rishabhaiover

I love the fact that Steven Seagal made it.

That's true. The noise is being generated by people who are directly or indirectly incentivized to talk about it.

> coming up with the right projects and producing a vertically differentiated product to what already exists is.

Agreed but not all engineers are involved with this aspect of the business and the concern applies to them.

For the longest time, the joy of creation in programming came from solving hard problems. The pursuit of a challenge meant something. Now, that pursuit seems to be short-circuited by an animated being racing ahead under a different set of incentives. I see a tsunami at the beach, and I’m not sure whether I can run fast enough.

Sorry for not being clear. We had two different CUDA functions, one was for Attention and one was for the MLP. Here's the kernel code: https://github.com/sankirthk/GPT2-Kernel-Fusion/blob/main/ke...

We saw different results of pipelining with the Attention kernel vs the MLP kernel (since MLP W1 has to project the attention results into a much higher dimension, the arithmetic intensity shifts towards compute bound characteristics)

I did an experiment on FlashAttention in Triton to measure the impact of caching tiles in the Shared Memory. Surprisingly, it had a non-monotonic relationship with prefetching these tiles and it was kernel dependent. Attention kernel benefits from prefetching caches while MLP W1 doesn't.

Well said. I don't think anybody's stopping anything. I wish I knew how to prepare for it.

Predictions over historical data in a landscape with fragile priors doesn't seem like a strong metric to me (it's a useful approximation at best)

I've practiced a healthy skepticism of the recent boom but I can't reason why the long horizon time wouldn't stretch to 8 hours or a week worth's of effort from next year. After Opus-4.5, governments and organizations should really figure out a path out of this storm because we're in it now.

After I saw Opus 4.5 search through zig's std io because it wasn't aware of a breaking change in the recent release, I fell in love with claude-code and I don't see a strong enough reason to switch to codex at the moment.

exactly. they need to bring in spotify level of caching of streaming music that it just works if you're in a subway. Constant availability should be table stakes for them.

NATS be trippin, no CAP.

It seems like the demise of the possibility of great art in the next 50 years. Maybe my bias I find everything made by Apple or Netflix almost perfect but not it. Every moment is curated for maximum something, but not the feeling I get I used to get, even with filler episodes in between.

> it is like C in that you can fit the whole language in your head.

This is exactly why I find Go to be an excellent language. Most of the times, Go is the right tool.

Rust doesn't feel like a tool. Ceremonial yet safe and performant.

Interesting concept, though I’m not sure I’d install something called `browserbuddy` on my phone.

Exactly! In the Java ecosystem, your intelligence is measured by how elaborate an interface hell you can conjure just to do CRUD.

I'm shocked by people and state using the crutch of cyber crime or scams to push a totalitarian solution to a problem that is better solved by improved education and targeted campaigns against common security pitfalls.

I abhor any decision that robs even a grain of my individual freedom.

If there's anyone who can sell you to travel, it's Bourdain. Travel deep, travel in mundane streets without the thirst to capture each moment.

I hope Airbus only uses Honeywell or Collins in their newer planes.

While I think there's obvious merit to their skepticism over the race towards agi, Sutskever's goal doesn't seem practical to me. As Dwarkesh also said, we reach to a safe and eventually perfect system by deploying it in public and iterating over it until optimal convergence dictated by users in a free market. Hence, I trust that Google, OpenAI or Anthropic will reach there, not SSI.

The child within me likes to dream and this is the dream I have!

I like to think of RLHF as a technique that I, as a student, used to apply to score good marks in my exam. As soon as I started working, I realized that out-of-distribution generalization can't be only achieved from practicing in an environment with verifiable rewards.

I wish I didn't know how Cursor works. It's a great product for 90% of programmers out there no doubt.

I ain’t paying a penny more than the $20 I already do. I got cracks in my boots, brother.

If you're given a finite context window, what's the most efficient token to present for a programming task? sloppy prompts or actual code (using it with autocomplete)

I suspect Cursor is not the right platform to write code on. IMO, humans are lazy and would never code on Cursor. They default to code generation via prompt which is sub-optimal.

damn, I need a MAX sub for this.

I use claude-code extensively to plan and study for my college using the socrates learning mode. It's a great way to learn for me. I wanted to test the new model's capabilities on that front.

Is this available on claude-code?

I was working with the assumption in this model the attestation is signed by ephemeral keys (OIDC) which would reveal the bad actor or give breadcrumbs. Enough to reduce incentives to hijack packages.

Why can't package managers enforce attestations backed by a transparent log for each commit made to a public repository?