fdeage

See also this very nice teardown of an iPhone charger by Ken Shirriff (2012): http://www.righto.com/2012/05/apple-iphone-charger-teardown-...

I was rereading this old (2014) post from Tonsky (tonsky.me/blog/reinventing-git-interface), complaining about the fact that no git CLI manages to make the git interface sucks less.

Well, I have to disagree: tig does it! It's in the terminal, is lightning fast (written in C, can handle million-commit repos), configurable, and supports a shortcut for any action you can think of.

I think I type the "tig" command about 100 times a day. To me tig is a hidden gem that every dev should know about. No more complex GUIs, gitk, or byzantine git commands. Just `alias tig='tig --all'` and you're good forever with git.

Thank you. The quality of the answer it gave you is on par with what I was expecting, except I am very surprised it didn't recognize Obama.

Also, do you know why this response differ so much from ( and is far less accurate than) this response? https://news.ycombinator.com/item?id=37674968

In the previous link, GPT4V seems to have understood everything there was to understand about the picture (suspiciously, btw. As someone else said it, the picture and its text are almost certainly in the training data).

Could someone with access tell me what GPT-4V has to say of this image?

http://karpathy.github.io/assets/obamafunny.jpg

Andrej Karpathy used it in 2012 as an example of an image he thought would be extremely hard for a model to interpret. I'm wondering how this holds 11 years later...

You are absolutely right, my mistake. I saw "2012" in another submission and didn't care to check myself. I can't edit the title myself, though, so hopefully dang will correct it.

Thanks, I'll have a look!

Around last March I was intrigued (https://news.ycombinator.com/item?id=35236635) that ChatGPT's URL was a bit convoluted ("http://chat.openai.com/chat"). OpenAI fixed it in April.

This got me thinking: what takes so long to deploy a minor feature like this one on a product with 100M+ users?

(To be clear: this is not to point out incompetence on OpenAI's side, of course. On the contrary, I want to understand what I am missing)

Yes, me too, it can be an excellent starting point (in addition of being an excellent forum generally).

"light" in the sense of: "I read it for fun and do not care too much if I don't understand most of it", or in the sense of "I understand most of it"?

Yes, this would definitely be a game changer for almost all companies. Considering how huge the market is, I guess it's pretty difficult to do, or it would be done already.

I certainly don't expect a nice drag-and-drop interface to put my Office files and then ask questions about it coming in 2023. Maybe 2024?

Yes, but what if they are very large documents that exceed the maximum context size, say, a 200-page PDF? In that case won't you be forced to do some form of fine-tuning, in order to avoid a very slow/computationally expensive on-the-fly retrieval?

Edit: spelling

I saw it, but it only mentions "applications" (whatever that means) and not bare documents. Does this mean companies might be able to upload, say, PDFs, and fine-tune the model on that?

Interesting, but I am a bit disappointed that this release doesn't include fine-tuning on an enterprise corpus of documents. This only looks like a slightly more convenient and privacy-friendly version of ChatGPT. Or am I missing something?

I just realized you posted this 8 days ago... Yes, I'm wondering too! After reading "Sparks of Artificial General Intelligence: Early experiments with GPT-4" (on ArXiv) my guess is that GPT-4 is not far from explaining the picture perfectly.

This is Andrej Karpathy (AI superstar and ex-director of AI and Autopilot at Tesla) discussing what he thought was an impossibly hard picture to interpret for machines. I'm wondering how this holds with GPT-4...

Yes, also this. Weird that noone noticed.

The same happened to me.

Actually, I begin to wonder if maybe Google might be switching to some "high-quality" mode when it detects that you are a "technical" user?

Seems like a bit of competition was all Google needed to be good again.

How about the ads, though? Did you see as many as usual?

Depends on the query I guess. My first results were:

- elixir-lang.org (the official Elixir website),

- elixirschool.com (high-quality Elixir content),

- til.hashrocket.com (also high-quality Elixir content IMO),

- hexdocs.pm (the official documentation of elixir modules/packages),

- graffino.com (relevant TIL page about the date sigil),

- a GitHub issue page about a very close sigil in Elixir.

My point was: do you think Google switches to some "high-quality" mode when it detects that you are a technical user?

It's been released yesterday :) https://phoenixframework.org/blog/phoenix-1.7-final-released

With Phoenix 1.7 being released, will there be a new edition of "Programming Phoenix"? The latest one covers Phoenix 1.4, which is already slightly outdated. I think having an up-to-date book would be of great value for Phoenix beginners :)

> He was utterly dominant; indeed it is hard to think of a competitor in all of history more dominant over his sport or game than Tinsley.

Very interesting comment. This sentence about dominance in a field made me think of Stu Ungar, who dominated Gin Rummy so completely that he had to switch to Poker (where he became a 3-time world champion) to meet interesting adversaries.

I couldn't find an exact reference for the following quote, but still: "Some day, I suppose it's possible for someone to be a better No Limit Hold'em player than me. I doubt it, but it could happen. But, I swear to you, I don't see how anyone could ever play gin better than me."

Not sure about this, but related: https://news.ycombinator.com/item?id=22098473

The thing is, there doesn't appear to be any way to know whether this is the case.

If you are interested in a code review of git architecture: https://fabiensanglard.net/git_code_review/index.php

I do see some FB engineers sometimes, but mostly working on open source projects there.

There is, but it's pre-pandemic data and about the general public, not the HN crowd: https://www.theverge.com/2020/3/2/21144680/verge-tech-survey...

Probably. But I am afraid that losing valuable contributions from FB people leads to groupthink and ideological bubbles. I realize that I myself have a very negative opinion of FB the company, and I think this is mostly due to the stuff I read on HN. Most of my friends don't see the problem and consider Facebook as yet another big company, ruthless and profit-seeking maybe, but not more than the other ones, and that's probably true.

EDIT: typo

I wasn't talking about his death specifically, although the fact that so many people wrote about how they felt personally affected definitely says something. Name one CEO who would trigger the same reaction.

Also, I also wasn't talking about the legacy of the company he founded, or about the fact that Tim Cook managed to create more shareholder value than he did.

I was only thinking about the influence he had, and keeps having, on tech products and companies today.

I never said he was a good person. A certain Adolf from Austria is also certainly an important historical figure. I'm just saying that whether we like him or not, Steve Jobs matters, and that a century from now it will probably be helpful to read stuff about him to understand our society, the same way we might read today about, say, Galileo to understand science in the 17th century. I don't think this qualifies as being a fanboy.

I see your point, but I think history doesn't care about merit. Even if he was only the best phone seller in history, selling phone has become so important in our times that this achievement would still make him a historical figure IMO.

Also, definitely not sure about Woz. He would probably have found a nice job, sure, but it's a far stretch to say he would have revolutionized as many industries as Apple did.