Keep on to blur preview images; turn off to show them clearly
I have never seen in my life a line moving so fast. Kudos to the organizers for having set up the GPU of registration with like 20 desks open. @NeurIPSConf
Research Scientist @meta (FAIR), Prof. @Unige_en, co-founder @nc_shape. I like reality.
New @MistralAI models! Apache 2.0 and available in small sizes: 3B, 8B, and 14B. Perfect for iPhone, iPad, and Mac, respectively. Hoping to be able to add them to @LocallyAIApp soon.
iOS Developer @Match @MeeticGroup. AI & ML enthusiast. Building @LocallyAIApp for iPhone, a private and local AI chat.
To make this happen, we've also started work on Active Record Search with full-text implementations for both SQLite and MySQL (Trilogy). https://t.co/Gi1yaVFB2O
Father of three, Creator of Ruby on Rails + Omarchy, Co-owner & CTO of 37signals, Shopify director, NYT best-selling author, and Le Mans 24h class-winner.
We are a 4.5 months old company, and we have a long way to go. This is solving Erdos problems - mathematical Pokemon hunting. Real mathematics is deep discoveries in all branches of math - especially those that have a deep literature. Onwards!
@axiommathai : careers@axiommath.ai
this is a great question to think about and follows some other interesting research question: 1. is there an in-context learning ceiling? 2. is RL the way to break through that ceiling? 3. how much knowledge is already in the weights that we don’t know how to reliably extract? text based optimization definitely works in some capacity. for example the flow of: Agent Harness —> Eval —> Mine Traces —> Improve Harness lots of interesting experiments to do on sample efficiency, cost/compute, comparing text based and RL based optimization another thing is that post-RL, how much will we again re-tweak your harness which brings you back to Text based optimization again?
building agents and harnesses, prev @awscloud, phd cs @ temple
Nah. Nothing happened. Sleep well, Zvi… (nothing new happened. «A factor of two» price reduction… some more post-training… this was, of course, all baked in. If V3.2-exp didn't pass the triage, why would 3.2?)
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
