Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
est
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
9 ms
·
181.
▲
by
est
7mo ago
This brings back memories back in the days I mess around Resource Hacker for Win32 EXEs. I miss those days where you can tweak all kinds of software GUI by your self. Change icons, menus, shortcut keys, etc.
182.
▲
Tmux enables AIs to operate servers safely
(blog.est.im)
4 points
by
est
7mo ago
|
0 comments
183.
▲
by
est
7mo ago
also kagi exclude non-English sites. Sad for mixed language blogs like mine.
184.
▲
by
est
7mo ago
> proves nothing remotely like the question that was asked I am not an expert, but by my understanding, the paper prooves that a computationally bounded "observer" may fail to extract all the structure present in the model in o
185.
▲
by
est
7mo ago
> the architect → developer → reviewer pipeline actually produces better results than just... talking to one strong model in one session? There's a 63 pages paper with mathematical proof if you really into this. https://a
186.
▲
by
est
7mo ago
idk if this was the exact quote but: H.R. McMaster: Trump’s knowledge was like a series of islands. He might know a lot about one specific thing, but there were no bridges between the islands, no way to connect one thought to another
187.
▲
Show HN: Realtime Chat with Curl Command
(chat.est.im)
1 points
by
est
7mo ago
|
0 comments
188.
▲
by
est
7mo ago
I wonder what happens if Claude was exploited by hackers and all chat logs were released one day. Nearly every corp secrets would be instant leaked.
189.
▲
by
est
7mo ago
> quantized a 100B parameter model to 1 trit I had the same question, after some debates with Chatgpt, it's not the "quantize" for post-training we often witness these days, you have to use 1 trit in the beginning since pr
190.
▲
by
est
7mo ago
> I tapped into Pangram. Pangram is a remarkably good, conservative model for detecting LLM-generated text I tried it against some of my AI generated articles. It says 100% human Turns out if one manually write a structure and a core id
191.
▲
by
est
7mo ago
Seems a bit complicated. Why not use git for social networking ;) https://github.com/est/gitweets
192.
▲
by
est
7mo ago
Even if you have open training, the corpus were compiled from millions of sources, labeled by experts manually, by AI, by outsourcing, etc. Is it "open by first principle" ?
193.
▲
by
est
7mo ago
I don't understand, how exactly does `npm install github:cline/cline#b181e0` work? b181e0 is literally a commit, a few deleted lines. npm could parse that as a legit script ???
194.
▲
by
est
7mo ago
`rm -rf` is a bit harsh. Let's do `chmod -R 000 /` instead.
195.
▲
by
est
7mo ago
so could software engineering sommehow catch those crashes?
196.
▲
by
est
7mo ago
I won't call that forging, but commission. btw you can make git commits with AI as author and you as commiter. Which makes git blame easier
197.
▲
by
est
7mo ago
Uh, patricide? The key leap from gpt3 to gpt-3.5 (aka ChatGPT) was code-davinci-002, which is trained upon Github source code after OpenAI-Microsoft partnership. Open source code contributed much to LLM's amazing CoT consistency. If th
198.
▲
by
est
7mo ago
big corp politics. qwen started as the core team of Tongyi dept, which was part of the algorithm & model offerings from Alibaba-Cloud (aliyun) https://tongyi.aliyun.com/ - qwen series for nlp/LLM - wan series for c
199.
▲
by
est
7mo ago
on a side note you can use many surprising non-standard HTTP verbs, but many CDNs like Cloudflare filter them
200.
▲
by
est
7mo ago
That's a fair point. And CRM system should take notes. Not everyone lives a 9 to 5 schedule.
201.
▲
by
est
7mo ago
how so? Eastern parts of China gets up at 05:00 AM and westtern part gets up at 10:00 AM. People get used to it.
202.
▲
by
est
7mo ago
is this somehow related ? https://www.anthropic.com/research/persona-selection-model
203.
▲
by
est
7mo ago
History rhymes. The same happened in the 30s to Albert Einstein, Max Born, Lise Meitner, Hans Bethe, Edward Teller, Leo Szilard, etc.
204.
▲
by
est
7mo ago
like how WWII started after excluding Jewish scientists?
205.
▲
by
est
7mo ago
I used to avoid talking to people because it always turns out to be an argument Later I realized this is wrong on my part, talking is all about talking, let the vibe continue and don't let it die.
206.
▲
by
est
7mo ago
obligatory: git notes Lots of comments mentioned this, for those who aren't aware, please checkout Git Notes: Git's coolest, most unloved feature (2022) https://news.ycombinator.com/item?id=44345334 I think it&#x
207.
▲
by
est
7mo ago
Why do I need to download & run to checkout? Can I just submit my gear spec in some dropdowns to find out?
208.
▲
by
est
7mo ago
>Can someone explain what the hell is going on here? Someone at Chromium team is launching rapidly for an promotion
209.
▲
by
est
7mo ago
I think one of the reasons why AI companies are valued this high is one can actually inspect what user inputs & outputs are. It's basically an OSINT siphon. In this Chinese case, the tokens are leaked at least twice. ChatGPT offers
210.
▲
by
est
8mo ago
It says 富士康科技, Foxconn Tech
More ›