Docs 2/n (#15521)
* merge master * merge master * merge master * merge master * install colors * install colors * install colors * install colors * install colors * install colors * install colors * install colors * docs * docs * docs * docs * docs * docs * docs * docs * Revert "docs" This reverts commitc83d9854fb. * Revert "docs" This reverts commita2bb66d2f0. * docs * docs * remove source-lit * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * precommit * files * folder Co-authored-by: Jirka <jirka.borovec@seznam.cz> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
This commit is contained in:
William Falcon
GitHub
parent
4750e221d1
commit
69bac9ffd9
|
|
@ -72,6 +72,7 @@ repos:
|
|||
hooks:
|
||||
- id: isort
|
||||
name: Format imports
|
||||
exclude: docs/source-app
|
||||
|
||||
- repo: https://github.com/psf/black
|
||||
rev: 22.6.0
|
||||
|
|
@ -97,11 +98,11 @@ repos:
|
|||
#- mdformat-black
|
||||
- mdformat_frontmatter
|
||||
exclude: |
|
||||
(?x)^(
|
||||
src/pytorch_lightning/CHANGELOG.md|
|
||||
src/lightning_app/CHANGELOG.md|
|
||||
src/lightning_lite/CHANGELOG.md
|
||||
)$
|
||||
(?x)^(
|
||||
src/pytorch_lightning/CHANGELOG.md|
|
||||
src/lightning_app/CHANGELOG.md|
|
||||
src/lightning_lite/CHANGELOG.md
|
||||
)$
|
||||
|
||||
- repo: https://github.com/PyCQA/flake8
|
||||
rev: 4.0.1
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@ Add an Interactive Demo
|
|||
|
||||
.. _add_an_interactive_Demo:
|
||||
|
||||
**Required background:** Basic Python familiarity and complete the :ref:`install` guide.
|
||||
**Required background:** Basic Python familiarity and complete the install guide.
|
||||
|
||||
**Goal:** We'll walk you through the 4 key steps to run a Lightning App that trains and demos a model.
|
||||
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@
|
|||
Build and Train a Model
|
||||
#######################
|
||||
|
||||
**Required background:** Basic Python familiarity and complete the :ref:`install` guide.
|
||||
**Required background:** Basic Python familiarity and complete the guide.
|
||||
|
||||
**Goal:** We'll walk you through the creation of a model using PyTorch Lightning.
|
||||
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@ Start from an ML system template
|
|||
|
||||
.. _go_beyond_training:
|
||||
|
||||
**Required background:** Basic Python familiarity and complete the :ref:`install` guide.
|
||||
**Required background:** Basic Python familiarity and complete the install guide.
|
||||
|
||||
**Goal:** We'll walk you through the 4 key steps to run a Lightning App that trains and demos a model.
|
||||
|
||||
|
|
|
|||
|
|
@ -25,7 +25,7 @@ Here are `the entire App's code <https://github.com/Lightning-AI/lightning-quick
|
|||
Step 1: Install Lightning
|
||||
*************************
|
||||
|
||||
If you are using a :ref:`virtual environment<install_beginner>`, don't forget to activate it before running commands.
|
||||
If you are using a virtual env, don't forget to activate it before running commands.
|
||||
You must do so in every new shell.
|
||||
|
||||
.. tip:: We highly recommend using virtual environments.
|
||||
|
|
|
|||
|
|
@ -11,7 +11,7 @@ Build models and full stack AI apps ⚡ *Lightning fast*.
|
|||
.. join_slack::
|
||||
:align: left
|
||||
|
||||
**Featured examples of what you can build with Lightning:**
|
||||
**Featured examples of what you can do with Lightning:**
|
||||
|
||||
|
|
||||
|
||||
|
|
@ -22,7 +22,7 @@ Build models and full stack AI apps ⚡ *Lightning fast*.
