dnrops
/
burn
mirror of https://github.com/tracel-ai/burn.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add Jupyter notebook examples (#651)

This commit is contained in:
Dilshod Tadjibaev 2023-08-16 17:17:12 -05:00 committed by GitHub
parent 8448611908
commit 294a319557
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 337 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Cargo.toml
View File

@ -22,6 +22,10 @@ members = [
"examples/*",
]
exclude = [
"examples/notebook",
]
[workspace.dependencies]
bytemuck = "1.13"
const-random = "0.1.15"

47
examples/notebook/README.md Normal file
View File

@ -0,0 +1,47 @@
# Jupyter Notebook Examples with Burn
This directory includes Jupyter Notebook examples showcasing the usage of the Burn deep learning
framework in Rust through
[Evcxr Jupyter](https://github.com/evcxr/evcxr/blob/main/evcxr_jupyter/README.md). The examples are
systematically organized based on the specific Burn features they illustrate.
## Viewing Options
You can explore the examples in different ways:
- **Notebook Viewer:** If you prefer not to set up the entire crate package, you can view the
examples in a notebook viewer or run them to see images and other media outputs.
- **Visual Studio Code (vscode):** If you're using vscode, you already have access to a built-in
notebook viewer, enabling you to open and interact with the notebook files directly.
For other editors, you can utilize the [Jupyter Notebook Viewer](https://nbviewer.jupyter.org/).
## Getting Started with Rust and Evcxr
To execute the Rust code within the notebooks, you must install the Evcxr kernel. Here's how to get
started:
### Install Evcxr Kernel
1. **Build Evcxr Kernel:** Install the required package with the following command:
```shell
cargo install evcxr_jupyter
```
2. **Install and Register the Kernel to Jupyter:**
```shell
evcxr_jupyter --install
```
### Open and Run Notebooks
Once the kernel is installed, you can open the notebook files in your preferred editor and run the
code. Ensure that the kernel is set to `Rust` within the notebook for proper execution.
## Additional Reading Resources
- [Notebook Special Commands for Evcxr](https://github.com/evcxr/evcxr/blob/main/COMMON.md): Learn
about the unique commands and functionalities offered by Evcxr for a more efficient workflow with
Jupyter Notebooks.

