catethos/onnx_percentile.ipynb

## onnx_percentile.ipynb
{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "9fcfd4ff-6340-4ef1-a501-ece7fdf138a1",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "id": "a29b0d3a-f593-4684-bba9-77e18e8669e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import beta\n",
    "import numpy as np\n",
    "from torch import nn\n",
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "import onnxruntime as ort"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38db0175-8d59-4bdc-967c-50ecd23c73e9",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "8e0dbad0-841b-4b68-b271-d32aae16a0a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "x = np.linspace(0,1,150).reshape(150,1).astype(np.float32)\n",
    "x = torch.from_numpy(x)\n",
    "\n",
    "y = beta(1,2).cdf(x).astype(np.float32)\n",
    "y = torch.from_numpy(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8184ce8d-efa9-449b-bc2d-ee93e2c28005",
   "metadata": {},
   "source": [
    "# Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "id": "15fe497e-762f-4527-8cd1-9dd7764f9745",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Network(nn.Module):\n",
    "    \n",
    "  def __init__(self):\n",
    "    super().__init__()\n",
    "    self.layers = nn.Sequential(\n",
    "      nn.Linear(1, 25),\n",
    "      nn.ReLU(),\n",
    "      nn.Linear(25, 1)\n",
    "    )\n",
    "\n",
    "\n",
    "  def forward(self, x):\n",
    "    return self.layers(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8c38d77-eef5-4551-9eb4-3d23b5ce3b68",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "c9751d33-b9ce-4261-a15e-126a12ef99ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "net = Network()\n",
    "loss_function = nn.MSELoss()\n",
    "optimizer = torch.optim.Adam(net.parameters(), lr=1e-2)\n",
    "\n",
    "for epoch in range(0, 500):\n",
    "    optimizer.zero_grad()\n",
    "    outputs = net(x)\n",
    "    loss = loss_function(outputs, y)\n",
    "    loss.backward()\n",
    "    optimizer.step()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93994203-db33-4ea8-808c-e981049565d0",
   "metadata": {},
   "source": [
    "## Exporting to ONNX"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "id": "1a2236f1-ce20-46dd-b92d-bdc43578c1c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.onnx.export(net, x, \"percentile.onnx\",\n",
    "                 input_names = ['input'],\n",
    "                 output_names = ['output'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74fa9197-94d3-4bf9-85b4-7ddfca5d1aaa",
   "metadata": {},
   "source": [
    "# Inference Time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6de4285b-6b79-4117-9155-171a81964806",
   "metadata": {},
   "outputs": [],
   "source": [
    "sess = ort.InferenceSession(\"percentile.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "id": "31e6d34f-5f1f-4358-a320-59f956576a73",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 110 µs, sys: 5 µs, total: 115 µs\n",
      "Wall time: 78.7 µs\n",
      "CPU times: user 1.18 ms, sys: 8 µs, total: 1.19 ms\n",
      "Wall time: 939 µs\n"
     ]
    }
   ],
   "source": [
    "# onnx is so much faster\n",
    "%time y_est  = sess.run(None, {'input': x.numpy()})[0]\n",
    "%time y_true = beta(1,2).cdf(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "id": "400350cf-8c99-443e-8912-4bcf7bd05675",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f09319506a0>]"
      ]
     },
     "execution_count": 247,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(y_est, y_true)\n",
    "plt.plot([0,1],[0,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d9711541-2169-4cee-9b38-c3eec6d3b11e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "290f61ae-d229-4a28-a9d7-da19508c1ff3",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "analysis",
   "language": "python",
   "name": "analysis"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}
	{
	"cells": [
	{
	"cell_type": "markdown",
	"id": "9fcfd4ff-6340-4ef1-a501-ece7fdf138a1",
	"metadata": {
	"tags": []
	},
	"source": [
	"# Libraries"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 239,
	"id": "a29b0d3a-f593-4684-bba9-77e18e8669e2",
	"metadata": {},
	"outputs": [],
	"source": [
	"from scipy.stats import beta\n",
