bjourne

10 ms·

151.
by bjourne 5mo ago
People who think unions don't benefit workers are the flat earthers of the 21st century.
152.
by bjourne 5mo ago
Python, strip ANSI escape codes regexp. Copy-paste regexp. Bam, done. I'm comfortable with delegating such menial, low-value tasks. I can focus on the important tasks. I'm a much stronger coder than the bot but also lazy.
153.
by bjourne 5mo ago
Nope. I want to get the most out of all cpus that will run my code. This is a combinatorial problem that grows exponentially by the number of relevant extensions. So, yes, you need to know the hardware, but accounting for combinations of
154.
by bjourne 5mo ago
Bruh, the "average person" won't buy a riscv-based computer in decades. The average bystander to the riscv project indeed writes high-performance code for their, so far, mostly non-existent or emulated riscv processors.
155.
by bjourne 5mo ago
The average bystander might want to write high-performance code for their risc-v cpu. Then they must know precisely which instructions are available and what the performance implications of using them are. E.g., the difference between a sha
156.
Ask HN: Are toiled rolls the same everywhere?
2 points by bjourne 5mo ago|0 comments
157.
by bjourne 5mo ago
And I'm sure you have never even tried to acquire a PhD :p.
158.
by bjourne 5mo ago
How do you know it's ineffective? Other targeted taxes, such as alcohol tax, tobacco tax, congestion tax, etc., have been shown to be very effective.
159.
by bjourne 5mo ago
They didn't! But corporate development can be soul killing. Free software is really fun if u have time.
160.
The Silence That Meets the Rape of Palestinians(nytimes.com)
12 points by bjourne 5mo ago|1 comments
161.
by bjourne 5mo ago
Other than that? Exactly that!
162.
by bjourne 5mo ago
Eh... I think he puts the LLM down for his own ego's sake (as would I!). Curl may, next to the Linux kernel, be one of the most heavily audited codebases in existence. The LLM found something he and thousands of others missed. It'
163.
by bjourne 5mo ago
https://www.youtube.com/shorts/PkUOl2FRQss
164.
LLM Inference Series: 4. KV caching, a deeper look(medium.com)
1 points by bjourne 6mo ago|0 comments
165.
by bjourne 6mo ago
You will be subject to mandatory facial recognition technology with long-term storage, though. The US may certainly be the worst but Europe is also going in an authoritarian direction.
166.
by bjourne 6mo ago
You can't compare context to memory. Context is simply all the text the LLM can use to generate a likely continuation. Imagine you're a relationship expert and I'm asking you for relationship advice. You don't know me so
167.
by bjourne 6mo ago
Is that site for real? It almost seems like some kind of Monty Pythonesque humour site.
168.
by bjourne 6mo ago
That is false. The first thing everyone (at least everyone in CS---IDK about other fields) looks at are h-indexes, impact factors, number of papers per year, university rankings, and similar metrics. Researchers are most definitely select
169.
by bjourne 6mo ago
Conceptually no different from copy-pasting someone else's code.
170.
by bjourne 6mo ago
The only test that has worked 100% of the time for me is to read the candidate's code. Two hours is enough to precisely estimate the candidate's qualities as a software developer. I never understood why companies waste time with t
171.
by bjourne 6mo ago
Right, and everyone else unaware of this made up "informal credit assignment process".
172.
by bjourne 6mo ago
I see. The informal credit assignment process is something that only runs inside of your head.
173.
by bjourne 6mo ago
What "informal credit assignment"? It's automated and it runs entirely on quantitative data.
174.
by bjourne 6mo ago
Why? For all the automatic academic score tracking systems it doesn't matter one bit if it is Wulf et al. or Wulf and McKee.
175.
by bjourne 6mo ago
So softmax is e^x projection followed by l1 norm. Why is e^x projection useful?
176.
by bjourne 6mo ago
Open fast rest api. Plain html web with no js. NO AI BS! Do it, u profit.
177.
by bjourne 6mo ago
Probably? Reinforcement learning creates bots with specific styles. For example, ChatGPT is very fond of "typically", "unpack this", and "if you want".
178.
by bjourne 6mo ago
Each token is only routed through a few chosen (topk) experts during training. So not all expert weights are updated in the backward pass. Otoh, you may need more training to ensure all experts see enough tokens! I doubt MoE is actually wor
179.
by bjourne 6mo ago
CI catches all kinds of bugs.
180.
Ask HN: Will hardware ever be cheap again?
30 points by bjourne 6mo ago|8 comments

More ›