Latest News Feed
Tweeted
May 6, 2025
EXO Labs
@exolabs How long before this gets into the training data? x.com/alexocheema/st…
EXO Labs
@exolabs How long before this gets into the training data? x.com/alexocheema/st…
Tweeted
Apr 19, 2025
EXO Labs
@exolabs exo will be at @iclr_conf presenting SPARTA dm @MattBeton if you're going to be there too x.com/alexocheema/st…
EXO Labs
@exolabs exo will be at @iclr_conf presenting SPARTA dm @MattBeton if you're going to be there too x.com/alexocheema/st…
Tweeted
Mar 15, 2025
EXO Labs
@exolabs exo v2 generalises to any workload expressible in @__tinygrad__. This includes cryptography, e.g. FHE We published research on 1-bit FHE inference last year that exploits the matmul-free nature of 1-bit models. Overhead is still massive but a promising direction for private AI. x.com/alexocheema/st…
EXO Labs
@exolabs exo v2 generalises to any workload expressible in @__tinygrad__. This includes cryptography, e.g. FHE We published research on 1-bit FHE inference last year that exploits the matmul-free nature of 1-bit models. Overhead is still massive but a promising direction for private AI. x.com/alexocheema/st…
Tweeted
Mar 11, 2025
EXO Labs
@exolabs The full DeepSeek R1 has 671B parameters @ 8-bit = 671GB. 2 x M3 Ultra 512GB Mac Studio with exo (connected with TB5) should be enough to run it with a long context. x.com/alexocheema/st…
EXO Labs
@exolabs The full DeepSeek R1 has 671B parameters @ 8-bit = 671GB. 2 x M3 Ultra 512GB Mac Studio with exo (connected with TB5) should be enough to run it with a long context. x.com/alexocheema/st…
Tweeted
Mar 5, 2025
EXO Labs
@exolabs Who's building the 2T sparse MoE for this? How big will R2 be? x.com/GillVerd/statu…
EXO Labs
@exolabs Who's building the 2T sparse MoE for this? How big will R2 be? x.com/GillVerd/statu…