jafitc

7 ms·

31.
by jafitc 3y ago
Link from thread https://dev.to/zvone187/gpt-4-vs-claude-2-context-recall-ana...
32.
by jafitc 3y ago
My experience matched this as well. GPT-4 Turbo is more watered down on the details with long context But also it’s a newer feature for OpenAI, so they might catch up with next version
33.
by jafitc 3y ago
There are services online that can help you out. Google is your friend.
34.
by jafitc 3y ago
desperate times, desperate measure...ment practices
35.
by jafitc 3y ago
the new model is live
36.
by jafitc 3y ago
Bard currently is GPT-3.5 class model. Of course it's faster than GPT-4. (YMMV on specific examples of course)
37.
by jafitc 3y ago
I have a GPT-4 subscription, but not for Claude because GPT-4 is a better overall model. Still used both extensively. Claude just works better for insight extraction from long context. To say that it's "barely" doing what it&
38.
by jafitc 3y ago
On a constructive note, these things will trickle down into the models. Bing for example already does "thinking" step that is hidden from the user. Also see this quote from Ethan Mollick on twitter: > I have a strong suspicion
39.
by jafitc 3y ago
Sounds like you have a lot of firsthand experience with their model. Also like you "barely" read the article.
40.
by jafitc 3y ago
…when facing non-real-world adversarial scenarios.
41.
by jafitc 3y ago
No, what it’s showing is that synthetic tests where Claude didn’t perform well can still work if prompted right. But at the end of the day the test was still synthetic! Placing out-of-context things in a 200k document, needle in a hayst
42.
by jafitc 3y ago
It’s not where , it’s how .
43.
The Dark Matter of Programmers(johndcook.com)
3 points by jafitc 3y ago|0 comments
44.
Yann LeCun and Geoffrey Hinton disagree whether LLMs can reason(i.imgur.com)
4 points by jafitc 3y ago|6 comments
45.
by jafitc 3y ago
thread: https://twitter.com/geoffreyhinton/status/172920036786147771... (posted the image as twitter is not readable in a logged out state, and is annoying to many)
46.
by jafitc 3y ago
OpenAI's Future and Viability - OpenAI has damaged their brand and lost trust, but may still become a hugely successful company if they build great products - OpenAI looks stronger now with a more professional board, but has fundamenta
47.
by jafitc 3y ago
Sam Altman's Actions - Sam complained that Helen Toner's research paper criticized OpenAI's safety approach and praised Anthropic's, seeing it as dangerous for OpenAI - He reprimanded her for the paper, saying as a board
48.
Elon Musk says X will strip ability to block accounts(reuters.com)
8 points by jafitc 3y ago|6 comments
49.
Ask HN: What startup/technology is on your 'to watch' list?
3 points by jafitc 3y ago|1 comments
50.
by jafitc 3y ago
because people use it for 10+ years and never experienced outage? it's there to give your perspective add me to happy customers
51.
by jafitc 4y ago
Google: emergent capabilities of large language models
52.
by jafitc 4y ago
I agree with most of the article but displaying a static 5-star image on your website when your real rating is 4.6 stars (should be displayed as 4.5 star image) is rightly called out as disingenuous. If you really want it to be efficient di
53.
by jafitc 4y ago
if you're worried, you just need to buy a non-alcohol non-chlorhexidine based one. Listerine offers it too. https://www.listerine.com/mouthwash/listerine-zero
54.
by jafitc 5y ago
now count the upvotes versus the "original" posts about this
55.
by jafitc 5y ago
“This guy fucks”
56.
by jafitc 5y ago
do you really need to load 14,000px worth of content with no interaction? What if I just navigate away to the first article? The Nytimes website with the exact same look and functionality could be well under 1MB easily.
57.
by jafitc 5y ago
Is this something that can run in browser Workers via wasm or other means? Most of CSS OM is already implemented in Javascript which can come in handy for this project: https://github.com/NV/CSSOM https://gi
58.
by jafitc 5y ago
In the second video he gives a very interesting reason: Given 440 is what we’re all used to on an everyday basis, the effects of the OP experiment might be explained by the “novelty” of how 432 sounds to us.
59.
by jafitc 5y ago
Is it easy? Yes and no. Adding together integers is easy. Running a loop is easy. There are tremendous differences between programming technologies. Because the word “language” can distract you here. What you say in French and German and En
60.
by jafitc 5y ago
Thank you for the link! Yes and if you click “microScript programming” in the menu you are taken to the language itself. Didn’t find a way to link to that part but it’s just one click away.

More ›