|
|||
|
||||
.. app_card::
|
||||
:title: Develop and Train
|
||||
:description: Train an LLM (512 GPUs)
|
||||
:description: Train an LLM (64 GPUs)
|
||||
:width: 280
|
||||
:image: https://lightning-ai-docs.s3.amazonaws.com/develop_n_train_v1.jpg
|
||||
:preview: https://lightning.ai
|
||||
|
|
@ -34,10 +34,10 @@ Build models and full stack AI apps ⚡ *Lightning fast*.
|
|||
:title: Serve and deploy
|
||||
:description: Production-ready stable diffusion server (<2s latency)
|
||||
:width: 280
|
||||
:app_id: HvUwbEG90E
|
||||
:image: https://lightning-ai-docs.s3.amazonaws.com/serve_n_deploy_v1.jpg
|
||||
:preview: https://lightning.ai
|
||||
:deploy: https://lightning.ai
|
||||
:target: https://apple.com
|
||||
:target: https://01gbx4m78rbkpczdf5cpz2hpbh.litng-ai-03.litng.ai/root.api_component/
|
||||
:tags: App
|
||||
|
||||
.. app_card::
|
||||
|
|
@ -56,9 +56,12 @@ Build models and full stack AI apps ⚡ *Lightning fast*.
|
|||
|
||||
----
|
||||
|
||||
************************
|
||||
Build modular components
|
||||
************************
|
||||
****************************************
|
||||
Build self-contained, modular components
|
||||
****************************************
|
||||
Build modular, self-contained components and plug them into your existing workflows.
|
||||
A Lightning component organizes arbitrary code so it can run on the cloud. A component can train a model, deploy, or even host a web UI.
|
||||
The component manages its own infrastructure, cloud costs, networking and more, so you can focus on application logic and not engineering.
|
||||
|
||||
.. raw:: html
|
||||
|
||||
|
|
@ -66,27 +69,19 @@ Build modular components
|
|||
<img src="https://lightning-ai-docs.s3.amazonaws.com/intro_components.gif" style="max-width: 800px"></img>
|
||||
</div>
|
||||
|
||||
.. raw:: html
|
||||
|
|
||||
|
||||
<div class="row">
|
||||
<div class='col-md-5' style="padding-top: 18px">
|
||||
<p>
|
||||
Build modular, self-contained components that can train and deploy models, host a web UI or run arbitrary code on the cloud.
|
||||
A Lightning Component manages its own infrastructure, cloud costs, networking and more, so you can focus on application logic and not engineering.
|
||||
|
|
||||
|
||||
<br><br>
|
||||
Combine your components and prebuilt ones from <a href="https://lightning.ai/components" target="_blank">our gallery </a>
|
||||
to create full-stack AI apps, ⚡ <i>Lightning fast</i>.
|
||||
</p>
|
||||
</div>
|
||||
<div class='col-md-7'>
|
||||
**Run an example component on the cloud**:
|
||||
|
||||
.. lit_tabs::
|
||||
:code_files: landing_app.py;
|
||||
:highlights: 6
|
||||
:titles: Hello CPU world; Hello GPU (accelerated) world; Train PyTorch on a cloud GPU; Train PyTorch ⚡ on cloud GPUs; Deploy a model on cloud GPUs; Run a model script; XGBoost; XGBoost (GPU accelerated); Build a streamlit demo
|
||||
:code_files: ./levels/basic/hello_components/hello_world.py; ./levels/basic/hello_components/hello_world_gpu.py; ./levels/basic/hello_components/train_pytorch.py; ./levels/basic/hello_components/train_ptl.py; ./levels/basic/hello_components/deploy_model.py; ./levels/basic/hello_components/run_script.py; ./levels/basic/hello_components/xgboost.py; ./levels/basic/hello_components/xgboost_gpu.py; ./levels/basic/hello_components/build_demo.py
|
||||
:highlights: 7; 10, 11; 3, 6;3;4;5; 6, 9; 15, 20, 21; 10, 14, 29
|
||||
:app_id: abc123
|
||||
:height: 250px
|
||||
:code_only: True
|
||||
:tab_rows: 4
|
||||
:height: 550px
|
||||
|
||||
.. lit_tabs::
|
||||
:code_files: landing_app_run.bash
|
||||
|
|
@ -94,12 +89,7 @@ Build modular components
|
|||
:height: 150px
|
||||
:code_only: True
|
||||
|
||||
|
|
||||
|
||||
.. raw:: html
|
||||
|
||||
</div>
|
||||
</div>
|
||||
Explore pre-built community components in `our gallery <https://lightning.ai/components>`_.
