aabhay

5 ms·

31.
by aabhay 3mo ago
Not necessarily. Drop semantics are just syntactic sugar, and can thus be aggressively inlined or auto vectorized by the compiler.
32.
by aabhay 3mo ago
you're probably only allowed a subset of the query language to talk to an encrypted table. E.g. only range queries that target a sample size > 5% of rows, etc., with exceptions for searches that hit an exact index such as looking up
33.
by aabhay 3mo ago
What strikes me the most is just how many different tasks are involved in modern model design. It used to be the case that you come up with a new loss function, slight architecture changes, etc., run your train and eval loop, and publish th
34.
by aabhay 3mo ago
tags could be display only with no way of filtering for a certain tag.
35.
by aabhay 3mo ago
Community generated tags seems like the obvious solution to all of this. You could easily give a setting to turn them off, breaks no existing systems, and allows for a broad emergent taxonomy. Only surface tags above X community upvotes exc
36.
by aabhay 3mo ago
> We chose a different consensus algorithm for Meerkat, called QuePaxa, that aims to avoid the “tyranny of timeouts” imposed by protocols like Raft. QuePaxa is a subtle protocol, but here are the highlights. A client can contact any repl
37.
by aabhay 3mo ago
This product feels bad and sloppy, so I’ll give my hypothesis for why this was built: At this point, Anthropic is likely having Claude itself propose and build features autonomously based on providing it with raw user feedback. This could b
38.
by aabhay 3mo ago
Exactly. The spec should go in a git comment, NOT vcs. Or, you keep the document in history for the duration of the PR and remove it post review or pre merge
39.
by aabhay 3mo ago
Yep. Tokio uses it for their tests in CI as well last I checked.
40.
by aabhay 3mo ago
The article says that if your read hits a master there’s no consensus needed. Perhaps I read wrong
41.
by aabhay 3mo ago
Looking at the implementation sketches, this algorithm looks even trickier to implement than Paxos (already a notoriously tricky algorithm to implement) and on top of that, I think the failure case in this algorithm is subtle and different
42.
by aabhay 3mo ago
Its also because the CoT is probably unintelligible
43.
by aabhay 3mo ago
This has very little to do with “AI replacing jobs” and much much more to do with a bad product getting obsoleted by better ones. Human labeling is a two sided marketplace and so as any marketplace startup knows, both sides need to be const
44.
by aabhay 3mo ago
In Gemini at least, if you look at how they process PDFs, they do an OCR and then feed the text + image to the model, without charging you for the text tokens (I believe). So my guess is that Claude’s backend is doing the same — so this hac
45.
by aabhay 3mo ago
Ahhh my eyes the vibe coded readme
46.
by aabhay 3mo ago
The FAQ was exactly why I asked the question, since it made it seem like the answer is no.
47.
by aabhay 3mo ago
Can you speak to what makes this different from simply including or configuring various agent skills? Or is it simply the combination of lots of helpful defaults that makes this product useful?
48.
by aabhay 4mo ago
To all of those thinking that GLM/OSS will save you — keep in mind that the model size needed to compete here likely requires an NVL72 or similar — 72GPU dedicated infra to run a hosted model. This will almost certainly get regulated b
49.
by aabhay 4mo ago
This is equally true of a company with a bad product?
50.
by aabhay 4mo ago
I simply don’t see the value here. At the cost of these systems its inevitable that you hire at least one person full time to manage your rack infra, and that person would likely prefer the customizability of commodity hardware. Source: I r
51.
by aabhay 4mo ago
I think the beauty of this is that you know it can’t possibly have been ai generated
52.
by aabhay 4mo ago
Tauri doesn't lock you in to one JS ecosystem. In fact, it doesn't require you to use javascript at all. Also, we've had several developer framework startups get acquired -- Astro, Nuxt, UV, Bun, Vite. It doesn't exactly
53.
by aabhay 4mo ago
The benefit of Deno Desktop is it's like Tauri except for when you want it to be Electron???
54.
by aabhay 4mo ago
Claude code is a very small fraction of the code written by Anthropic and ultimately, despite being widely used, hardly dents Anthropic’s core priority of LLM performance. It’s because Anthropic doesn’t publish any of its core AI research t
55.
by aabhay 4mo ago
If you look at Anthropic's blogs about their model timelines, there is roughly a ~3m period between a model being in internal preview until release. That means that inside of Anthropic, the next version beyond Mythos/Fable is alre
56.
by aabhay 4mo ago
Yep looks like a go1
57.
by aabhay 4mo ago
I am guessing this is related to Anthropic’s recent acquisition of Coefficient Bio, and their interest on essentially using AGI to discover novel drugs
58.
by aabhay 4mo ago
Machines are historically more obedient than people — so employing millions of AI agents to maintain your empire isn’t as fragile as enslaving millions of people. Historically its been the revolts of people, not the commoditization of resou
59.
by aabhay 4mo ago
Banned Aid
60.
by aabhay 4mo ago
If you want to get inspired by good component DX, try looking at Bevy, the game engine. But essentially it comes down to traits, newtypes/enum variants, and macros.

More ›