OpenAI has introduced the OpenAI Microscope, a collection of visualisations of every significant layer and neuron in machine learning (ML) models.
OpenAI has announced on its official blog that Microscope will make it easier to analyse the features that form inside these neural networks. Made with the idea to encourage collaborative environments, OpenAI said it had developed this new tool to understand the behaviour of neurons and their interactions in detail, and share those observations.
Microscope systematically visualises every neuron in several commonly studied vision models, and makes all of those neurons linkable, said the company. Although these models and visualisations were already open source, visualising neurons was a “tedious” job. Microscope will try to change the feedback loop of exploring neurons from minutes to seconds. This quick feedback will help discover unexpected features like high-low frequency detectors in the ongoing circuits project.
Further, making models and neurons linkable will allow immediate scrutiny and further exploration of research making claims about those neurons.
OpenAI said just as biologists often focussed on the study of a few “model organisms,” Microscope focuses on exploring a small number of models in detail.
Our initial release includes nine frequently studied vision models, along with several visualization techniques we’ve found particularly useful in studying them. We plan to expand to other models and techniques in the coming months.
– OpenAI
Image credit: OpenAI