|
||||
|
||||
|
|
||||
|
||||
|
|
@ -112,7 +102,7 @@ Build modular components
|
|||
|
||||
.. displayitem::
|
||||
:header: Get started
|
||||
:description: Learn to go from Lightning components to full stack AI apps step-by-step.
|
||||
:description: Learn to build Lightning components step-by-step.
|
||||
:col_css: col-md-12
|
||||
:button_link: levels/basic/index.html
|
||||
:height: 160
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@
|
|||
Installation
|
||||
############
|
||||
|
||||
**Prerequisites**: Use Python 3.8.x or later (3.8.x, 3.9.x, 3.10.x). We also recommend you install in a virtual environment (`learn how <install_beginner.rst>`_).
|
||||
**Prerequisites**: Use Python 3.8.x or later (3.8.x, 3.9.x, 3.10.x). We also recommend you install in a virtual environment (learn how).
|
||||
|
||||
.. lit_tabs::
|
||||
:descriptions: Pip; Macs, Apple Silicon (M1/M2/M3); Windows
|
||||
|
|
|
|||
|
|
@ -2,4 +2,4 @@
|
|||
export GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1
|
||||
export GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1
|
||||
|
||||
python -m pip install -U lightning
|
||||
pip install lightning
|
||||
|
|
|
|||
|
|
@ -1 +1 @@
|
|||
python -m pip install -U lightning
|
||||
pip install lightning
|
||||
|
|
|
|||
|
|
@ -1,3 +1,5 @@
|
|||
:orphan:
|
||||
|
||||
###########################
|
||||
Example: Deploy a model API
|
||||
###########################
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
##########################################
|
||||
Package your code in a lightning component
|
||||
##########################################
|
||||
##############################################
|
||||
Level 1: Package code in a lightning component
|
||||
##############################################
|
||||
|
||||
**Prereqs:** You know *basic* Python.
|
||||
|
||||
|
|
@ -14,10 +14,13 @@ Package your code in a lightning component
|
|||
*********************************
|
||||
Why you need Lightning components
|
||||
*********************************
|
||||
ML workflows and full stack AI apps require many pieces working such as training, deploying, data annotation. However, this tight coupling
|
||||
can lead to monoliths that are hard to scale or many microservices that are hard to monitor, coordinate
|
||||
and scale. A Lightning component is a self-contained piece of code (ie: a microservice) that executes your code
|
||||
and manages its own infrastructure, auto-scaling and costs for you.
|
||||
A Lightning component organizes a piece of code into a self-contained, modular component that
|
||||
can be integrated into your existing workflows or assembled to form a Lightning app.
|
||||
A Lightning component manages its own infrastructure, auto-scaling, cost management, and more, so you
|
||||
can focus on the program logic and not the cloud engineering.
|
||||
|
||||
Components run on the cloud or your laptop without code changes 🤯🤯. Connect components using your current workflow management tools or use
|
||||
Lightning apps to build powerful sequential AND reactive workflows.
|
||||
|
||||
.. raw:: html
|
||||
|
||||
|
|
@ -27,9 +30,7 @@ and manages its own infrastructure, auto-scaling and costs for you.
|
|||
|
||||
|
|
||||
|
||||
By using Lightning components you can focus on application logic without dealing with cloud operations
|
||||
or even how components speak to each other. Together, Lightning components form a Lightning App that
|
||||
gives you these benefits:
|
||||
Organizing your code into Lightning components offers these benefits:
|
||||
|
||||
.. collapse:: Build systems not scripts
|
||||
|
||||
|
|
@ -38,7 +39,6 @@ gives you these benefits:
|
|||
The Lightning structure forces best practices so you don't have to be an expert production engineer.
|
||||
Although it feels like you're writing a script, you are actually building a production-ready system.
