Skip to content

Commit

Permalink
Add files via upload
Browse files Browse the repository at this point in the history
  • Loading branch information
@dataprofessor
dataprofessor authored Sep 6, 2020
1 parent 988c733 commit 2d8dfdb
Showing 1 changed file with 360 additions and 0 deletions.
360 changes: 360 additions & 0 deletions python/Hummingbird_ML.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,360 @@
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "Hummingbird-ML.ipynb",
"provenance": [],
"collapsed_sections": []
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "IpOdlr3WAPHJ",
"colab_type": "text"
},
"source": [
"# **Hummingbird-ML**\n",
"\n",
"[How to Harness GPU to Speed Up Machine Learning with Hummingbird-ML](https://www.youtube.com/watch?v=qN8jcUmo8TI)\n",
"\n",
"Adapted from: https://github.com/microsoft/hummingbird"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "ir3DZd5-_jiu",
"colab_type": "text"
},
"source": [
"# Install Hummingbird-ML"
]
},
{
"cell_type": "code",
"metadata": {
"id": "ra3JEgWN_bfp",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 408
},
"outputId": "4fae39de-26f0-4939-846d-039fb876725a"
},
"source": [
"! pip install hummingbird-ml[extra]"
],
"execution_count": 1,
"outputs": [
{
"output_type": "stream",
"text": [
"Collecting hummingbird-ml[extra]\n",
"\u001b[?25l Downloading https://files.pythonhosted.org/packages/ed/3b/cf1b8c1e7531377adead8de29e29b00b5aed380544ad0def4c0188b50d80/hummingbird_ml-0.0.5-py2.py3-none-any.whl (60kB)\n",
"\r\u001b[K |█████▌ | 10kB 16.6MB/s eta 0:00:01\r\u001b[K |███████████ | 20kB 1.8MB/s eta 0:00:01\r\u001b[K |████████████████▍ | 30kB 2.2MB/s eta 0:00:01\r\u001b[K |█████████████████████▉ | 40kB 2.5MB/s eta 0:00:01\r\u001b[K |███████████████████████████▎ | 51kB 2.0MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 61kB 1.8MB/s \n",
"\u001b[?25hRequirement already satisfied: numpy>=1.15 in /usr/local/lib/python3.6/dist-packages (from hummingbird-ml[extra]) (1.18.5)\n",
"Requirement already satisfied: torch>=1.4.* in /usr/local/lib/python3.6/dist-packages (from hummingbird-ml[extra]) (1.6.0+cu101)\n",
"Collecting onnxconverter-common>=1.6.0\n",
"\u001b[?25l Downloading https://files.pythonhosted.org/packages/fe/7a/7e30c643cd7d2ad87689188ef34ce93e657bd14da3605f87bcdbc19cd5b1/onnxconverter_common-1.7.0-py2.py3-none-any.whl (64kB)\n",
"\u001b[K |████████████████████████████████| 71kB 3.7MB/s \n",
"\u001b[?25hRequirement already satisfied: scikit-learn>=0.22.1 in /usr/local/lib/python3.6/dist-packages (from hummingbird-ml[extra]) (0.22.2.post1)\n",
"Requirement already satisfied: xgboost==0.90; extra == \"extra\" in /usr/local/lib/python3.6/dist-packages (from hummingbird-ml[extra]) (0.90)\n",
"Requirement already satisfied: lightgbm>=2.2; extra == \"extra\" in /usr/local/lib/python3.6/dist-packages (from hummingbird-ml[extra]) (2.2.3)\n",
"Requirement already satisfied: future in /usr/local/lib/python3.6/dist-packages (from torch>=1.4.*->hummingbird-ml[extra]) (0.16.0)\n",
"Collecting onnx\n",
"\u001b[?25l Downloading https://files.pythonhosted.org/packages/36/ee/bc7bc88fc8449266add978627e90c363069211584b937fd867b0ccc59f09/onnx-1.7.0-cp36-cp36m-manylinux1_x86_64.whl (7.4MB)\n",
"\u001b[K |████████████████████████████████| 7.4MB 16.0MB/s \n",
"\u001b[?25hRequirement already satisfied: protobuf in /usr/local/lib/python3.6/dist-packages (from onnxconverter-common>=1.6.0->hummingbird-ml[extra]) (3.12.4)\n",
"Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.6/dist-packages (from scikit-learn>=0.22.1->hummingbird-ml[extra]) (0.16.0)\n",
"Requirement already satisfied: scipy>=0.17.0 in /usr/local/lib/python3.6/dist-packages (from scikit-learn>=0.22.1->hummingbird-ml[extra]) (1.4.1)\n",
"Requirement already satisfied: typing-extensions>=3.6.2.1 in /usr/local/lib/python3.6/dist-packages (from onnx->onnxconverter-common>=1.6.0->hummingbird-ml[extra]) (3.7.4.3)\n",
"Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from onnx->onnxconverter-common>=1.6.0->hummingbird-ml[extra]) (1.15.0)\n",
"Requirement already satisfied: setuptools in /usr/local/lib/python3.6/dist-packages (from protobuf->onnxconverter-common>=1.6.0->hummingbird-ml[extra]) (49.6.0)\n",
