RD2024-10: Dev workflow setup and guides

.github/workflows/pr_checks.yaml → .github/workflows/main.yaml

@@ -1,34 +1,44 @@
-name: PR Checks
+name: Tests
 
-on: [push]
+on:
+  push:
+    branches:
+      - "main"
+  pull_request:
+    branches:
+      - "**"
 
 jobs:
   pytest_ruff:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up Python 3.10
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: "3.10"
 
+      - name: Install poetry build tool
+        run: |
+          pip install poetry
+
       - name: Install test dependencies
         run: |
-          pip install -r requirements/test.txt
+          poetry install --only dev
         continue-on-error: true
 
       - name: Lint with Ruff
         run: |
-          ruff --output-format=github .
+          poetry run ruff --output-format=github .
         continue-on-error: false
 
       - name: Install full dependencies
         run: |
-          pip install ".[all]"
+          poetry install
         continue-on-error: true
 
       - name: Run unit tests
         run: |
-          pytest
+          poetry run pytest
         continue-on-error: false

.gitignore

@@ -157,7 +157,10 @@ cython_debug/
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+.idea/
 
 # Ruff
 .ruff_cache
+
+# Poetry
+poetry.lock

CONTRIBUTING.md

@@ -0,0 +1,45 @@
+# Contributing
+
+## Code style
+
+This repository uses [Ruff](https://docs.astral.sh/ruff/) for Python formatting and linting.
+Ruff should be installed automatically in your environment as part of the package's
+development dependencies.
+
+You can execute Ruff by calling `ruff --fix .` or `ruff format .` from the workspace root.
+Ruff will pick up the configuration defined in the `pyproject.toml` file automatically.
+
+## Testing a development branch
+
+`flamingo` is intended to be installed as a pip requirement in the runtime environment of a Ray job.
+However, it is often desirable to test local branches on Ray before publishing a new version of the library.
+This is possible submitting a Ray job with a runtime environment that points to your 
+development branch of the `flamingo` repo.
+
+To do so, follow the steps:
+
+1. Export a copy of the package dependencies by running:
+
+    ```
+    poetry export --without-hashes --with finetuning,evaluation -o requirements.txt
+    ```
+
+    The following command will create a `requirements.txt` file in the repository 
+    that contains the dependencies for the `finetuning` and `evaluation` job groups:
+
+2. When submitting a job to cluster, specify in the Ray runtime environment the following:
+
+    - `py_modules`: Local path to the `flamingo` module folder (located at `src/flamingo` in the workspace).
+    - `pip`: Local path to the `requirements.txt` file generated above.
+
+3. Submit your job with an entrypoint command that invokes `flamingo` directly as a module, eg:
+
+    ```
+    python -m flamingo run finetuning --config config.yaml
+    ```
+
+    This is necessary because `py_modules` uploads the `flamingo` module
+    but does not install its entrypoint in the environment path.
+
+An example of this workflow can be found in the `examples/dev_workflow.ipynb` notebook.
+

README.md

@@ -4,20 +4,56 @@
     <img src="https://github.com/mozilla-ai/flamingo/blob/main/assets/flamingo.png" width="300">
 </p>
 
-## Installation
+## Getting started
 
-Install the package for local development in your chosen Python environment by running:
+### Minimum Python version
 
+This library is developed with the same Python version as the Ray cluster
+to avoid dependency/syntax errors when executing code remotely.
+Currently, installation requires Python between `[3.10, 3.11)` to match the global
+cluster environment (Ray cluster is running 3.10.8).
+
+### Installation
+
+This project is built using the [Poetry](https://python-poetry.org/docs/) build tool.
+First, install Poetry in your local environment via
 ```
-pip install -e ".[all]"
+curl -sSL https://install.python-poetry.org | python3 - -y
 ```
+or see the [installation guide](https://python-poetry.org/docs/#installation)
+for more instructions.
 
-Dependency groups are defined for the logical job groups accessible from the library.
-See `pyproject.toml` for exact information.
+Once Poetry is installed, you can install `flamingo` for development by running
+```
+poetry lock
+poetry install
+```
+This will install an editable version of the package along with all of its dependency groups.
+Poetry should recognize your active virtual environment during installation
+and install the package dependencies there.
 
-### Python version
+The `pyproject.toml` file defines dependency groups for the logical job types in the package.
+Individual dependency groups can be installed by running 
+`poetry install --with <group1>,<group2>` or `poetry install --only <group>`.
 
