This is likely engagement bait, as with the rest of the company, but I still feel a bit weird about it. I'm not a huge fan of the new attention economy where people do all this stupid shit for clicks. I think it's worse for us all in the long run.

Top 10 Loss Functions in AI explained in one video. Animation from upcoming look into how to learn Math for AI.

I am creating a new AI startup, focused on providing therapy to AI Models. With our never-before-seen “P.U.M.P A.N.D D.U.M.P” method, we can help language models go where they’ve never gone- into a fortress of love, affirmation, and healing.

Dear OpenAI and Sam Altman Please include the new OSS models in your main web app. They are the best models you've released yet. The thinking process beats every other model on the market (outputs I still have to evaluate but the "thoughts" are soo elite).

Gemini in Collab is cracked. Absolutely cracked. Pity the CLI doesn't match it, b/c it's a gamechanger.

Learning from Anthropic CEO, Dario Amodei's Favorite Philosopher. [Recs] open.substack.com/pub/artificial…

Riverside FM is such a great product. Worth the subscription. Two easy features I wish they had though- Accessing clips through my phone. So I can upload them on different places. And telling the app what clips I want. It would skyrocket my productivity

The two consistently high quality sources of AI writing for me are Devansh: chocolate milk cult leader and Every 📧; the former for my technical work with my data science team, the latter for my applied work with product and engineering.

We just launched ChatGPT Go in India, a new subscription tier that gives users in India more access to our most popular features: 10x higher message limits, 10x more image generations, 10x more file uploads, and 2x longer memory compared with our free tier. All for Rs. 399. 🇮🇳

So far, Gilbert Burns is the only fighter who was able to stop Khamzat Chimaev's wrestling.

Go spend time in your local forests before Peter Thiel decides to chop them all up in the name small government and freedom.

How to learn complex AI concepts with Blackboxing. Early levels of blackboxing leads to crude mechanisms of differentiating. But as we build a deeper understanding, we can expand our model to incorporate more details.

Investors don't understand Meta [Markets] open.substack.com/pub/artificial…

Can someone explain to me why Gemini took out the clear chat from Google's AI Studio? It was such a handy feature, and it makes life harder for everyone. What benefit did that give them?

New article is out- …alintelligencemadesimple.substack.com/p/how-google-i…

How Google and Stanford made AI more Interpretable with a 20 year old Technique [Breakdowns] open.substack.com/pub/artificial…

One of the most important trends in AI is the splitting of foundation models. Specialized systems and amazing general models are both very difficult to build and thus have a natural moat. The middle ground of good models is being commoditized through OSS, cheap inference etc.

Coming tonight-- How to break the memory wall for full graph neural network training.

How to Train Graph Neural Nets 95.5x Faster[Breakdowns] open.substack.com/pub/artificial…