"L 703 4441 \n",
"L 1819 4666 \n",
"L 2450 4666 \n",
"L 2450 531 \n",
"L 3481 531 \n",
"L 3481 0 \n",
"L 794 0 \n",
"L 794 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-31\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_4\">\n",
" <g id=\"line2d_7\">\n",
" <path d=\"M 202.98125 146.899219 \n",
"L 202.98125 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_8\">\n",
" <g>\n",
" <use xlink:href=\"#m237606db71\" x=\"202.98125\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_4\">\n",
" <!-- 1.5 -->\n",
" <g transform=\"translate(195.029688 161.497656)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-31\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"xtick_5\">\n",
" <g id=\"line2d_9\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_10\">\n",
" <g>\n",
" <use xlink:href=\"#m237606db71\" x=\"251.80625\" y=\"146.899219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_5\">\n",
" <!-- 2.0 -->\n",
" <g transform=\"translate(243.854688 161.497656)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-32\" d=\"M 1228 531 \n",
"L 3431 531 \n",
"L 3431 0 \n",
"L 469 0 \n",
"L 469 531 \n",
"Q 828 903 1448 1529 \n",
"Q 2069 2156 2228 2338 \n",
"Q 2531 2678 2651 2914 \n",
"Q 2772 3150 2772 3378 \n",
"Q 2772 3750 2511 3984 \n",
"Q 2250 4219 1831 4219 \n",
"Q 1534 4219 1204 4116 \n",
"Q 875 4013 500 3803 \n",
"L 500 4441 \n",
"Q 881 4594 1212 4672 \n",
"Q 1544 4750 1819 4750 \n",
"Q 2544 4750 2975 4387 \n",
"Q 3406 4025 3406 3419 \n",
"Q 3406 3131 3298 2873 \n",
"Q 3191 2616 2906 2266 \n",
"Q 2828 2175 2409 1742 \n",
"Q 1991 1309 1228 531 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-32\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"95.410156\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_6\">\n",
" <!-- epoch -->\n",
" <g transform=\"translate(138.928125 175.175781)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-65\" d=\"M 3597 1894 \n",
"L 3597 1613 \n",
"L 953 1613 \n",
"Q 991 1019 1311 708 \n",
"Q 1631 397 2203 397 \n",
"Q 2534 397 2845 478 \n",
"Q 3156 559 3463 722 \n",
"L 3463 178 \n",
"Q 3153 47 2828 -22 \n",
"Q 2503 -91 2169 -91 \n",
"Q 1331 -91 842 396 \n",
"Q 353 884 353 1716 \n",
"Q 353 2575 817 3079 \n",
"Q 1281 3584 2069 3584 \n",
"Q 2775 3584 3186 3129 \n",
"Q 3597 2675 3597 1894 \n",
"z\n",
"M 3022 2063 \n",
"Q 3016 2534 2758 2815 \n",
"Q 2500 3097 2075 3097 \n",
"Q 1594 3097 1305 2825 \n",
"Q 1016 2553 972 2059 \n",
"L 3022 2063 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-70\" d=\"M 1159 525 \n",
"L 1159 -1331 \n",
"L 581 -1331 \n",
"L 581 3500 \n",
"L 1159 3500 \n",
"L 1159 2969 \n",
"Q 1341 3281 1617 3432 \n",
"Q 1894 3584 2278 3584 \n",
"Q 2916 3584 3314 3078 \n",
"Q 3713 2572 3713 1747 \n",
"Q 3713 922 3314 415 \n",
"Q 2916 -91 2278 -91 \n",
"Q 1894 -91 1617 61 \n",
"Q 1341 213 1159 525 \n",
"z\n",
"M 3116 1747 \n",
"Q 3116 2381 2855 2742 \n",
"Q 2594 3103 2138 3103 \n",
"Q 1681 3103 1420 2742 \n",
"Q 1159 2381 1159 1747 \n",
"Q 1159 1113 1420 752 \n",
"Q 1681 391 2138 391 \n",
"Q 2594 391 2855 752 \n",
"Q 3116 1113 3116 1747 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-6f\" d=\"M 1959 3097 \n",
"Q 1497 3097 1228 2736 \n",
"Q 959 2375 959 1747 \n",
"Q 959 1119 1226 758 \n",
"Q 1494 397 1959 397 \n",
"Q 2419 397 2687 759 \n",
"Q 2956 1122 2956 1747 \n",
"Q 2956 2369 2687 2733 \n",
