Samuel Synthesized

A Quick Update

I’ve recently completed the scholars program at OpenAI and it’s been an amazing place to journey into deep learning. I’ve explored so many ideas and learned all about PyTorch, transformers, tuning and training models, as well as some of the implications that need to be considered when looking at…

Creating a Text Prediction Dataloader

When I was working on my project, I needed to train a transformer on an unsupervised task. In this case the task was text prediction. For this task I needed to break an input sequence into a source sequence and a target sequence, with the target sequence being…

Moving from Words to Bytes

I’m long overdue for an update of where I’ve been in my OpenAI journey! While learning about sequence and language models, I’ve been interested in sequence based approaches to other modalities, like images, videos and audio. How do you work with and learn from such large data? In…

Greatness cannot be planned

First things first, I found a good work cadence as well as a way to track…

Entering a new world

Reflecting on my last two weeks at OpenAI, it’s been a pretty interesting time…

On Bubbles

I‘ve been thinking about how I’ve only enjoyed you in bubbles. By bubbles I refer to separate spaces of existence, where we are in our own…