HNHacker News
TopNewBestAskShowJobs

lijok

2,257 karma · joined May 31, 2023

lijok@pm.me
submissionscomments
lijok··on Workers Cache
I don’t have access to my flavor of LLM on the train nor the time or budget to have it do the research and summary for me

Why are we all of a sudden pretending like pre-LLM era blogs were these pristinely well written pieces of art or even that effort and care was put into them? In most cases they were significantly less coherent and incomplete. Don’t get me started on the mess that was the communication of this particular company or one of their competitors like AWS.

lijok··on Workers Cache
Can you elaborate? I read it, found the concepts well explained, walked away better informed.

Responding to alleged slop with more slop doesn’t decrease the total amount of slop on the internet.

lijok··on Parse, Don't Validate – In a Language That Doesn't Want You To
Should they disclose the use of a spellchecker? A translation app? Gramarly? A writing tutor?
lijok··on Parse, Don't Validate – In a Language That Doesn't Want You To
Why should they disclose how much AI was used to write an article?
lijok··on How to Passive-Aggressively Shame People Who Use LLMs Selfishly
Just reply “I gave this to Claude and it didn’t understand what you meant - can you rephrase?”
lijok··on Finding the best dog treat with statistics
Bebop is clearly being paid off by Big Chicken to skew the results
lijok··on Codex logging bug may write TBs to local SSDs
Good programmers know symlinks are not portable
lijok··on Tesco moving 40k server workloads off VMware amid Broadcom's abusive conduct
What’s so special about Java/JVM solutions? What is for example the Go ecosystem missing in comparison?
lijok··on What job interviews taught me about Kubernetes
Which LLMs?
lijok··on Statement on US government directive to suspend access to Fable 5 and Mythos 5
Switching between LLMs takes all of 30 seconds - there will be no hesitation to adopt whichever LLM is performing the best.
lijok··on Software Is Made Between Commits
I swear a lot in my chats with Claude..
lijok··on AWS Bedrock to require sharing data with Anthropic for Mythos and future models
You would be very, very surprised
lijok··on What it feels like to work with Mythos
Hot damn - is that the floor of what you consider useful?
lijok··on Are you expected to run five Python type-checkers now?
ML

Data tooling

Talent pool

Libraries for customers

Brownfield codebases

Academics

I can keep going…

lijok··on Elixir v1.20: Now a gradually typed language
People no can Rust so people no use Rust. Simple as.
lijok··on Uber caps employee AI spending after blowing through budget in four months
Are you describing finetuning?
lijok··on Reverting the incremental GC in Python 3.14 and 3.15
What about the projects that python handles better?
lijok··on Reverting the incremental GC in Python 3.14 and 3.15
We should all go back to writing assembly
lijok··on Cloudflare to cut about 20% of its workforce
Or maybe you don’t understand what it means because you’re not the target audience?
lijok··on Cloudflare to cut about 20% of its workforce
Not really. This is all so new, noone is using it correctly, because noone knows how to yet. We’re all just kind of flailing our arms around with it, but it’s clearly a force multiplier and its increased use is an actionable signal
lijok··on Cloudflare to cut about 20% workforce
Noone knows what the correct structure for this new world looks like. We’ll see what they end up hiring for. But it’s fairly standard to lay off a bunch of people and hire new, rather than retrain, when you need to restructure
lijok··on Cloudflare to cut about 20% of its workforce
Which part of that sentence was confusing? I found it perfectly clear. Their internal AI use is exploding, which is a signal that they need to structure for that, and so they’re laying people off as one of the first steps towards actioning that signal.

Nowhere did they indicate there is less work to do, in fact quite the opposite.

lijok··on Cloudflare to cut about 20% of its workforce
Zirp ended over 4 years ago, what are you talking about, the us economy is collapsing? What? Care to elaborate on any of this?
lijok··on Higher usage limits for Claude and a compute deal with SpaceX
Because it’s allegedly more expensive
lijok··on Mythos is the best cybersecurity news in a decade
It is not possible to make a piece of software completely secure because software sits atop hardware and hardware introduces its own security vulnerabilities that leak into software without possible recourse.
lijok··on Higher usage limits for Claude and a compute deal with SpaceX
Why exactly would the H100s get retired for Blackwells if specifically power and DC permits were the bottleneck?
lijok··on Higher usage limits for Claude and a compute deal with SpaceX
Not millions of dollars per sat. Are you being intentionally obtuse?
lijok··on Higher usage limits for Claude and a compute deal with SpaceX
The current bottleneck is not silicon. There is plenty of silicon locked up in previous gen GPUs that are no longer efficient enough to run relative to newer models. The bottleneck is the economics of owning the older GPU models - which is why all the GPU neoclouds are gonna go bust unless they can get customers to continue renting old GPUs.

The economics are vastly different when opex is near zero for these things

lijok··on Higher usage limits for Claude and a compute deal with SpaceX
Millions of dollars? Where did you get that number from?
lijok··on Higher usage limits for Claude and a compute deal with SpaceX
There are over 10k starlink sats in orbit already. They’re obtaining approval for another ~30k.

So clearly not a problem for them.

← PreviousPage 2 of 30Next →