|
||||
|
||||
|
||||
.. collapse:: Cost control
|
||||
|
||||
|
|
||||
|
|
@ -54,7 +54,6 @@ gives you these benefits:
|
|||
coding experience. You can write code like you normally do, and the Lightning structure
|
||||
ensures your code is implicitely production ready... even if you're just doing research.
|
||||
|
||||
|
||||
.. collapse:: For experts: Scale with full control
|
||||
|
||||
|
|
||||
|
|
@ -63,6 +62,14 @@ gives you these benefits:
|
|||
scaling logic, fault-tolerance and even pre-provisioning, all from Python. We even give you
|
||||
full flexibility to use tools like `terraform <../../cloud/customize_a_lightning_cluster.html>`_ to optimize cloud clusters for your Lightning apps.
|
||||
|
||||
.. collapse:: Integrate into your current workflow tools
|
||||
|
||||
|
|
||||
|
||||
Lightning components are self-contained pieces of funcionality. Add them to your current workflow
|
||||
tools to quickly fill in gaps in your ML workflow such as monitoring drift, training LLMs and more.
|
||||
You can (optionally) use the Lightning App to integrate components into a cohesive workflow.
|
||||
|
||||
.. collapse:: Packaged code
|
||||
|
||||
|
|
||||
|
|
@ -91,45 +98,25 @@ Install Lightning
|
|||
*****************
|
||||
First, install Lightning.
|
||||
|
||||
.. code:: bash
|
||||
|
||||
python -m pip install -U lightning
|
||||
|
||||
.. collapse:: Mac M1/M2/M3 and Windows users
|
||||
|
||||
|
|
||||
|
||||
**Mac**
|
||||
|
||||
To install on Mac, set these 2 environment variables
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
# needed for M1/M2/M3
|
||||
export GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1
|
||||
export GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1
|
||||
|
||||
python -m pip install -U lightning
|
||||
|
||||
**Windows users**
|
||||
|
||||
To install on Windows:
|
||||
|
||||
- setup an alias for Python: python=python3
|
||||
- Add the root folder of Lightning to the Environment Variables to PATH
|
||||
.. lit_tabs::
|
||||
:descriptions: Pip; Macs, Apple Silicon (M1/M2/M3); Windows
|
||||
:code_files: /install/pip.bash; /install/mac.bash; /install/windows.bash
|
||||
:tab_rows: 4
|
||||
:height: 180px
|
||||
|
||||
----
|
||||
|
||||
**************************
|
||||
Build your first component
|
||||
**************************
|
||||
A Lightning component organizes Python code so it can run on the cloud and be connected with other components to form a Lightning App.
|
||||
Pick one of these components to run:
|
||||
A Lightning component organizes Python code into a self-contained module so it can run on the cloud.
|
||||
|
||||
**Run one of these components!**
|
||||
|
||||
.. lit_tabs::
|
||||
:titles: Hello CPU world; Hello GPU (accelerated) world; Train PyTorch on a cloud GPU; Train PyTorch ⚡ on cloud GPUs; Deploy a model on cloud GPUs; Run a model script; Build a model web UI
|
||||
:code_files: ./hello_components/hello_world.py; ./hello_components/hello_world_gpu.py; ./hello_components/train_pytorch.py; ./hello_components/train_ptl.py; ./hello_components/deploy_model.py; ./hello_components/run_script.py; ./hello_components/build_demo.py
|
||||
:highlights: 7; 10, 11; 3, 7-23;3;4;5;6
|
||||
:titles: Hello CPU world; Hello GPU (accelerated) world; Train PyTorch (cloud GPU); Train PyTorch with Lightning Trainer (cloud GPUs); Deploy a model on cloud GPUs; Run a model script; XGBoost; XGBoost (GPU accelerated); Build a streamlit demo
|
||||
:code_files: ./hello_components/hello_world.py; ./hello_components/hello_world_gpu.py; ./hello_components/train_pytorch.py; ./hello_components/train_ptl.py; ./hello_components/deploy_model.py; ./hello_components/run_script.py; ./hello_components/xgboost.py; ./hello_components/xgboost_gpu.py; ./hello_components/build_demo.py
|
||||
:highlights: 7; 10, 11; 3, 6;3;4;5; 6, 9; 15, 20, 21; 10, 14, 29
|
||||
:app_id: abc123
|
||||
:tab_rows: 4
|
||||
:height: 550px
|
||||
|
|
@ -142,6 +129,7 @@ Components run the same on the cloud and locally on your choice of hardware.
