quote: hey @JensenHuang @NaderLikeLadder! i'm almost certain you'll find this interesting -- we built a system where anyone can take an open-ended problem, openly improve the best-known solution, and have every result independently verified. we first used it to improve Shor's algorithm -- the work was later published through IEEE. more recently, we made Laguna inference 2.3x faster on MLX -- without speculative decoding or aggressive quantization. we share a lot of the same values as Nvidia around open-weight models and open-source inference -- but right now, hardware is our biggest bottleneck. we'd love to work with Nvidia. it would be incredible if y'all could sponsor some rented GPU capacity so we can run an open optimization challenge for Nemotron -- or any model your team is excited to accelerate. we can also take on broader research problems together. our goal is to help the open-source world explode with an abundance of intelligence -- and make that intelligence useful for solving any problem in the world. | i see a beautiful hill being climbed… open innovation accelerating open intelligence… i really want to now do CUDA & AMD -- would be amazing if someone can sponsor us an entire rack.

quote: hey @JensenHuang @NaderLikeLadder! i'm almost certain you'll find this interesting -- we built a system where anyone can take an open-ended problem, openly improve the best-known solution, and have every result independently verified. we first used it to improve Shor's algorithm -- the work was later published through IEEE. more recently, we made Laguna inference 2.3x faster on MLX -- without speculative decoding or aggressive quantization. we share a lot of the same values as Nvidia around open-weight models and open-source inference -- but right now, hardware is our biggest bottleneck. we'd love to work with Nvidia. it would be incredible if y'all could sponsor some rented GPU capacity so we can run an open optimization challenge for Nemotron -- or any model your team is excited to accelerate. we can also take on broader research problems together. our goal is to help the open-source world explode with an abundance of intelligence -- and make that intelligence useful for solving any problem in the world. | i see a beautiful hill being climbed… open innovation accelerating open intelligence… i really want to now do CUDA & AMD -- would be amazing if someone can sponsor us an entire rack.

Source:https://x.com/eigencloud/status/2083726829551341835