kaizenith
banner
kaizenith.bsky.social
kaizenith
@kaizenith.bsky.social
M&A banking escapee on a Sisyphean quest to reclaim the inner mathematician I forsake

Duke ‘22
Real power-play from Amazon with the way it’s thrusted Nova into the picture. Everyone hung up on the fact that it’s < Sonnet 3.6 when the $$ is the differentiator.

One GPT-4o generation ≈ 1 Generate -> Critique -> Improve loop with a similarly capable model like Nova Pro

OAI’s move now 👀
December 4, 2024 at 6:10 PM
One of the sickest things I’ve seen all year.
Nous Research announces the pre-training of a 15B parameter language model over the internet, using Nous DisTrO and heterogeneous hardware contributed by our partners at Oracle, Lambda Labs, Northern Data Group, Crusoe, and the Andromeda Cluster.

You can watch the run LIVE: distro.nousresearch.com
December 3, 2024 at 4:43 AM
Reposted by kaizenith
So someone asked me yesterday whether LLM-Modulo can be used for information retrieval/factual queries.
My answer: "𝕠𝕗 𝕔𝕠𝕦𝕣𝕤𝕖, 𝕨𝕙𝕒𝕥 𝕕𝕠 𝕪𝕠𝕦 𝕥𝕙𝕚𝕟𝕜 ℝ𝔸𝔾 𝕚𝕤?" Let me elaborate..

www.linkedin.com/posts/subbar...
Subbarao Kambhampati on LinkedIn: 𝐕𝐢𝐞𝐰𝐢𝐧𝐠 𝐑𝐀𝐆 𝐚𝐬 𝐚 𝐤𝐢𝐧𝐝 𝐨𝐟 𝐋𝐋𝐌-𝐌𝐨𝐝𝐮𝐥𝐨 𝐟𝐨𝐫…
𝐕𝐢𝐞𝐰𝐢𝐧𝐠 𝐑𝐀𝐆 𝐚𝐬 𝐚 𝐤𝐢𝐧𝐝 𝐨𝐟 𝐋𝐋𝐌-𝐌𝐨𝐝𝐮𝐥𝐨 𝐟𝐨𝐫 𝐟𝐚𝐜𝐭𝐮𝐚𝐥 𝐪𝐮𝐞𝐫𝐢𝐞𝐬. So someone asked me yesterday whether LLM-Modulo can be…
www.linkedin.com
November 23, 2024 at 3:26 PM
Reposted by kaizenith
November 22, 2024 at 4:43 AM