|
|||
:titles: Lightning Cloud (fully-managed); Your AWS account; Your own hardware
|
||||
:code_files: ./hello_components/code_run_cloud.bash; ./hello_components/code_run_cloud_yours.bash; ./hello_components/code_run_local.bash
|
||||
:tab_rows: 4
|
||||
:highlights: ; 5; 0
|
||||
:height: 195px
|
||||
|
||||
----
|
||||
|
|
@ -155,17 +143,17 @@ powerful Lightning app. Here are a few key features available to super-charge yo
|
|||
.. lit_tabs::
|
||||
:titles: 15+ accelerators; Auto-stop idle machines; Auto-timeout submitted work; Use spot machines (~70% discount); Work with massive datasets; Mount cloud storage; Use a custom container
|
||||
:code_files: ./key_features/accelerators.py; ./key_features/idle_machine.py; ./key_features/auto_timeout.py; ./key_features/spot.py; ./key_features/massive_dataset.py; ./key_features/mount_data.py; ./key_features/custom_container.py;
|
||||
:highlights: 10;10;10;10;10;2,6,9, 10; 7
|
||||
:highlights: 11;11;11;11;11;2,7,10, 11; 11
|
||||
:app_id: abc123
|
||||
:tab_rows: 3
|
||||
:height: 430px
|
||||
|
||||
----
|
||||
|
||||
***************************
|
||||
Use the community ecosystem
|
||||
***************************
|
||||
Lightning has a vibrant collection of community-built components you can use as templates or to inspire you.
|
||||
******************************
|
||||
Use community-built components
|
||||
******************************
|
||||
Lightning has a vibrant collection of `community-built components <https://lightning.ai/components>`_ you can use as templates or to inspire you.
|
||||
|
||||
|
||||
----
|
||||
|
|
|
|||
|
|
@ -11,10 +11,13 @@ Level 2: Connect components into a full stack AI app
|
|||
****************************
|
||||
What is a full stack AI app?
|
||||
****************************
|
||||
A full stack AI app coordinates 2 or more `Lightning components <build_a_lightning_component.html>`_ together.
|
||||
We call this system of components interacting with each other a Lightning App.
|
||||
In the ML world, workflows coordinate multiple pieces of code working together. In Lightning,
|
||||
when we coordinate 2 or more `Lightning components <build_a_lightning_component.html>`_ working together,
|
||||
we instead call it a Lightning App. The difference will become more obvious when we introduce reactive
|
||||
workflows in the advanced section.
|
||||
|
||||
In this guide, we'll coordinate 2 components together and explain how it works.
|
||||
For the time being, we'll go over how to coordinate 2 components together in a traditional workflow setting
|
||||
and explain how it works.
|
||||
|
||||
.. note:: If you've used workflow tools for Python, this page describes conventional DAGs.
