I’ve tried to spin off tensorflow locally and play with GPT-Neo on my M1 mac. Found some official reference online beforehand, so my intentions should have been supported out of the box without any headaches. Buuut, apparently this doesn’t totally apply for MacOS 12 (Monterey). That’s why I’ve fixed the process and I’m documenting it (for myself) down below. Setting up the env Steps: install xcode & brew MiniForge3 (don’t use Anaconda) -> download the arm64 sh from GitHub setup conda for your shell (fish in my case) create a new virtualenv with conda and switch on it set up tensorflow + tensorflow-metal for GPU (in case you plan to use it afterwards) install jupyter install happytransformer and use it for generating text leveraging EleutherAI/gpt-neo Some detailed steps # set up conda for fish ~/miniforge3/bin/conda init fish # create and use a new env conda create --name tf python=3....
Software developer usually working on distributed systems within various (programming) languages or paradigms.
- experienced with ecommerce systems
- nodeJS, golang, php, python, react, rust, typescript, (?)
- terraform, ansible, packer, jenkins
- MySQL, PGSQL, OracleDB, MongoDB, Redis, ClickHouseDB
- kubernetes, apache kafka, rabbitmq, grafana, prometheus, helm
- arch linux <3, redhat
- AWS, GCP, OCI, Azure
- various cloud related certifications
- yeah, that’s my cat