Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ed
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
18 ms
·
211.
▲
by
ed
3y ago
There was a period in the 90’s when it was necessary to raise money and assemble a team just to make web products. Frameworks didn’t exist, we didn’t have the patterns we do now, everything was built for the first time and as such was 100%
212.
▲
by
ed
3y ago
Google has a cached copy but you’ll need to copy and paste the URL http://webcache.googleusercontent.com/search?q=cache%3Ahttps... TL;DR - it’s 13% faster, and better at providing relevant context for the LLM (suggestion ac
213.
▲
by
ed
3y ago
You didn’t specify but I assume you are only required to add [shopify] (exact match) as a negative KW, which is very different from shopify (broad match).
214.
▲
by
ed
3y ago
Atlassian has this policy for 3rd-party developers too. It feels pretty reasonable, to be honest. Developers still get to advertise against phrases relevant to their niche (and I have to imagine the keyword “shopify” would be a poor perform
215.
▲
by
ed
3y ago
This question becomes more relevant when you consider the vision pro is essentially a head-mounted iPad. I wish the author would say how they tried to get a working development environment. E.g.: VSCode Server? https://code.visua
216.
▲
by
ed
3y ago
These could train GPT-3 in 46 hours. (Edited from "11 mins" per https://news.ycombinator.com/item?id=36500154 ) An equivalent number of V100's (GPT-3's original GPU) would've taken about 36 days [0]
217.
▲
by
ed
3y ago
https://web.archive.org/web/20230615202717/https://www.sfchr...
218.
▲
by
ed
3y ago
https://web.archive.org/web/20230602051336/http://jsomers.ne...
219.
▲
by
ed
3y ago
The model was open source before the weights leaked
220.
▲
by
ed
3y ago
> Even worse he never tells us the number of humans being examined You might want to actually open the paper. N = 282,541. They have data for 98% of the population of wales. Of the ~3m people there, 282,541 were in the relevant age brack
221.
▲
by
ed
3y ago
Wish it were possible to “deploy” an example agent without signing up. Even if the output were faked, it would help a lot to understand what this does. (I assume it’s “Geocities for AI agents” but couldn’t confirm.)
222.
▲
by
ed
3y ago
For context https://www.metabunk.org/threads/debunked-rumsfeld-says-2-3-...
223.
▲
by
ed
3y ago
This isn’t free?! Feels like 1Password is the main beneficiary in this relationship. Consider: 1pass owns the auth, and who knows how they’ll wield that power (need to deliver a return on their massive $920m VC round!), the burden to educat
224.
▲
by
ed
3y ago
Nit, but the 6TB version includes a lot of forks and duplicated code so I assume StarCoder was trained on the deduped version, which is 2.9TB.
225.
▲
by
ed
3y ago
Cool! The DX is tricky to nail, when combined with LLM's tendency to hallucinate. I asked it to extract some text from an image, which it dutifully tried to do. However the generated python kept throwing errors. There's no image -
226.
▲
by
ed
3y ago
They do not own your content. > from the tos: “non-exclusive” A non-exclusive right and license means you can still grant similar rights to other parties, and you are not restricted from using the content yourself or licensing it to othe
227.
▲
by
ed
4y ago
This is an April fools joke. From the linked paper: > It is important to note that the concept of GPTrillion and its capabilities are entirely fictional and were created as an April Fools joke for the AI community. GPTrillion does not ac
228.
▲
by
ed
4y ago
Thanks for the quote. Friendly reminder to folks to use the Flag button to get this off the front page.
229.
▲
by
ed
4y ago
Here’s the (empty, for now) GitHub repo - https://github.com/microsoft/JARVIS
230.
▲
by
ed
4y ago
This. And “What did you do right?” is often informative, regardless of overall success.
231.
▲
Silicon Valley Bank profit squeeze in tech downturn attracts short sellers
(ft.com)
2 points
by
ed
4y ago
|
0 comments
232.
▲
by
ed
4y ago
Wow, that is super helpful. Thank you for releasing it!
233.
▲
by
ed
4y ago
Unlike many llm’s this was trained using public training sets (and cited in their paper), to let anyone with the $$$ independently generate the weights
234.
▲
by
ed
4y ago
The procedure was: For 7 minutes... 1. Rate your happiness 2. Stare at a cross on the screen for 15s 3. GOTO 1 They found reported happiness decreased over time.
235.
▲
Microsoft brings its new AI-powered Bing to the Windows 11 taskbar
(theverge.com)
1 points
by
ed
4y ago
|
0 comments
236.
▲
by
ed
4y ago
How is this different from e.g. a python agent like https://github.com/hwchase17/langchain-hub/blob/master/promp... ?
237.
▲
by
ed
4y ago
Cool! Looks like an improved version of Natbot https://github.com/nat/natbot
238.
▲
by
ed
4y ago
> "You could parachute [Sam Altman] into an island full of cannibals and come back in 5 years and he'd be the king. If you're Sam Altman, you don't have to be profitable to convey to investors that you'll succeed
239.
▲
Genesis, a Crypto Lending Firm, Files for Bankruptcy
(nytimes.com)
6 points
by
ed
4y ago
|
0 comments
240.
▲
FTX’s Sam Bankman-Fried Said to Be Arrested in the Bahamas
(nytimes.com)
13 points
by
ed
4y ago
|
1 comments
More ›