|
||||
In `level 5 <../intermediate/run_lightning_work_in_parallel.html>`_, we introduce reactive workflows that generalize beyond DAGs
|
||||
|
|
@ -33,10 +36,10 @@ on a separate CPU machine. We save money by stopping the GPU machine when the wo
|
|||
:titles: Import Lightning; Define Component 1; Define Component 2; Orchestrator; Connect component 1; Connect component 2; Implement run; Train; Analyze; Define app placeholder
|
||||
:descriptions: First, import Lightning; This component trains a model on a GPU machine; This component analyzes a model on a CPU machine; Define the LightningFlow that orchestrates components; Component 1 will run on a CPU machine; Component 2 will run on an accelerated GPU machine; Describe the workflow in the run method; Training runs first and completes; Analyze runs after training completes; This allows the app to be runnable
|
||||
:code_files: ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py
|
||||
:highlights: 2; 4-6; 8-10; 12; 15; 16; 18; 19; 20; 22
|
||||
:highlights: 2; 5-7; 9-11; 13; 16; 17; 19; 20; 21; 23
|
||||
:app_id: abc123
|
||||
:tab_rows: 4
|
||||
:height: 450px
|
||||
:height: 460px
|
||||
|
||||
|
|
||||
|
||||
|
|
@ -64,10 +67,10 @@ Without going out of your way, you're now doing the following: (Hint: Click **vi
|
|||
|
||||
.. lit_tabs::
|
||||
:titles: Orchestration; Distributed cloud computing; Multi-machine communication; Multi-machine communication; Multi-cloud;
|
||||
:descriptions: Define orchestration in Python with full control-flow; The two pieces of independent Python code ran on separate machines 🤯🤯; The text "GPU machine 1" was sent from the flow machine to the machine running the TrainComponent; The text "CPU machine 2" was sent from the flow machine to the machine running the AnalyzeComponent; The full Lightning app can move across clusters and clouds
|
||||
:descriptions: Define orchestration in Python with full control-flow; The two pieces of independent Python code ran on separate machines 🤯🤯; The text "CPU machine 1" was sent from the flow machine to the machine running the TrainComponent; The text "GPU machine 2" was sent from the flow machine to the machine running the AnalyzeComponent; The full Lightning app can move across clusters and clouds
|
||||
:code_files: ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./level_2_scripts/hello_app.py; ./hello_components/multi_cloud.bash
|
||||
:tab_rows: 4
|
||||
:highlights: 18-20; 15-16; 19; 20; 2
|
||||
:highlights: 19-21; 16-17; 20; 21; 2, 6, 10
|
||||
:images: <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/orchestration.gif" style="max-height: 430px; width: auto"></img> | <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/distributed_computing.gif" style="max-height: 430px; width: auto"></img> | <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/multi_machine_comms.gif" style="max-height: 430px; width: auto"></img> | <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/multi_machine_comms.gif" style="max-height: 430px; width: auto"></img> | <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/multi_cloud.gif" style="max-height: 430px; width: auto"></img>
|
||||
:height: 450px
|
||||
|
||||
|
|
@ -81,13 +84,30 @@ Although we've abstracted the infrastructure, you still have full control when y
|
|||
.. lit_tabs::
|
||||
:titles: Scheduler; Crontab syntax; Auto-scaling; Organized Python; Full terraform control;
|
||||
:descriptions: Although you can use Python timers, we have a scheduler short-hand; You can also use full cron syntax; Code your own auto-scaling syntax (Lightning plays well with Kubernetes); *Remember* components organize ANY Python code which can even call external non-python scripts such as optimized C++ model servers ;Experts have the option to use terraform to configure Lightning clusters
|
||||
:code_files: ./level_2_scripts/hello_app_scheduler.py; ./level_2_scripts/hello_app_cron.py; ./level_2_scripts/hello_app_auto_scale.py; ./level_2_scripts/organized_app_python.py; ./hello_components/terraform_example.bash;
|
||||
:code_files: ./level_2_scripts/hello_app_scheduler.py; ./level_2_scripts/hello_app_cron.py; ./level_2_scripts/hello_app_auto_scale.py; ./level_2_scripts/organized_app_python.py; ./level_2_scripts/tr.bash
|
||||
:tab_rows: 4
|
||||
:highlights: 22, 23; 22, 23; 20, 23, 26, 27; 8-11, 15-17; 2
|
||||
:highlights: 24; 24; 21, 24, 27, 28; 9, 16, 17; 5
|
||||
:height: 680px
|
||||
|
||||
----
|
||||
|
||||
****************************************
|
||||
Use your own orchestrator (experts only)
|
||||
****************************************
|
||||
If you have your own orchestrator, feel free to coordinate individual Lightning components with
|
||||
those existing tools.
|
||||
|
||||
The Lightning orchestrator described below is optimized for advanced patterns using regular python
|
||||
control-flow and an embedded state system that is discussed in the advanced sections.