	"import numpy as np\n",
	"from torch import nn\n",
	"import torch\n",
	"import matplotlib.pyplot as plt\n",
	"import onnxruntime as ort"
	]
	},
	{
	"cell_type": "markdown",
	"id": "38db0175-8d59-4bdc-967c-50ecd23c73e9",
	"metadata": {
	"tags": []
	},
	"source": [
	"# Data"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 180,
	"id": "8e0dbad0-841b-4b68-b271-d32aae16a0a5",
	"metadata": {},
	"outputs": [],
	"source": [
	"x = np.linspace(0,1,150).reshape(150,1).astype(np.float32)\n",
	"x = torch.from_numpy(x)\n",
	"\n",
	"y = beta(1,2).cdf(x).astype(np.float32)\n",
	"y = torch.from_numpy(y)"
	]
	},
	{
	"cell_type": "markdown",
	"id": "8184ce8d-efa9-449b-bc2d-ee93e2c28005",
	"metadata": {},
	"source": [
	"# Model"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 190,
	"id": "15fe497e-762f-4527-8cd1-9dd7764f9745",
	"metadata": {},
	"outputs": [],
	"source": [
	"class Network(nn.Module):\n",
	" \n",
	" def __init__(self):\n",
	" super().__init__()\n",
	" self.layers = nn.Sequential(\n",
	" nn.Linear(1, 25),\n",
	" nn.ReLU(),\n",
	" nn.Linear(25, 1)\n",
	" )\n",
	"\n",
	"\n",
	" def forward(self, x):\n",
	" return self.layers(x)"
	]
	},
	{
	"cell_type": "markdown",
	"id": "c8c38d77-eef5-4551-9eb4-3d23b5ce3b68",
	"metadata": {},
	"source": [
	"# Training"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 220,
	"id": "c9751d33-b9ce-4261-a15e-126a12ef99ce",
	"metadata": {},
	"outputs": [],
	"source": [
	"net = Network()\n",
	"loss_function = nn.MSELoss()\n",
	"optimizer = torch.optim.Adam(net.parameters(), lr=1e-2)\n",
	"\n",
	"for epoch in range(0, 500):\n",
	" optimizer.zero_grad()\n",
	" outputs = net(x)\n",
	" loss = loss_function(outputs, y)\n",
	" loss.backward()\n",
	" optimizer.step()"
	]
	},
	{
	"cell_type": "markdown",
	"id": "93994203-db33-4ea8-808c-e981049565d0",
	"metadata": {},
	"source": [
	"## Exporting to ONNX"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 230,
	"id": "1a2236f1-ce20-46dd-b92d-bdc43578c1c9",
	"metadata": {},
	"outputs": [],
	"source": [
	"torch.onnx.export(net, x, \"percentile.onnx\",\n",
	" input_names = ['input'],\n",
	" output_names = ['output'])"
	]
	},
	{
	"cell_type": "markdown",
	"id": "74fa9197-94d3-4bf9-85b4-7ddfca5d1aaa",
	"metadata": {},
	"source": [
	"# Inference Time"
	]
	},
	{
	"cell_type": "code",
	"execution_count": null,
	"id": "6de4285b-6b79-4117-9155-171a81964806",
	"metadata": {},
	"outputs": [],
	"source": [
	"sess = ort.InferenceSession(\"percentile.onnx\")"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 248,
	"id": "31e6d34f-5f1f-4358-a320-59f956576a73",
	"metadata": {},
	"outputs": [
	{
	"name": "stdout",
	"output_type": "stream",
	"text": [
	"CPU times: user 110 µs, sys: 5 µs, total: 115 µs\n",
	"Wall time: 78.7 µs\n",
	"CPU times: user 1.18 ms, sys: 8 µs, total: 1.19 ms\n",
	"Wall time: 939 µs\n"
	]
	}
	],
	"source": [
	"# onnx is so much faster\n",
	"%time y_est = sess.run(None, {'input': x.numpy()})[0]\n",
	"%time y_true = beta(1,2).cdf(x)"
	]
	},
	{
	"cell_type": "code",
	"execution_count": 247,
	"id": "400350cf-8c99-443e-8912-4bcf7bd05675",
	"metadata": {},
	"outputs": [
	{
	"data": {
	"text/plain": [
	"[<matplotlib.lines.Line2D at 0x7f09319506a0>]"
	]
	},
	"execution_count": 247,
	"metadata": {},
	"output_type": "execute_result"
	},
	{
	"data": {
	"image/png": "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\n",
	"text/plain": [
	"<Figure size 432x288 with 1 Axes>"
	]
	},
	"metadata": {
	"needs_background": "light"
	},
	"output_type": "display_data"
	}
	],
	"source": [
	"plt.plot(y_est, y_true)\n",
	"plt.plot([0,1],[0,1])"
	]
	},
	{
	"cell_type": "code",
	"execution_count": null,
	"id": "d9711541-2169-4cee-9b38-c3eec6d3b11e",
	"metadata": {},
	"outputs": [],
	"source": []
	},
	{
	"cell_type": "code",
	"execution_count": null,
	"id": "290f61ae-d229-4a28-a9d7-da19508c1ff3",
	"metadata": {},
	"outputs": [],
	"source": []
	}
	],
	"metadata": {
	"kernelspec": {
	"display_name": "analysis",
	"language": "python",
	"name": "analysis"
	},
	"language_info": {
	"codemirror_mode": {
	"name": "ipython",
	"version": 3
	},
	"file_extension": ".py",
	"mimetype": "text/x-python",
	"name": "python",
	"nbconvert_exporter": "python",
	"pygments_lexer": "ipython3",
	"version": "3.9.7"
	}
	},
	"nbformat": 4,
	"nbformat_minor": 5
	}