"Installing collected packages: onnx, onnxconverter-common, hummingbird-ml\n",
"Successfully installed hummingbird-ml-0.0.5 onnx-1.7.0 onnxconverter-common-1.7.0\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "YnA-PmeA_q70",
"colab_type": "text"
},
"source": [
"# Import libraries"
]
},
{
"cell_type": "code",
"metadata": {
"id": "lkIThThi_puf",
"colab_type": "code",
"colab": {}
},
"source": [
"import numpy as np\n",
"from sklearn.ensemble import RandomForestClassifier\n",
"from hummingbird.ml import convert"
],
"execution_count": 2,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "rFw_4cGa_-tF",
"colab_type": "text"
},
"source": [
"# Create some random data for binary classification"
]
},
{
"cell_type": "code",
"metadata": {
"id": "hGGngPPp__mx",
"colab_type": "code",
"colab": {}
},
"source": [
"num_classes = 2\n",
"X = np.random.rand(100000, 28)\n",
"y = np.random.randint(num_classes, size=100000)"
],
"execution_count": 3,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "WusxNKH4AHII",
"colab_type": "text"
},
"source": [
"# Create and train a model (scikit-learn RandomForestClassifier)"
]
},
{
"cell_type": "code",
"metadata": {
"id": "GMRJRuBwAGeV",
"colab_type": "code",
"colab": {}
},
"source": [
"skl_model = RandomForestClassifier(n_estimators=10, max_depth=10)"
],
"execution_count": 4,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "M_kGo80yAYTn",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "aa863652-02f8-4578-8fb7-e3b028685cd7"
},
"source": [
"%%timeit\n",
"skl_model.fit(X, y)"
],
"execution_count": 5,
"outputs": [
{
"output_type": "stream",
"text": [
"1 loop, best of 3: 4.78 s per loop\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "Hp4a8I0tAbBl",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "4e083fd5-981f-4238-9158-3f4500585560"
},
"source": [
"%%timeit\n",
"skl_model.predict(X)"
],
"execution_count": 6,
"outputs": [
{
"output_type": "stream",
"text": [
"10 loops, best of 3: 85.6 ms per loop\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "mNiBvy9BA7wR",
"colab_type": "text"
},
"source": [
"# Use Hummingbird to convert the model to PyTorch"
]
},
{
"cell_type": "code",
"metadata": {
"id": "vcAOpuxxAzPc",
"colab_type": "code",
"colab": {}
},
"source": [
"model = convert(skl_model, 'pytorch')"
],
"execution_count": 7,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "dpt6_4l8BF7e",
"colab_type": "text"
},
"source": [
"# Run predictions on CPU"
]
},
{
"cell_type": "code",
"metadata": {
"id": "_BiU63hNBDu-",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "1bd8b158-a62b-4fe0-be09-ca382c817247"
},
"source": [
"%%timeit\n",
"model.predict(X)"
],
"execution_count": 8,
"outputs": [
{
"output_type": "stream",
"text": [
"1 loop, best of 3: 174 ms per loop\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "F10tJEMKBPZG",
"colab_type": "text"
},
"source": [
"# Run predictions on GPU"
]
},
{
"cell_type": "code",
"metadata": {
"id": "l2PUbqoHBJBX",
"colab_type": "code",
"colab": {}
},
"source": [
"model.to('cuda')"
],
"execution_count": 9,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "-AB23_VTBRMP",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 51
},
"outputId": "b9efea7d-913c-4326-c14a-6b6ca0e9c063"
},
"source": [
"%%timeit\n",
"model.predict(X)"
],
"execution_count": 10,
"outputs": [
{
"output_type": "stream",
"text": [
"The slowest run took 5.22 times longer than the fastest. This could mean that an intermediate result is being cached.\n",
"100 loops, best of 3: 14.8 ms per loop\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "dbkQU69JDt7T",
"colab_type": "text"
},
"source": [
"# Calculation Time"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Hr1R_9nwDwpc",
"colab_type": "text"
},
"source": [
"Methods | Timing | Performance\n",
"--|--|--\n",
"scikit-learn | 85.6 ms | -\n",
"PyTorch (CPU) | 174 ms | 2 X slower than scikit-learn\n",
"PyTorch (GPU) | 14.8 ms | Almost 6 X faster than scikit-learn; Almost 12 X faster than PyTorch (CPU)"
]
},
{
"cell_type": "code",
"metadata": {
"id": "9lmR3LHoEzhl",
"colab_type": "code",
"colab": {}
},
"source": [
""
],
"execution_count": null,
"outputs": []
}
]
}

0 comments on commit 2d8dfdb

Please sign in to comment.