173
examples/notebook/basic-tensor-op.ipynb Normal file
View File

@ -0,0 +1,173 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# This notebook demonstrates basic tensor operations in Burn."
]
},
{
"cell_type": "code",
"execution_count": 36,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [],
"source": [
"// Dependency declarations for the notebook. WARNING: It may take a while to compile the first time.\n",
"\n",
"// The syntax is similar to the one used in the Cargo.toml file. Just prefix with :dep\n",
"// See: https://github.com/evcxr/evcxr/blob/main/COMMON.md\n",
"\n",
":dep burn = {path = \"../../burn\"}\n",
":dep burn-ndarray = {path = \"../../burn-ndarray\"}"
]
},
{
"cell_type": "code",
"execution_count": 37,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [],
"source": [
"// Import packages\n",
"use burn::tensor::Tensor;\n",
"use burn_ndarray::NdArrayBackend;\n",
"\n",
"// Type alias for the backend\n",
"type B = NdArrayBackend<f32>;"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Tensor creation"
]
},
{
"cell_type": "code",
"execution_count": 38,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Empty tensor: Tensor {\n",
" data: [[[0.0, 0.0, 0.0], [0.0, 0.0, 0.0]]],\n",
" shape: [1, 2, 3],\n",
" device: Cpu,\n",
" backend: \"ndarray\",\n",
" kind: \"Float\",\n",
" dtype: \"f32\",\n",
"}\n",
"Tensor from slice: Tensor {\n",
" data: [[1.0, 2.0], [3.0, 4.0]],\n",
" shape: [2, 2],\n",
" device: Cpu,\n",
" backend: \"ndarray\",\n",
" kind: \"Float\",\n",
" dtype: \"f32\",\n",
"}\n",
"Random tensor: Tensor {\n",
" data: [0.16685265, 0.7217095, 0.35741878, 0.49403405, 0.27360022],\n",
" shape: [5],\n",
" device: Cpu,\n",
" backend: \"ndarray\",\n",
" kind: \"Float\",\n",
" dtype: \"f32\",\n",
"}\n"
]
}
],
"source": [
"// Create an empty tensor for a given shape\n",
"let tensor: Tensor<B, 3> = Tensor::empty([1, 2, 3]);\n",
"println!(\"Empty tensor: {}\", tensor);\n",
"\n",
"// Create a tensor from a slice of floats\n",
"let tensor: Tensor<B, 2> = Tensor::from_floats([1.0, 2.0, 3.0, 4.0]).reshape([2, 2]);\n",
"println!(\"Tensor from slice: {}\", tensor);\n",
"\n",
"// Create a random tensor\n",
"use burn::tensor::Distribution;\n",
"let tensor: Tensor<B, 1> = Tensor::random([5], Distribution::Default);\n",
"println!(\"Random tensor: {}\", tensor);\n",
"\n",
"// Create a tensor using fill values, zeros, or ones\n",
"let tensor: Tensor<B,2> = Tensor::full([2, 2], 7.0);\n",
"let tensor: Tensor<B,2> = Tensor::zeros([2, 2]);\n",
"let tensor: Tensor<B,2> = Tensor::ones([2, 2]);\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Tensor Operations\n"
]
},
{
"cell_type": "code",
"execution_count": 39,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"x3 = Tensor {\n",
" data: [[8.0, 8.0], [8.0, 8.0]],\n",
" shape: [2, 2],\n",
" device: Cpu,\n",
" backend: \"ndarray\",\n",
" kind: \"Float\",\n",
" dtype: \"f32\",\n",
"}\n"
]
}
],
"source": [
"let x1: Tensor<B,2> = Tensor::ones([2, 2]);\n",
"let x2: Tensor<B,2> = Tensor::full([2, 2], 7.0);\n",
"\n",
"let x3 = x1 + x2;\n",
"\n",
"println!(\"x3 = {}\", x3);"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Rust",
"language": "rust",
"name": "rust"
},
"language_info": {
"codemirror_mode": "rust",
"file_extension": ".rs",
"mimetype": "text/rust",
"name": "Rust",
"pygment_lexer": "rust",
"version": ""
},
"orig_nbformat": 4
},
"nbformat": 4,
"nbformat_minor": 2
}

113
examples/notebook/plots.ipynb Normal file
View File

@ -0,0 +1,113 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# This notebook demonstrates basic tensor operations in Burn."
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [],
"source": [
"// Dependency declarations for the notebook. WARNING: It may take a while to compile the first time.\n",
"\n",
"// The syntax is similar to the one used in the Cargo.toml file. Just prefix with :dep\n",
"// See: https://github.com/evcxr/evcxr/blob/main/COMMON.md\n",
"\n",
":dep burn = {path = \"../../burn\"}\n",
":dep burn-ndarray = {path = \"../../burn-ndarray\"}\n",
"\n",
"// The following dependencies are used for plotting\n",
":dep image = \"0.23\"\n",
":dep evcxr_image = \"1.1\""
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [],
"source": [
"// Import packages\n",
"use burn::tensor::Tensor;\n",
"use burn_ndarray::NdArrayBackend;\n",
"\n",
"// Import plotting library\n",
"use evcxr_image::ImageDisplay;\n",
"\n",
"// Type alias for the backend\n",
"type B = NdArrayBackend<f32>;"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Image from tensor"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"vscode": {
"languageId": "rust"
}
},
"outputs": [
{
"data": {
"image/png": "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"
},
"execution_count": 15,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"// Create a random tensor\n",
"use burn::tensor::Distribution;\n",
"let tensor: Tensor<B, 3> = Tensor::random([3, 256, 256], Distribution::Default);\n",
"\n",
"// TODO Use tenso to display plots\n",
"image::ImageBuffer::from_fn(256, 256, |x, y| {\n",
" if (x as i32 - y as i32).abs() < 3 {\n",
" image::Rgb([0, 0, 255])\n",
" } else {\n",
" image::Rgb([0, 0, 0])\n",
" }\n",
"})\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Rust",
"language": "rust",
"name": "rust"
},
"language_info": {
"codemirror_mode": "rust",
"file_extension": ".rs",
"mimetype": "text/rust",
"name": "Rust",
"pygment_lexer": "rust",
"version": ""
},
"orig_nbformat": 4
},
"nbformat": 4,
"nbformat_minor": 2
}