Latent Space: The AI Engineer PodcastLatent.SpaceAll episodes

FlashAttention 2: making Transformers 800% faster w/o approximation - with Tri Dao of Together AI

55 min

Wed 26 Jul 2023, 16:463 years ago

AboutTechnology & AI

Hosted by

FlashAttention was first published by Tri Dao in May 2022 and it had a deep impact in the large language models space. Most open models you’ve heard of (RedPajama, MPT, LLaMA, Falcon, etc) all leverage it for faster inference. Tri came on the podcast t...
Full show notes and links

Episode page ↗

Also about technology and AI

La banca centrale dell’AI ha perso il suo governatore
The AI central bank has lost its governor

9 hours agoBlack Box - La scatola nera della finanza15 min
De 7 Extra | Exit Jensen Huang: einde van een tijdperk bij AI-boegbeeld Nvidia
De 7 Extra | Exit Jensen Huang: end of an era at AI flagship Nvidia

9 hours agoDe 724 min
Raport o sztucznej inteligencji - Moja szefowa AI
Report on artificial intelligence - My AI boss

10 hours agoRaport o stanie świata Dariusza Rosiaka1h 18m
America, power and markets after the midterms | Morning Finance

14 hours agoBlack Box - La scatola nera della finanza27 min
Why is Google Holding Schools Hostage?

15 hours agoThe Cal Newport Show20 min
Liquidating a whale by hand, 22 times

16 hours agoComplex Systems with Patrick McKenzie (patio11)22 min

Every episode about Technology & AI

More from Latent Space: The AI Engineer Podcast

LLMs Everywhere: Running 70B models in browsers and iPhones using MLC — with Tianqi Chen of CMU / OctoML

Aug 2023 · 3 years ago52 min
[AI Breakdown] Summer AI Technical Roundup: a Latent Space x AI Breakdown crossover pod!

Aug 2023 · 3 years ago59 min
Llama 2: The New Open LLM SOTA (ft. Nathan Lambert, Matt Bornstein, Anton Troynikov, Russell Kaplan, Whole Mars Catalog et al.)

Jul 2023 · 3 years ago1h 20m
AI Fundamentals: Datasets 101

Jul 2023 · 3 years ago1h 01m