|
||||
|
||||
To run the app from a CI/CD job, CLI or other orchestrators, run it without opening the UI (unless you need the UI)
|
||||
|
||||
.. code:: bash
|
||||
|
||||
lightning run app app.py --open-ui=false
|
||||
|
||||
----
|
||||
|
||||
***************************
|
||||
Next step: Build a real app
|
||||
***************************
|
||||
|
|
|
|||
|
|
@ -1,3 +1,5 @@
|
|||
:orphan:
|
||||
|
||||
############################
|
||||
Example: Create a model demo
|
||||
############################
|
||||
|
|
|
|||
|
|
@ -1,3 +1,5 @@
|
|||
:orphan:
|
||||
|
||||
###########################
|
||||
Example: Deploy a model API
|
||||
###########################
|
||||
|
|
|
|||
|
|
@ -1,15 +1,41 @@
|
|||
# A hello world component
|
||||
# pip install streamlit omegaconf scipy
|
||||
# pip install torch
|
||||
# app.py
|
||||
|
||||
import lightning as L
|
||||
import torch
|
||||
from io import BytesIO
|
||||
from functools import partial
|
||||
from scipy.io.wavfile import write
|
||||
import streamlit as st
|
||||
|
||||
|
||||
class YourComponent(L.LightningWork):
|
||||
def run(self):
|
||||
print('RUN ANY PYTHON CODE HERE')
|
||||
class LitStreamlit(L.app.components.ServeStreamlit):
|
||||
def build_model(self):
|
||||
sample_rate = 48000
|
||||
|
||||
model, _ = torch.hub.load(
|
||||
repo_or_dir='snakers4/silero-models',
|
||||
model='silero_tts',
|
||||
speaker="v3_en",
|
||||
)
|
||||
|
||||
return partial(
|
||||
model.apply_tts,
|
||||
sample_rate=sample_rate,
|
||||
speaker="en_0",
|
||||
), sample_rate
|
||||
|
||||
# run on a cloud machine
|
||||
compute = L.CloudCompute("cpu")
|
||||
worker = YourComponent(cloud_compute=compute)
|
||||
app = L.LightningApp(worker)
|
||||
def render(self):
|
||||
st.title("Text To Speech")
|
||||
text = st.text_input("Text:", "Lightning Apps are the best!")
|
||||
|
||||
if text:
|
||||
model, sample_rate = self.model
|
||||
audio_numpy = model(text).numpy()
|
||||
audio = BytesIO()
|
||||
write(audio, sample_rate, audio_numpy)
|
||||
audio.seek(0)
|
||||
st.audio(audio)
|
||||
|
||||
app = L.LightningApp(LitStreamlit())
|
||||
|
|
|
|||
|
|
@ -1,8 +1,6 @@
|
|||
# app.py
|
||||
import torch
|
||||
|
||||
import lightning as L
|
||||
|
||||
import torch
|
||||
|
||||
class PyTorchComponent(L.LightningWork):
|
||||
def run(self):
|
||||
|
|
|
|||
|
|
@ -0,0 +1,22 @@
|
|||
# pip install sklearn xgboost
|
||||
# app.py
|
||||
|
||||
import lightning as L
|
||||
from sklearn import datasets
|
||||
from sklearn.model_selection import train_test_split
|
||||
from xgboost import XGBClassifier
|
||||
|
||||
class XGBoostComponent(L.LightningWork):
|
||||
def run(self):
|
||||
iris = datasets.load_iris()
|
||||
X, y = iris.data, iris.target
|
||||
|
||||
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
|
||||
|
||||
bst = XGBClassifier(verbosity=3)
|
||||
bst.fit(X_train, y_train)
|
||||
preds = bst.predict(X_test)
|
||||
print(f'preds: {preds}')
|
||||
|
||||
|
||||
app = L.LightningApp(XGBoostComponent())
|
||||
|
|
@ -0,0 +1,22 @@
|
|||
# pip install sklearn xgboost
|
||||
# conda install py-xgboost-gpu
|
||||
# app.py
|
||||
import lightning as L
|
||||
from sklearn import datasets
|
||||
from sklearn.model_selection import train_test_split
|
||||
from xgboost import XGBClassifier