"Q 2419 3097 1959 3097 \n",
"z\n",
"M 1959 3584 \n",
"Q 2709 3584 3137 3096 \n",
"Q 3566 2609 3566 1747 \n",
"Q 3566 888 3137 398 \n",
"Q 2709 -91 1959 -91 \n",
"Q 1206 -91 779 398 \n",
"Q 353 888 353 1747 \n",
"Q 353 2609 779 3096 \n",
"Q 1206 3584 1959 3584 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-63\" d=\"M 3122 3366 \n",
"L 3122 2828 \n",
"Q 2878 2963 2633 3030 \n",
"Q 2388 3097 2138 3097 \n",
"Q 1578 3097 1268 2742 \n",
"Q 959 2388 959 1747 \n",
"Q 959 1106 1268 751 \n",
"Q 1578 397 2138 397 \n",
"Q 2388 397 2633 464 \n",
"Q 2878 531 3122 666 \n",
"L 3122 134 \n",
"Q 2881 22 2623 -34 \n",
"Q 2366 -91 2075 -91 \n",
"Q 1284 -91 818 406 \n",
"Q 353 903 353 1747 \n",
"Q 353 2603 823 3093 \n",
"Q 1294 3584 2113 3584 \n",
"Q 2378 3584 2631 3529 \n",
"Q 2884 3475 3122 3366 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-68\" d=\"M 3513 2113 \n",
"L 3513 0 \n",
"L 2938 0 \n",
"L 2938 2094 \n",
"Q 2938 2591 2744 2837 \n",
"Q 2550 3084 2163 3084 \n",
"Q 1697 3084 1428 2787 \n",
"Q 1159 2491 1159 1978 \n",
"L 1159 0 \n",
"L 581 0 \n",
"L 581 4863 \n",
"L 1159 4863 \n",
"L 1159 2956 \n",
"Q 1366 3272 1645 3428 \n",
"Q 1925 3584 2291 3584 \n",
"Q 2894 3584 3203 3211 \n",
"Q 3513 2838 3513 2113 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-65\"/>\n",
" <use xlink:href=\"#DejaVuSans-70\" x=\"61.523438\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"125\"/>\n",
" <use xlink:href=\"#DejaVuSans-63\" x=\"186.181641\"/>\n",
" <use xlink:href=\"#DejaVuSans-68\" x=\"241.162109\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"matplotlib.axis_2\">\n",
" <g id=\"ytick_1\">\n",
" <g id=\"line2d_11\">\n",
" <path d=\"M 56.50625 141.672296 \n",
"L 251.80625 141.672296 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_12\">\n",
" <defs>\n",
" <path id=\"mec308516a0\" d=\"M 0 0 \n",
"L -3.5 0 \n",
"\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </defs>\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"141.672296\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_7\">\n",
" <!-- 0.225 -->\n",
" <g transform=\"translate(20.878125 145.471514)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_2\">\n",
" <g id=\"line2d_13\">\n",
" <path d=\"M 56.50625 115.53768 \n",
"L 251.80625 115.53768 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_14\">\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"115.53768\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_8\">\n",
" <!-- 0.250 -->\n",
" <g transform=\"translate(20.878125 119.336899)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_3\">\n",
" <g id=\"line2d_15\">\n",
" <path d=\"M 56.50625 89.403065 \n",
"L 251.80625 89.403065 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_16\">\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"89.403065\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_9\">\n",
" <!-- 0.275 -->\n",
" <g transform=\"translate(20.878125 93.202284)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-37\" d=\"M 525 4666 \n",
"L 3525 4666 \n",
"L 3525 4397 \n",
"L 1831 0 \n",
"L 1172 0 \n",
"L 2766 4134 \n",
"L 525 4134 \n",
"L 525 4666 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-37\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_4\">\n",
" <g id=\"line2d_17\">\n",
" <path d=\"M 56.50625 63.26845 \n",
