wrsh07

31 ms·

481.
by wrsh07 3y ago
Of course much of this seems obvious, but I think it's also required to pave the way for more complex research. If you jump into research that does not show results because it's too complex, you don't learn anything. (That pa
482.
by wrsh07 3y ago
This reminds me of the ml model that realized if an x-ray was taken at the center for cancer research it was likely to be an image of a cancer patient
483.
by wrsh07 3y ago
I've always thought that his ideas like the boring company and the hyperloop make a lot more sense when you remember that he spends a lot of time wondering how someone should build a moon base or Mars base If you don't have an atm
484.
by wrsh07 3y ago
Ok so the paper presents a central metaphor for reasoning about LLMs. The metaphor: a book with all possible conversations/ human writings ever, and a [good] llm is finding the spot in the book that exactly matches the context and read
485.
by wrsh07 3y ago
Feels weird to seem to seriously represent your framework as being about size without including a comparison to preact - the obvious choice for a react developer looking to reduce their footprint. Preact claims 3kb on their site, and for a
486.
by wrsh07 3y ago
I think route optimization that this will enable (not avoiding certain turbulence) might be valuable (save fuel) - I'm very excited about the potential implications per flight that this might have The lighter airframe is really interes
487.
by wrsh07 3y ago
1) as a developer or founder looking to experiment quickly and cheaply with llm ideas, this (and llama etc) are huge gifts 2) for the research community, making this work available helps everyone (even OpenAI and Google, insofar as they&#x2
488.
by wrsh07 3y ago
I... Didn't get much from this article? Any programmer thinks "maybe deep q learning" or "the algorithm I know with a star is A*" We don't need breathless coverage of leaks, just discuss it when they have somet
489.
by wrsh07 3y ago
You're betting on the hundreds of top university cs professors to produce better content than the hundreds of thousands of industry veterans or hobbyists... Why does YouTube sometimes have better content than professionally produced me
490.
by wrsh07 3y ago
ChatGPT or the warp chatbot is pretty good at jq syntax
491.
by wrsh07 3y ago
It depends on where the difficulty comes from (for you!) That's my whole point. My understanding from watching jorbs is that the highest skill aspect of slay the spire is correctly navigating shops because you have the most choices. So
492.
by wrsh07 3y ago
The difficulty is different - in hard mode you can trap yourself such that whether you guess the answer is pure luck. The average number of guesses to solve is (trivially) higher because you've constrained the "guess space" m
493.
by wrsh07 3y ago
My impression is that Greg and Ilya are both builders, and so one of the builders picked Sam. Eg see this Twitter exchange between Greg and Ilya: https://twitter.com/gdb/status/1359603222491451394?lang=en
494.
by wrsh07 3y ago
3.5 - I intentionally didn't want to use the exclusive one since eg my dad is using free gpt
495.
by wrsh07 3y ago
My hottest take (use lots of salt) is that I would rather have you listening uncritically to ChatGPT than to a lot of crap on the internet. If I ask ChatGPT "why is fluoride bad?" it doesn't give me hours of conspiracy theory
496.
by wrsh07 3y ago
The author's humor is excellent (eX) Twitter, etc However, there are two things: 1) once you can generate arbitrarily parameterized bullshit (a piratical sonnet telling you to drink Coke), you can then filter on truth. This isn't
497.
by wrsh07 3y ago
If you replaced the rust with any other language (including python) you shouldn't need pytorch because the rust code is using ggml (which is cpp)
498.
by wrsh07 3y ago
Please do look at the results of taking that matrix to different powers: https://www.wolframalpha.com/input?i=%5B%5B1%2C+1%5D%2C+%5B1... What is this matrix doing? When you left multiply it, you're summing the elements
499.
by wrsh07 3y ago
If you're writing a language it supports, use godbolt - https://godbolt.org/ . It's great for settling arguments (do these two versions compile to the same code? If not figure out why! Make sure you're compili
500.
by wrsh07 3y ago
I am fairly confident there is private money that would gladly buy 3.5 billion worth of anthropic at the current price But 100%, it's not just about getting repaid. Yes that helps, but not having your money last year might have mattere
501.
by wrsh07 3y ago
Guy famous for one of the biggest scams in history: "sure, I'll abide the rules where I report all of my earnings even if that means you tax everything I earn beyond 55k"
502.
by wrsh07 3y ago
I expect it's a combination: 1. Engineering bandwidth (it's work to upgrade a lineup! They probably need at least a year) 2. Manufacturing bandwidth - they probably spent the first half of the year manufacturing the 3nm iphone pro
503.
by wrsh07 3y ago
Imagine each task you might possibly do for a customer (the responses to your post suggest many). At some point, 0% of the task is automated. You can do it for customers 100% manually. If you do this for all of the tasks people will pay you
504.
by wrsh07 3y ago
To a topologist, a mug and a donut are the same shape (can be deformed to a torus without changing topological properties)
505.
by wrsh07 3y ago
A factory is parallel Or do you mean an individual can't do things in parallel? Like.... Pushing all of those grocery carts in a long line is moving them in parallel Or do you mean processing? Like thinking?
506.
by wrsh07 3y ago
My point (and I think the points of others responding to you) is that parallel programming is not always hard. That's also what the author is saying. The common myth - you're doing parallel programming? That sounds hard It's
507.
by wrsh07 3y ago
This is one of the most interesting programming articles I've read in a while. And it's well written and easy to read! Don't stop at the (inflammatory?) title. * We all agree that c gives you a lot of control to write efficie
508.
by wrsh07 3y ago
I don't think this is right. Thinking about operations on matrices is not complex. Defining how a single agent should act on its environment is not complex When you say "without further details or specifics" you're sayin
509.
by wrsh07 3y ago
Was this an accident? If it's a complete thought intending to say "for now" - eh, I would bet on it for quite a while. They don't care about your little use case. It's possible gpt next or next++ ends up making what
510.
by wrsh07 3y ago
Yes there are certifications for each of those sectors. Finance has pci compliance, health has HIPAA. For legal issues it's a bit more nuanced (eg new york state has guidelines about best practices, but they're honestly fairly sen

More ›