|
||||
|
||||
class XGBoostComponent(L.LightningWork):
|
||||
def run(self):
|
||||
iris = datasets.load_iris()
|
||||
X, y = iris.data, iris.target
|
||||
|
||||
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
|
||||
|
||||
bst = XGBClassifier(tree_method='gpu_hist', gpu_id=0, verbosity=3)
|
||||
bst.fit(X_train, y_train)
|
||||
preds = bst.predict(X_test)
|
||||
print(f'preds: {preds}')
|
||||
|
||||
compute = L.CloudCompute('gpu')
|
||||
app = L.LightningApp(XGBoostComponent(cloud_compute=compute))
|
||||
|
|
@ -6,11 +6,6 @@
|
|||
|
||||
build_a_lightning_component
|
||||
connect_lightning_components
|
||||
train_pytorch_on_the_cloud
|
||||
deploy_ai_model_api
|
||||
run_jupyter_notebook_on_the_cloud
|
||||
create_a_model_demo
|
||||
build_a_dag
|
||||
|
||||
############
|
||||
Basic skills
|
||||
|
|
@ -33,7 +28,7 @@ Every new user should start here.
|
|||
|
||||
.. displayitem::
|
||||
:header: Level 1: Package code in a Lightning component
|
||||
:description: Learn to build a Lightning component which is the core building block of a full stack AI app (Lightning App).
|
||||
:description: Learn to build a Lightning component which you can use in existing workflows or turn into a Lightning app.
|
||||
:button_link: build_a_lightning_component.html
|
||||
:col_css: col-md-6
|
||||
:height: 170
|
||||
|
|
|
|||
|
|
@ -1,6 +1,5 @@
|
|||
import os
|
||||
|
||||
import lightning as L
|
||||
import os
|
||||
|
||||
|
||||
class YourComponent(L.LightningWork):
|
||||
|
|
|
|||
|
|
@ -17,7 +17,7 @@ class WorkflowOrchestrator(L.LightningFlow):
|
|||
self.analyze = AnalyzeComponent(cloud_compute=L.CloudCompute('gpu'))
|
||||
|
||||
def run(self):
|
||||
self.train.run("GPU machine 1")
|
||||
self.analyze.run("CPU machine 2")
|
||||
self.train.run("CPU machine 1")
|
||||
self.analyze.run("GPU machine 2")
|
||||
|
||||
app = L.LightningApp(WorkflowOrchestrator())
|
||||
|
|
|
|||
|
|
@ -1,6 +1,5 @@
|
|||
# app.py
|
||||
import subprocess
|
||||
|
||||
import lightning as L
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -0,0 +1,5 @@
|
|||
# custom control for optimized clusters with tools like terraform
|
||||
# are only supported on the enterprise tier (support@lightning.ai)
|
||||
|
||||
# once the cluster is created you can run any app on it
|
||||
lightning run app app.py --cloud my-custom-optimized-cluster
|
||||
|
|
@ -1,3 +1,5 @@
|
|||
:orphan:
|
||||
|
||||
#############################################
|
||||
Example: Develop a Jupyter Notebook component
|
||||
#############################################
|
||||
|
|
|
|||
|
|
@ -1,3 +1,5 @@
|
|||
:orphan:
|
||||
|
||||
###################################
|
||||
Example: Train PyTorch on the cloud
|
||||
###################################
|
||||
|
|
|
|||
|
|
@ -16,7 +16,7 @@ and uses `Gradio <https://gradio.app>`_ to serve it.
|
|||
Step 1 - Installation
|
||||
**********************
|
||||
|
||||
First, you'll need to install Lightning. You can find the complete guide here: :ref:`install`.
|
||||
First, you'll need to install Lightning. You can find the complete guide here.
|
||||
|
||||
Then, you'll need to install the `Lightning Quick Start package <https://github.com/Lightning-AI/lightning-quick-start>`_.
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue