Cherry pick doc PRs #34614 #34615 #34435 #34505 #34617 #34623 #34660 (#…

…34676)

doc/source/_static/js/custom.js

@@ -102,3 +102,22 @@ window.onload = function() {
  localStorage.removeItem("scroll");
  }
 };
+
+// When the document is fully loaded
+document.addEventListener("DOMContentLoaded", function() {
+ // find all the code blocks' copy buttons
+ let codeButtons = document.querySelectorAll(".copybtn");
+ for (let i = 0; i < codeButtons.length; i++) {
+ const button = codeButtons[i];
+ // and add a click event listener to each one for Google Analytics.
+ button.addEventListener("click", function() {
+ gtag("event", "code_copy_click", {
+ "send_to": "UA-110413294-1",
+ "event_category": "ray_docs_copy_code",
+ "event_label": "URL: " + document.URL
+ + " Button: " + button.getAttribute("data-clipboard-target"),
+ "value": 1,
+ });
+ });
+ }
+});

doc/source/ray-contribute/docs.ipynb

@@ -28,28 +28,27 @@
  "cd ray/doc\n",
  "```\n",
  "\n",
- "**Note**: If you are on an Apple Silicon (M1) read the instructions below for installing the dependencies.\n",
+ "**Note**: If you are using Apple Silicon (M1), follow the instructions below before continuing.\n",
  "\n",
- "Make sure you activate the Python environment you are using (e.g. venv, conda, etc.) and then to install the documentation dependencies, run the following command:\n",
+ "Activate the Python environment you are using (e.g., venv, conda, etc.). Install the documentation dependencies, with the following command:\n",
  "\n",
  "```shell\n",
  "pip install -r requirements-doc.txt\n",
  "```\n",
  "\n",
- "Additionally, it's best if you install the dependencies for our linters with\n",
+ "Install the dependencies for our linters to ensure your changes comply with our style guide.\n",
  "\n",
  "```shell\n",
  "pip install -r ../python/requirements_linters.txt\n",
  "```\n",
  "\n",
- "so that you can make sure your changes comply with our style guide.\n",
- "Building the documentation is done by running the following command:\n",
+ "Build the documentation by running the following command:\n",
  "\n",
  "```shell\n",
  "make develop\n",
  "```\n",
  "\n",
- "which will build the documentation into the `_build` directory.\n",
+ "Find the documentation build in the `_build` directory.\n",
  "After the build finishes, you can simply open the `_build/html/index.html` file in your browser.\n",
  "It's considered good practice to check the output of your build to make sure everything is working as expected.\n",
  "\n",

doc/source/ray-overview/use-cases.rst

@@ -5,6 +5,94 @@ Ray Use Cases
 
 This page indexes common Ray use cases for scaling ML. It contains highlighted references to blogs, examples, and tutorials also located elsewhere in the Ray documentation.
 
+.. _ref-use-cases-llm:
+
+LLMs and Gen AI
+---------------
+
+Large language models (LLMs) and generative AI are rapidly changing industries, and demand compute at an astonishing pace. Ray provides a distributed compute framework for scaling these models, allowing developers to train and deploy models faster and more efficiently. With specialized libraries for data streaming, training, fine-tuning, hyperparameter tuning, and serving, Ray simplifies the process of developing and deploying large-scale AI models.
+
+.. figure:: /images/llm-stack.png
+
+Learn more about how Ray scales LLMs and generative AI with the following resources.
+
+.. panels::
+ :container: container pb-3
+ :column: col-md-3 px-1 py-1
+ :img-top-cls: p-2 w-75 d-block mx-auto fixed-height-img
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: https://www.anyscale.com/blog/ray-common-production-challenges-for-generative-ai-infrastructure
+ :type: url
+ :text: [Blog] How Ray solves common production challenges for generative AI infrastructure
+ :classes: btn-link btn-block stretched-link webCrawler
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: https://www.anyscale.com/blog/training-175b-parameter-language-models-at-1000-gpu-scale-with-alpa-and-ray
+ :type: url
+ :text: [Blog] Training 175B Parameter Language Models at 1000 GPU scale with Alpa and Ray
+ :classes: btn-link btn-block stretched-link webCrawler
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: https://www.anyscale.com/blog/faster-stable-diffusion-fine-tuning-with-ray-air
+ :type: url
+ :text: [Blog] Faster stable diffusion fine-tuning with Ray AIR
+ :classes: btn-link btn-block stretched-link webCrawler
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: https://www.anyscale.com/blog/how-to-fine-tune-and-serve-llms-simply-quickly-and-cost-effectively-using
+ :type: url
+ :text: [Blog] How to fine tune and serve LLMs simply, quickly and cost effectively using Ray + DeepSpeed + HuggingFace
+ :classes: btn-link btn-block stretched-link webCrawler
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: https://www.businessinsider.com/openai-chatgpt-trained-on-anyscale-ray-generative-lifelike-ai-models-2022-12
+ :type: url
+ :text: [Blog] How OpenAI Uses Ray to Train Tools like ChatGPT
+ :classes: btn-link btn-block stretched-link chatgpt
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: /ray-air/examples/gptj_deepspeed_fine_tuning
+ :type: ref
+ :text: [Example] GPT-J-6B Fine-Tuning with Ray AIR and DeepSpeed
+ :classes: btn-link btn-block stretched-link antServing
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: /ray-air/examples/dreambooth_finetuning
+ :type: ref
+ :text: [Example] Fine-tuning DreamBooth with Ray AIR
+ :classes: btn-link btn-block stretched-link rayForward
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: /ray-air/examples/stablediffusion_batch_prediction
+ :type: ref
+ :text: [Example] Stable Diffusion Batch Prediction with Ray AIR
+ :classes: btn-link btn-block stretched-link rayForward
+
+ ---
+ :img-top: /images/ray_logo.png
+
+ .. link-button:: /ray-air/examples/gptj_serving
+ :type: ref
+ :text: [Example] GPT-J-6B Serving with Ray AIR
+ :classes: btn-link btn-block stretched-link webCrawler
+
 .. _ref-use-cases-batch-infer:
 
