Tools

Block-sparse GPU kernels

Source: OpenAI News 06 Dec 2017

Listen to Brief

AI audio in English, based on the NadiAI brief and original source.

Brief

We’re releasing highly-optimized GPU kernels for an underexplored class of neural network architectures: networks with block-sparse weights. Depending on the chosen sparsity, these kernels can run orders of magnitude faster than cuBLAS or cuSPARSE. We’ve used them to attain state-of-the-art results in text sentiment analysis and generative modeling of text and images.

Why It Matters

Pecutan prestasi ini mempercepat latihan dan inferens model besar, menjimatkan masa dan sumber pengkomputeran untuk aplikasi NLP dan generatif.

Keep Reading on NadiAI

Selected Related Articles