An underappreciated fact about concept embeddings : those that are useful across a wide range of tasks must by necessity be relatively low-dimensional, because you need the space to be as structured as possible. You need simple, clear relationships between elements.

I just learned my student/TA/RA got admitted to MIT's PhD program. This gave me an idea of how Acalonia will support grad school applicants across disciplines: It will run free live online statistics courses and help top performers join top grad schools. DM me if interested

匹臼尺弓三刀

Sharing this amazing graphic drawn by Oleg Avdeev I predict this will become the canonical graphic in all MLOps talks going forward

My past students got to PhD programs at Stanford, MIT, Princeton, Columbia, or UTokyo. You may plan to apply to PhD or master's programs or for postdocs where statistics is relevant, such as CS, ML/AI, biology, or economics. Here is how I and Acalonia can support you. (1/3)

Somehow managed to shake my way up 1505 places to get the last gold spot in the Jigsaw Rate Severity of Toxic Comments competition!

I started entering Kaggle competitions about a year ago, and was recently able to reach the rank of Competitions Master, so I decided to write a blog about my experiences and a few things I've learned so far: amontgomerie.github.io/2022/02/10/kag…

The level at which AI researchers discuss philosophy is almost as bad as the level at which philosophers discuss AI

Looking forward to this!

However, most tech workers don’t face those circumstances. Most meetings are not amazing brainstorming sessions in the honeymoon stage of your project. Most tech workers are not millionaires who can afford to live anywhere they want. 2/

Hi everyone 👋 I am #OpenToWork ! 😍 👀 Looking for a mid-career ML/DL/CV Engineering positions, preferably in Tokyo. 👉 Check out my portfolio, CV (alisher-ai.github.io) and LinkedIn (linkedin.com/in/alisher-ai). 🙏 Please let me know if you have suggestions and Retweet 🙌

Hamel Husain on a slightly related note, tools and pokemon are sounding similar these days (i played up to like diamond and pearl) 😅 pixelastic.github.io/pokemonorbigda…

I refuse to be delegated to the "skeptics box" in someone else's framing of a debate. Here is my response to Steven Johnson 's NYT Magazine article about LLMs and OpenAI. On NYT Magazine on AI: Resist the Urge to be Impressed link.medium.com/rag8kbGujpb

that’s what i have been trying to tell you. these systems confabulate spread misinformation constantly because they have no conception whatsoever of truth. *all* they know is the probabilities of word sequences. everything else is anthropomorphism.