"L 251.80625 63.26845 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_18\">\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"63.26845\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_10\">\n",
" <!-- 0.300 -->\n",
" <g transform=\"translate(20.878125 67.067668)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-33\" d=\"M 2597 2516 \n",
"Q 3050 2419 3304 2112 \n",
"Q 3559 1806 3559 1356 \n",
"Q 3559 666 3084 287 \n",
"Q 2609 -91 1734 -91 \n",
"Q 1441 -91 1130 -33 \n",
"Q 819 25 488 141 \n",
"L 488 750 \n",
"Q 750 597 1062 519 \n",
"Q 1375 441 1716 441 \n",
"Q 2309 441 2620 675 \n",
"Q 2931 909 2931 1356 \n",
"Q 2931 1769 2642 2001 \n",
"Q 2353 2234 1838 2234 \n",
"L 1294 2234 \n",
"L 1294 2753 \n",
"L 1863 2753 \n",
"Q 2328 2753 2575 2939 \n",
"Q 2822 3125 2822 3475 \n",
"Q 2822 3834 2567 4026 \n",
"Q 2313 4219 1838 4219 \n",
"Q 1578 4219 1281 4162 \n",
"Q 984 4106 628 3988 \n",
"L 628 4550 \n",
"Q 988 4650 1302 4700 \n",
"Q 1616 4750 1894 4750 \n",
"Q 2613 4750 3031 4423 \n",
"Q 3450 4097 3450 3541 \n",
"Q 3450 3153 3228 2886 \n",
"Q 3006 2619 2597 2516 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_5\">\n",
" <g id=\"line2d_19\">\n",
" <path d=\"M 56.50625 37.133834 \n",
"L 251.80625 37.133834 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_20\">\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"37.133834\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_11\">\n",
" <!-- 0.325 -->\n",
" <g transform=\"translate(20.878125 40.933053)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-32\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"ytick_6\">\n",
" <g id=\"line2d_21\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #b0b0b0; stroke-width: 0.8; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"line2d_22\">\n",
" <g>\n",
" <use xlink:href=\"#mec308516a0\" x=\"56.50625\" y=\"10.999219\" style=\"stroke: #000000; stroke-width: 0.8\"/>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_12\">\n",
" <!-- 0.350 -->\n",
" <g transform=\"translate(20.878125 14.798437)scale(0.1 -0.1)\">\n",
" <use xlink:href=\"#DejaVuSans-30\"/>\n",
" <use xlink:href=\"#DejaVuSans-2e\" x=\"63.623047\"/>\n",
" <use xlink:href=\"#DejaVuSans-33\" x=\"95.410156\"/>\n",
" <use xlink:href=\"#DejaVuSans-35\" x=\"159.033203\"/>\n",
" <use xlink:href=\"#DejaVuSans-30\" x=\"222.65625\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"text_13\">\n",
" <!-- loss -->\n",
" <g transform=\"translate(14.798438 88.607031)rotate(-90)scale(0.1 -0.1)\">\n",
" <defs>\n",
" <path id=\"DejaVuSans-6c\" d=\"M 603 4863 \n",
"L 1178 4863 \n",
"L 1178 0 \n",
"L 603 0 \n",
"L 603 4863 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" <path id=\"DejaVuSans-73\" d=\"M 2834 3397 \n",
"L 2834 2853 \n",
"Q 2591 2978 2328 3040 \n",
"Q 2066 3103 1784 3103 \n",
"Q 1356 3103 1142 2972 \n",
"Q 928 2841 928 2578 \n",
"Q 928 2378 1081 2264 \n",
"Q 1234 2150 1697 2047 \n",
"L 1894 2003 \n",
"Q 2506 1872 2764 1633 \n",
"Q 3022 1394 3022 966 \n",
"Q 3022 478 2636 193 \n",
"Q 2250 -91 1575 -91 \n",
"Q 1294 -91 989 -36 \n",
"Q 684 19 347 128 \n",
"L 347 722 \n",
"Q 666 556 975 473 \n",
"Q 1284 391 1588 391 \n",
"Q 1994 391 2212 530 \n",
"Q 2431 669 2431 922 \n",
"Q 2431 1156 2273 1281 \n",
"Q 2116 1406 1581 1522 \n",
"L 1381 1569 \n",
"Q 847 1681 609 1914 \n",
"Q 372 2147 372 2553 \n",
"Q 372 3047 722 3315 \n",
"Q 1072 3584 1716 3584 \n",
"Q 2034 3584 2315 3537 \n",
"Q 2597 3491 2834 3397 \n",
"z\n",
"\" transform=\"scale(0.015625)\"/>\n",
" </defs>\n",
" <use xlink:href=\"#DejaVuSans-6c\"/>\n",
" <use xlink:href=\"#DejaVuSans-6f\" x=\"27.783203\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"88.964844\"/>\n",
" <use xlink:href=\"#DejaVuSans-73\" x=\"141.064453\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <g id=\"line2d_23\">\n",
" <path d=\"M 73.007467 -1 \n",
"L 82.54625 62.440437 \n",
"L 95.56625 101.125442 \n",
"L 108.58625 115.212926 \n",
"L 121.60625 119.800377 \n",
"L 134.62625 122.585943 \n",
"L 147.64625 121.387646 \n",
"L 160.66625 116.834716 \n",
"L 173.68625 118.22928 \n",
"L 186.70625 121.939493 \n",
"L 199.72625 118.857904 \n",
"L 212.74625 123.159671 \n",
"L 225.76625 122.540552 \n",
"L 238.78625 119.650837 \n",
"L 251.80625 120.455439 \n",
"\" clip-path=\"url(#pc59c6bafe9)\" style=\"fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_3\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 56.50625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_4\">\n",
" <path d=\"M 251.80625 146.899219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_5\">\n",
" <path d=\"M 56.50625 146.899219 \n",
"L 251.80625 146.899219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" <g id=\"patch_6\">\n",
" <path d=\"M 56.50625 10.999219 \n",
"L 251.80625 10.999219 \n",
"\" style=\"fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square\"/>\n",
" </g>\n",
" </g>\n",
" </g>\n",
" <defs>\n",
" <clipPath id=\"pc59c6bafe9\">\n",
" <rect x=\"56.50625\" y=\"10.999219\" width=\"195.3\" height=\"135.9\"/>\n",
" </clipPath>\n",
" </defs>\n",
"</svg>\n"
],
"text/plain": [
"<Figure size 252x180 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"def yogi(params, states, hyperparams):\n",
" beta1, beta2, eps = 0.9, 0.999, 1e-3\n",
" for p, (v, s) in zip(params, states):\n",
" with torch.no_grad():\n",
" v[:] = beta1 * v + (1 - beta1) * p.grad\n",
" s[:] = s + (1 - beta2) * torch.sign(\n",
" torch.square(p.grad) - s) * torch.square(p.grad)\n",
" v_bias_corr = v / (1 - beta1 ** hyperparams['t'])\n",
" s_bias_corr = s / (1 - beta2 ** hyperparams['t'])\n",
" p[:] -= hyperparams['lr'] * v_bias_corr / (torch.sqrt(s_bias_corr)\n",
" + eps)\n",
" p.grad.data.zero_()\n",
" hyperparams['t'] += 1\n",
"\n",
"data_iter, feature_dim = d2l.get_data_ch11(batch_size=10)\n",
"d2l.train_ch11(yogi, init_adam_states(feature_dim),\n",
" {'lr': 0.01, 't': 1}, data_iter, feature_dim);"
]
},
{
"cell_type": "markdown",
"id": "4feb3739",
"metadata": {
"origin_pos": 17
},
"source": [
"## 小结\n",
"\n",
"* Adam算法将许多优化算法的功能结合到了相当强大的更新规则中。\n",
"* Adam算法在RMSProp算法基础上创建的,还在小批量的随机梯度上使用EWMA。\n",
"* 在估计动量和二次矩时,Adam算法使用偏差校正来调整缓慢的启动速度。\n",
"* 对于具有显著差异的梯度,我们可能会遇到收敛性问题。我们可以通过使用更大的小批量或者切换到改进的估计值$\\mathbf{s}_t$来修正它们。Yogi提供了这样的替代方案。\n",
"\n",
"## 练习\n",
"\n",
"1. 调节学习率,观察并分析实验结果。\n",
"1. 试着重写动量和二次矩更新,从而使其不需要偏差校正。\n",
"1. 收敛时为什么需要降低学习率$\\eta$\n",
"1. 尝试构造一个使用Adam算法会发散而Yogi会收敛的例子。\n"
]
},
{
"cell_type": "markdown",
"id": "475fc1bc",
"metadata": {
"origin_pos": 19,
"tab": [
"pytorch"
]
},
"source": [
"[Discussions](https://discuss.d2l.ai/t/4331)\n"
]
}
],
"metadata": {
"language_info": {
"name": "python"
},
"required_libs": []
},
"nbformat": 4,
"nbformat_minor": 5
}