-This library is developed with the same Python version as the Ray cluster
-to avoid dependency/syntax errors when executing code remotely.
-Currently, installation requires at least Python 3.10 to match the global
-cluster environment (Ray cluster is running 3.10.8).
+See the [contributing](CONTRIBUTING.md) guide for more information on development workflows.
+
+### Usage
+
+`flamingo` exposes a simple CLI with a few commands, one for each Ray job type.
+Jobs are expected to take as input a YAML configuration file 
+that contains all necessary parameters/settings for the work.
+See the `examples/configs` folder for examples of the configuration structure.
+
+Once installed in your environment, usage is as follows:
+```
+# Simple test
+flamingo run simple --config simple_config.yaml
+
+# LLM finetuning
+flamingo run finetuning --config finetuning_config.yaml
+
+# LLM evaluation
+flamingo run lm-harness --config lm_harness_config.yaml
+```
+When submitting a job to Ray, the above commands should be used as your job entrypoints.

examples/configs/simple_config.yaml

@@ -0,0 +1 @@
+magic_number: 42

examples/dev_workflow.ipynb

@@ -0,0 +1,161 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "123e34e9-70f8-42ab-b790-b59ddc01b1f3",
+   "metadata": {},
+   "source": [
+    "# Development Workflow"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "8c0f15ed-77dc-44ce-adb6-d1b59368f03c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Required imports\n",
+    "import os\n",
+    "from pathlib import Path\n",
+    "from ray.job_submission import JobSubmissionClient\n",
+    "\n",
+    "# flamingo should be installed in your development environment\n",
+    "import flamingo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "969884e5-d815-42d9-9d4e-3b8f890657e2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Create a submission client bound to a Ray cluster\n",
+    "# Note: You will likely have to update the cluster address shown below\n",
+    "client = JobSubmissionClient(f\"http://10.146.174.91:8265\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "3258bb97-d3c6-4fee-aa0c-962c1411eaa7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Determine local module path for the flamingo repo\n",
+    "flamingo_module = Path(flamingo.__file__).parent"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "id": "1db3b9aa-99a4-49d9-8773-7b91ccf89c85",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "SimpleJobConfig(magic_number=42)"
+      ]
+     },
+     "execution_count": 19,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Load and inspect the config file\n",
+    "# Not mandatory for job submission, but helpful when debugging\n",
+    "from flamingo.jobs.simple import SimpleJobConfig\n",
+    "\n",
+    "CONFIG_DIR = Path(\"configs\")\n",
+    "CONFIG_FILE = \"simple_config.yaml\"\n",
+    "\n",
+    "config = SimpleJobConfig.from_yaml_file(CONFIG_DIR / CONFIG_FILE)\n",
+    "config"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "b81b36be-35ce-4398-a6d4-ac1f719f5c95",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Construct the runtime environment for your job submission\n",
+    "# py_modules contains the path to the local flamingo module directory\n",
+    "# pip contains an export of the dependencies for the flamingo package (see CONTRIBUTING.md for how to generate)\n",
+    "runtime_env = {\n",
+    "    \"working_dir\": str(CONFIG_DIR),\n",
+    "    \"env_vars\": {\"WANDB_API_KEY\": os.environ[\"WANDB_API_KEY\"]}, # If running a job that uses W&B\n",
+    "    \"py_modules\": [str(flamingo_module)],\n",
+    "    \"pip\": \"/path/to/flamingo/requirements.txt\"\n",
+    "}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "4bd300f9-b863-4413-bd3a-430601656816",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "2024-01-20 15:32:25,620\tINFO dashboard_sdk.py:385 -- Package gcs://_ray_pkg_ba0036a72fdb32af.zip already exists, skipping upload.\n",
+      "2024-01-20 15:32:25,814\tINFO dashboard_sdk.py:385 -- Package gcs://_ray_pkg_8f96eb40a239b233.zip already exists, skipping upload.\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "'raysubmit_tWfixDMGHavrhHPF'"
+      ]
+     },
+     "execution_count": 17,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Submit the job to the Ray cluster\n",
+    "# Note: flamingo is invoked by 'python -m flamingo' since the CLI is not installed in the environment\n",
+    "client.submit_job(\n",
+    "    entrypoint=f\"python -m flamingo run simple --config {CONFIG_FILE}\",\n",
+    "    runtime_env=runtime_env\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3c82892d-bcdf-42e6-b95e-2393e01ab7d6",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "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": "3.10.13"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}