Monicaxie (@Monica_XieY) started following @zhiyuanli_ on Sep 27, 2024

RT @M3LWorkshop: 📡Join us at the 2nd workshop on Mathematics of Modern Machine Learning (M3L) at #NeurIPS2024! https://t.co/crLL1KQZMV Sub…

This is happening now! I will be poster #913 presenting this work. Happy to chat if you are interested!

Link to book here: https://t.co/IWMLwzONPP

The shock therapy used in Poland known as “Plan Balcerowicza” was heavily influenced by Jeffrey Sachs (recently in a @theallinpod panel to rave reviews). Balcerowicz’s memoirs are worth a read and hopefully someone will translate them. A win for liberal democracy nation building https://t.co/zMoCadN259

RT @amanmathur_: what does the inside of a website look like? leslie and I explored this in "translucent website" https://t.co/vDQyLVINF1

RT @thejamescad: Day 1: Profound HQ, Union Square NYC. Gonna be spending a lot of time here with @dbabbs. https://t.co/FRDjuGyhdw

RT @arjunkhemani: .@elonmusk (worth $250B) answers why he’s still working: I think it's a good question you asked, because it goes to, lik…

🦅🇺🇸

Claude Console 之前是我使用最高频的 AI 工具之一，主要是因为它的 Prompt Generator 功能，可以帮你直接优化生成一个生产级的 Prompt，之前也向 @oran_ge 兄大力推荐。后面 Claude 把它开源了，flowith 也在产品中集成了 Prompt Generator 模式，你现在可以随便输入一个简单/狗屎的 prompt，AI

Claude Console was my go-to AI tool, mainly for its Prompt Generator. Thanks to @AnthropicAI for open-sourcing it. @flowith_ai now has a Prompt Generator mode - type any prompt, let AI optimize it. Often, a great prompt + Claude 3.5 Sonnet still beats O1-Preview. ⬇️ https://t.co/fVjL9cSlR7

We found you on the app https://t.co/SjeQAULxnF

From the organizing committee and everyone involved we hope that you had an amazing time at ICML https://t.co/UIK68Wr2OZ

RT @RWKV_AI: RWKV.cpp - is now being deployed to half a billion systems worldwide Making it one of the world's most widely deployed, truly…

RT @jandotai: Integration to run a GGUF model in Jan? Oh, not needed! - Find the GGUF model on @huggingface - Copy the link & paste it in…

Code is a (graph) data structure. Storing it in text files is convenient for IDEs, but has many downsides: ❌ Import hell ❌ Large files ❌ Poor tree shaking ... etc. I think we'll soon eliminate the "file" abstraction and instead operate directly on the symbol graph.

Terence Tao is giving commentary on o1's math capabilities on Mastodon, and has mixed but overall optimistic takeaways World's foremost expert is seriously evaluating a model vs. his PhD students w/ mixed results. Rate of improvement is incredible. https://t.co/CsGf5weNX2 https://t.co/tjqJVT8dxt

Something I don’t understand about test time compute: why steer MCTS with RM vs. steering MCTS with logprobs on a PPO’d model?

I’m enjoying Wayward Pines atm https://t.co/V1h3OwuseG

world, meet world labs! language is only one facet of human intelligence -- even more visceral is is spatial intelligence, or understanding and reasoning about the physical world. we @radicalvcfund are thrilled to partner with this truly world-class team - @drfeifei @jcjohnss

can't say enough about this team and the opportunity they're tackling. go @Nominal_io ! 🚀

RT @molwelch: world, meet world labs! language is only one facet of human intelligence -- even more visceral is is spatial intelligence, o…

RT @annbordetsky: I remember going to an AI talk by @drfeifei about 7 years ago, long before this GenAI wave... it was mind blowing Super…

5/ All evals are here: https://t.co/NlMKVjYx7s

6/ Read our full blog post for more! https://t.co/pDDMp8uv8v

RT @LeptonAI: Here is the fastest diffuser in the west https://t.co/MlGIBicC6k by collaboration with @hippoml_com 😝 #sdxl #diffusion #GPU

RT @KubernetesPod: What did you do for your internship? Xiang Li (@xiangli0227) created the distributed key-value store @etcdio. #etcd n…

RT @gravicle: #LumaDreamMachine 1.5 is here and it's a great leap in text-to-video quality and intelligence! This new model understands tex…

RT @PatronusAI: 1/ Introducing Lynx v1.1: an 8B State-of-the-Art RAG hallucination detection model 🚀 - Beats Claude-3.5-Sonnet on HaluBenc…

too many people let their first 25 years define them. you can get cracked, master an instrument, land your first dunk, publish a paper, whatever. 10k hours is only 2 years of grinding.

altera beta players have now created thousands of agent friends to play games with. here i create, play with, and get killed by ironman within two minutes. https://t.co/2dMxLy1HuQ

Great work resonating our recent research on agent learning (https://t.co/7fe7yOYYBn). Love to see more inspiring works on this new research direction!

RT @rongwu_xu: ☕️New paper 👉Our latest paper delves into LLMs' ability to perform safety self-correction, namely COURSE-CORRECTION. In th…