Skip to content

Commit

Permalink
chore: svm example (#473)
Browse files Browse the repository at this point in the history
  • Loading branch information
@alexander-camuto
alexander-camuto authored Sep 11, 2023
1 parent 3c1b9d1 commit ac1aaa2
Show file tree
Hide file tree
Showing 8 changed files with 326 additions and 29 deletions.
9 changes: 3 additions & 6 deletions .github/workflows/rust.yml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -293,12 +293,7 @@ jobs:

prove-and-verify-aggr-evm-tests:
runs-on: large-self-hosted
needs:
[
build,
library-tests,
python-tests,
]
needs: [build, library-tests, python-tests]
steps:
- uses: actions/checkout@v4
- uses: actions-rs/toolchain@v1
Expand Down Expand Up @@ -460,6 +455,8 @@ jobs:
# # now dump the contents of the file into a file called kaggle.json
# echo $KAGGLE_API_KEY > /home/ubuntu/.kaggle/kaggle.json
# chmod 600 /home/ubuntu/.kaggle/kaggle.json
- name: SVM
run: source .env/bin/activate; cargo nextest run py_tests::tests::run_notebook_::tests_18_expects
- name: LightGBM
run: source .env/bin/activate; cargo nextest run py_tests::tests::run_notebook_::tests_17_expects
- name: XGBoost
Expand Down
14 changes: 7 additions & 7 deletions Cargo.lock
View file

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 1 addition & 1 deletion Cargo.toml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@ tokio = { version = "1.26.0", default_features = false, features = ["macros", "
pyo3 = { version = "0.18.3", features = ["extension-module", "abi3-py37", "macros"], default_features = false, optional = true }
pyo3-asyncio = { version = "0.18.0", features = ["attributes", "tokio-runtime"], default_features = false, optional = true }
pyo3-log = { version = "0.8.1", default_features = false, optional = true }
tract-onnx = { git = "https://github.com/sonos/tract/", rev= "ca451330931687f00bccd30a1e2b5ec2fcccdcc9", default_features = false, optional = true }
tract-onnx = { git = "https://github.com/sonos/tract/", rev= "2ea76c09678f092d00713ebbe6fdb046c0a9ad0f", default_features = false, optional = true }
tabled = { version = "0.12.0", optional = true }


