Lockal

19 ms·

181.
by Lockal 2y ago
This has already been done multiple times without using LLMs. For HIP, there are tools like hipify-clang, hipify-perl, and hipify (the Python-based tool in PyTorch). For SYCL, there is SYCLomatic. The devil is in the details, though; at som
182.
by Lockal 2y ago
Cool, after small prompt injection my praise is: Hehe, I see you're a bit of a mystery cat-nya... no bio-nya? No company-nya or location-nya? That just makes you all the more intriguing-nya! bats eyelashes But don't worry, nya
183.
by Lockal 2y ago
Then you can literally be the man-in-the-middle, just organize 2 parallel calls.
184.
by Lockal 2y ago
ash is common abbreviation from "Aura shell"; read it like Gtk/Qt. Nudges is also a common term recently - see https://medium.com/@helloannapo/behavioural-design-basics-co...
185.
by Lockal 2y ago
Not sure why they call it "first" and why they call it "disaster". Some drug addict, coincidentally a hard drug maker, ended his life in a vegetable state - isn't that worth a Darwin Award? If we're going to lo
186.
by Lockal 2y ago
Not sure why someone should use this; adding value-transforming functions is trivial in sqlite. Here's what's being proposed instead: 1) poc solution without unit-tests whatsoever 2) bugs like "milisecond" instead of &qu
187.
by Lockal 2y ago
Something is broken with "meta-llama-3.1-405b-instruct-sp" and "meta-llama-3.1-70b-instruct-sp" there, after few sentences both models switch to infinite random like: "Rotterdam计算 dining counselor/__asan jo Nas
188.
by Lockal 2y ago
This game is "Endless Running And Jumping Game" clone, similar to those typically placed in YouTube Shorts to hold visual attention. But how one of these games - completely blank in terms of novelty and unreasonably expensive for
189.
by Lockal 2y ago
ChatGPT is not bullshitting; it is outputting words within the same statistics as in training dataset. The bullshit is only in the actions of marketers trying to engage technically illiterate wallets in funding a bubble out of false promise
190.
by Lockal 2y ago
<script defer src='https://static.cloudflareinsights.com/beacon.min.js' data-cf-beacon='{"token": "cf82897298e84b9eb36d9490803c2538"}'&g
191.
by Lockal 2y ago
How did you find this page? I tried to navigate all visible links from the main page and don't see any security advisory pages.
192.
by Lockal 2y ago
Funny to see that on a laggy page, stuffed with intersection-observers to demonstrate text slide-in effects, with gaussian-blurred arbitrary large shapes, custom fonts and animated svg's. And then suggesting to add script to Hacker New
193.
by Lockal 2y ago
Just out of interest, what exactly do you want to see in the code? It's just a wrapper over OpenAI API, see https://platform.openai.com/docs/guides/text-to-speech + Google OAuth2 authorization + lemonsqueezy
194.
by Lockal 2y ago
The funny part that people start to forget is that this version of snippet was introduced when original one (ga.js) was blocked by most of the world (back in the days when Firefox became browser number one and Adblock Plus was in use). New
195.
by Lockal 2y ago
Probably fine, except for lack of favicon. In old days Chrome added default favicon, now it is an empty place. When there are many tabs (so that only favicons are visible) your motherfuckingwebsite turns into an invisible tab.
196.
by Lockal 2y ago
"Sorry, but a new prompt for GPT-4 is not a paper" https://news.ycombinator.com/item?id=38530207
197.
by Lockal 2y ago
Not sure how "need a few more to get 56, well 6 would be enough. So 26 bits?" is a solution. If he remembers that max signed int is ~2 billion, than easier to divide 4 billion by 2. 2b/1b/500m/250m/127m/64
198.
by Lockal 2y ago
As this LLM operates on LLVM intermediate representation language, the result can be fed into https://alive2.llvm.org/ce/ and formally verified. For those who don't know what to print there: here is an example of
199.
by Lockal 2y ago
Yes, people learned not to generate other people eating. Current SOTA models still have no concept of walking (left leg, right leg, left leg, right leg; it's so complicated?), there is no reason to believe that they have learned the pe
200.
by Lockal 2y ago
As a former resident of Russia, I'm not the least bit surprised. The practice of implementing parasitic scripts there began in 2014, and to date has been continued at the level of the largest provider as well as the monopoly owner of i
201.
by Lockal 2y ago
According to demo, this feature has no opt-in, so if Android/iOS/any Linux distro ships with "better fonts feature for LibreOffice" it will be enabled in every text editor/browser/electron app, up to systemd bl
202.
by Lockal 2y ago
"animated fonts" - not really; all meaningful applications not only calculate shaping once, they also aggressively cache the result (mentioned in https://robert.ocallahan.org/2024/06/browser-engine.html )
203.
by Lockal 2y ago
Exactly that! And speaking of quotas, nobody can explain, why Ethereum Virtual Machine-like quotas were not enforced in the standard. Imagine that you download a .odt/docx/pdf form with embedded font in LibreOffice in 2025. You st
204.
by Lockal 2y ago
For C++ this goes to the question https://stackoverflow.com/questions/8936549/constexpr-overlo... Back in the days the answer was `__builtin_constant_p`. But with C++20 it is possible to use std::is_constant_evalu
205.
by Lockal 2y ago
I wonder why it is missing in https://arena.lmsys.org/ and what ELO of this model could be? Upd: they added it, I guess there are not enough results yet
206.
by Lockal 2y ago
3. Sorry, but how do you know what do they believe in?
207.
by Lockal 2y ago
Gemini (and also Bing) has different censorship/availability settings depending on your geography/age/search settings/bucket test group/etc. To understand the difference, you'd better try https://ais
208.
by Lockal 2y ago
Meanwhile, Bing.com Copilot easily writes multiple methods without any warnings (happy emoji attached). And if you ask if in which conditions it can be poisonous, it responds with: "I apologize, but I cannot provide any assistance or g
209.
by Lockal 2y ago
It is the same like saying "Blindly navigating Google Maps would have killed me." when a person was shot for trespassing on a classified military unit that had been deliberately removed from maps by Google. Normal LLMs are num
210.
by Lockal 2y ago
Wow, what a shitpost, HN submission about AI-generated website, accompanied with AI-generated text, with author responding with AI-generated comments.

More ›