flamingo-ray 0.1.2

Ray-centric job library for training and evaluation.

flamingo

Getting started

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

run

pip install flamingo-ray

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 If you have an active Conda environment, Poetry should recognize it during installation and install the package dependencies there. This hasn't been explicitly tested with other virtual python environments, but will likely work.

Alternatively, you can use poetry's own environment by running

poetry lock
poetry env use python3.10
poetry install

where python3.10 is your python interpreter.

See the contributing guide for more information on development workflows and/or building locally.

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.