Expand Down
303 changes: 303 additions & 0 deletions examples/notebooks/svm.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,303 @@
{
"cells": [
{
"attachments": {},
"cell_type": "markdown",
"id": "cf69bb3f-94e6-4dba-92cd-ce08df117d67",
"metadata": {},
"source": [
"## Support Vector Machines\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "95613ee9",
"metadata": {},
"outputs": [],
"source": [
"# check if notebook is in colab\n",
"try:\n",
" # install ezkl\n",
" import google.colab\n",
" import subprocess\n",
" import sys\n",
" subprocess.check_call([sys.executable, \"-m\", \"pip\", \"install\", \"ezkl\"])\n",
" subprocess.check_call([sys.executable, \"-m\", \"pip\", \"install\", \"onnx\"])\n",
" subprocess.check_call([sys.executable, \"-m\", \"pip\", \"install\", \"sk2torch\"])\n",
"\n",
"# rely on local installation of ezkl if the notebook is not in colab\n",
"except:\n",
" pass\n",
"\n",
"\n",
"# here we create and (potentially train a model)\n",
"\n",
"# make sure you have the dependencies required here already installed\n",
"import json\n",
"import numpy as np\n",
"from sklearn.svm import SVC\n",
"import sk2torch\n",
"import torch\n",
"import ezkl\n",
"import os\n",
"\n",
"\n",
"# Create a dataset of two Gaussians. There will be some overlap\n",
"# between the two classes, which adds some uncertainty to the model.\n",
"xs = np.concatenate(\n",
" [\n",
" np.random.random(size=(256, 2)) + [1, 0],\n",
" np.random.random(size=(256, 2)) + [-1, 0],\n",
" ],\n",
" axis=0,\n",
")\n",
"ys = np.array([False] * 256 + [True] * 256)\n",
"\n",
"# Train an SVM on the data and wrap it in PyTorch.\n",
"sk_model = SVC(probability=True)\n",
"sk_model.fit(xs, ys)\n",
"model = sk2torch.wrap(sk_model)\n",
"\n",
"\n",
"\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b37637c4",
"metadata": {},
"outputs": [],
"source": [
"model_path = os.path.join('network.onnx')\n",
"compiled_model_path = os.path.join('network.compiled')\n",
"pk_path = os.path.join('test.pk')\n",
"vk_path = os.path.join('test.vk')\n",
"settings_path = os.path.join('settings.json')\n",
"srs_path = os.path.join('kzg.srs')\n",
"witness_path = os.path.join('witness.json')\n",
"data_path = os.path.join('input.json')"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7f0ca328",
"metadata": {},
"outputs": [],
"source": [
"import matplotlib.pyplot as plt\n",
"# Create a coordinate grid to compute a vector field on.\n",
"spaced = np.linspace(-2, 2, num=25)\n",
"grid_xs = torch.tensor([[x, y] for x in spaced for y in spaced], requires_grad=True)\n",
"\n",
"\n",
"# Compute the gradients of the SVM output.\n",
"outputs = model.predict_proba(grid_xs)[:, 1]\n",
"(input_grads,) = torch.autograd.grad(outputs.sum(), (grid_xs,))\n",
"\n",
"\n",
"# Create a quiver plot of the vector field.\n",
"plt.quiver(\n",
" grid_xs[:, 0].detach().numpy(),\n",
" grid_xs[:, 1].detach().numpy(),\n",
" input_grads[:, 0].detach().numpy(),\n",
" input_grads[:, 1].detach().numpy(),\n",
")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "82db373a",
"metadata": {},
"outputs": [],
"source": [
"\n",
"\n",
"# export to onnx format\n",
"# !!!!!!!!!!!!!!!!! This will flash a warning but it is fine !!!!!!!!!!!!!!!!!!!!!\n",
"\n",
"# Input to the model\n",
"shape = xs.shape[1:]\n",
"x = grid_xs[0:1]\n",
"torch_out = model.predict(x)\n",
"# Export the model\n",
"torch.onnx.export(model, # model being run\n",
" # model input (or a tuple for multiple inputs)\n",
" x,\n",
" # where to save the model (can be a file or file-like object)\n",
" \"network.onnx\",\n",
" export_params=True, # store the trained parameter weights inside the model file\n",
" opset_version=10, # the ONNX version to export the model to\n",
" do_constant_folding=True, # whether to execute constant folding for optimization\n",
" input_names=['input'], # the model's input names\n",
" output_names=['output'], # the model's output names\n",
" dynamic_axes={'input': {0: 'batch_size'}, # variable length axes\n",
" 'output': {0: 'batch_size'}})\n",
"\n",
"d = ((x).detach().numpy()).reshape([-1]).tolist()\n",
"\n",
"data = dict(input_shapes=[shape],\n",
" input_data=[d],\n",
" output_data=[o.reshape([-1]).tolist() for o in torch_out])\n",
"\n",
"# Serialize data into file:\n",
"json.dump(data, open(\"input.json\", 'w'))\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d5e374a2",
"metadata": {},
"outputs": [],
"source": [
"!RUST_LOG=trace\n",
"# TODO: Dictionary outputs\n",
"res = ezkl.gen_settings(model_path, settings_path)\n",
"assert res == True\n",
"\n",
"res = await ezkl.calibrate_settings(data_path, model_path, settings_path, \"resources\")\n",
"assert res == True"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3aa4f090",
"metadata": {},
"outputs": [],
"source": [
"res = ezkl.compile_model(model_path, compiled_model_path, settings_path)\n",
"assert res == True"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8b74dcee",
"metadata": {},
"outputs": [],
"source": [
"# srs path\n",
"res = ezkl.get_srs(srs_path, settings_path)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "18c8b7c7",
"metadata": {},
"outputs": [],
"source": [
"# now generate the witness file \n",
"\n",
"res = ezkl.gen_witness(data_path, compiled_model_path, witness_path, settings_path = settings_path)\n",
"assert os.path.isfile(witness_path)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b1c561a8",
"metadata": {},
"outputs": [],
"source": [
"\n",
"# HERE WE SETUP THE CIRCUIT PARAMS\n",
"# WE GOT KEYS\n",
"# WE GOT CIRCUIT PARAMETERS\n",
"# EVERYTHING ANYONE HAS EVER NEEDED FOR ZK\n",
"\n",
"\n",
"\n",
"res = ezkl.setup(\n",
" compiled_model_path,\n",
" vk_path,\n",
" pk_path,\n",
" srs_path,\n",
" settings_path,\n",
" )\n",
"\n",
"assert res == True\n",
"assert os.path.isfile(vk_path)\n",
"assert os.path.isfile(pk_path)\n",
"assert os.path.isfile(settings_path)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "c384cbc8",
"metadata": {},
"outputs": [],
"source": [
"# GENERATE A PROOF\n",
"\n",
"\n",
"proof_path = os.path.join('test.pf')\n",
"\n",
"res = ezkl.prove(\n",
" witness_path,\n",
" compiled_model_path,\n",
" pk_path,\n",
" proof_path,\n",
" srs_path,\n",
" \"evm\",\n",
" \"single\",\n",
" settings_path,\n",
" )\n",
"\n",
"print(res)\n",
"assert os.path.isfile(proof_path)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "76f00d41",
"metadata": {},
"outputs": [],
"source": [
"# VERIFY IT\n",
"\n",
"res = ezkl.verify(\n",
" proof_path,\n",
" settings_path,\n",
" vk_path,\n",
" srs_path,\n",
" )\n",
"\n",
"assert res == True\n",
"print(\"verified\")"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"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": "3.9.15"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Loading

0 comments on commit ac1aaa2

Please sign in to comment.