 Batch Inference
@@ -530,13 +618,6 @@ The following highlights feature projects leveraging Ray Core's distributed APIs
  ---
  :img-top: /images/ray_logo.png
 
- .. link-button:: https://www.businessinsider.com/openai-chatgpt-trained-on-anyscale-ray-generative-lifelike-ai-models-2022-12
- :type: url
- :text: [Blog] How OpenAI Uses Ray to Train Tools like ChatGPT
- :classes: btn-link btn-block stretched-link chatgpt
- ---
- :img-top: /images/ray_logo.png
-
  .. link-button:: https://www.anyscale.com/blog/building-highly-available-and-scalable-online-applications-on-ray-at-ant
  :type: url
  :text: [Blog] Highly Available and Scalable Online Applications on Ray at Ant Group

doc/source/serve/getting_started.md

@@ -37,18 +37,16 @@ First, let's take a look at our text-translation model. Here's its code:
 :start-after: __start_translation_model__
 :end-before: __end_translation_model__
 :language: python
-:linenos: true
 ```
 
 The Python file, called `model.py`, uses the `Translator` class to translate English text to French.
 
-- The `self.model` variable on line 8 inside `Translator`'s `__init__` method
+- The `self.model` variable inside `Translator`'s `__init__` method
  stores a function that uses the [t5-small](https://huggingface.co/t5-small)
  model to translate text.
 - When `self.model` is called on English text, it returns translated French text
  inside a dictionary formatted as `[{"translation_text": "..."}]`.
-- The `Translator`'s `translate` method extracts the translated text on
- line 15 by indexing into the dictionary.
+- The `Translator`'s `translate` method extracts the translated text by indexing into the dictionary.
 
 You can copy-paste this script and run it locally. It translates `"Hello world!"`
 into `"Bonjour Monde!"`.
@@ -133,7 +131,6 @@ Here's the full Ray Serve script that we built:
 :start-after: __deployment_full_start__
 :end-before: __deployment_full_end__
 :language: python
-:linenos: true
 ```
 
 We can run our script with the `serve run` CLI command. This command takes in an import path
@@ -201,7 +198,7 @@ For example, let's deploy a machine learning pipeline with two steps:
 You can copy-paste this script and run it locally. It summarizes the snippet from _A Tale of Two Cities_ to `it was the best of times, it was worst of times .`
 
 ```console
-$ python model.py
+$ python summary_model.py
 
 it was the best of times, it was worst of times .
 ```
@@ -212,10 +209,9 @@ Here's a Ray Serve deployment graph that chains the two models together. The gra
 :start-after: __start_graph__
 :end-before: __end_graph__
 :language: python
-:linenos: true
 ```
 
-This script contains our `Summarizer` class converted to a deployment and our `Translator` class with some modifications. In this script, the `Summarizer` class contains the `__call__` method since requests are sent to it first. It also takes in the `Translator` as one of its constructor arguments, so it can forward summarized texts to the `Translator` deployment. The `__call__` method also contains some new code on lines 44 and 45:
+This script contains our `Summarizer` class converted to a deployment and our `Translator` class with some modifications. In this script, the `Summarizer` class contains the `__call__` method since requests are sent to it first. It also takes in the `Translator` as one of its constructor arguments, so it can forward summarized texts to the `Translator` deployment. The `__call__` method also contains some new code:
 
 ```python
 translation_ref = await self.translator.translate.remote(summary)

python/ray/tune/tests/tutorial.py

@@ -12,6 +12,7 @@
 import torch.nn.functional as F
 
 from ray import air, tune
+from ray.air import session
 from ray.tune.schedulers import ASHAScheduler
 # __tutorial_imports_end__
 # fmt: on
@@ -105,7 +106,7 @@ def train_mnist(config):
  acc = test(model, test_loader)
 
  # Send the current training result back to Tune
- tune.report(mean_accuracy=acc)
+ session.report({"mean_accuracy": acc})
 
  if i % 5 == 0:
  # This saves the model to the trial directory