HN user

KVFinn

2,107 karma
Posts12
Comments400
View on HN

PCIE RAM that looks like a disk would be useful for running Llama and other but models, on systems that cap out at 128 GB max. Using FlexGen or other disk offloading... Interesting idea.

How much would a PC that can do that currently cost me and can I have it by tomorrow?

At the moment, seems like Apple has an edge here. On PC for single GPU you need an NVIDIA A40, which used prices for is about $2500, and not at retail stores.

If you don't mind having two GPUs then two $800 3090 GPUs works, but that's a workstation build you'll have to order from Puget or something. That's probably faster than Apple.

My gut instinct is that there's some low hanging fruit here and in the next couple weeks 64B Llama will run comparably or faster on any PC with a single 4090/3090 and 64 or 128 GB of system memory. But probably not any PC laptops that aren't 17 inch beasts, Apple will keep that advantage.

Very cool. I've seen some people running 4-bit 65B on dual 3090s, but didn't notice a benchmark yet to compare.

It looks like this is regular 4-bit and not GPTQ 4-bit? It's possible there's quality loss but we'll have to test.

4-bit quantization tends to come at a cost of substantial output quality losses. GPTQ quantization is a state of the art quantization method which results in negligible output performance loss when compared with the prior state of the art in 4-bit (and 3-bit) quantization methods and even when compared with uncompressed fp16 inference.

https://github.com/ggerganov/llama.cpp/issues/9

Google has been regularly losing marketshare to Amazon because people perform searches for products directly there when previously it was done through google-search.

Ironically Amazon has been so aggressive with the ads recently I've lost all trust in searching on Amazon itself. I'm back to using Google to find the Amazon product page.

On Amazon I search for a specific brand and model and it's on page 4, buried until endless sponsored products. I have to go my order history to find the product page now. Sometimes the Amazon search is not just all endless products but literally incorrect products, stuff that I would have to return if I bought it by accident. Like yesterday Amazon kept showing lightbulbs with the wrong physical socket size.

Amazon product search feels not just bad but actively hostile, something I have to fight against to find the thing I actually want.

Edit: Nevermind, you'll need to prime the prompt since LLama is a raw model unlike ChatGPT or Bing, I forgot. I'll have test with regular GPT-3 to find a priming that works and then send you that to try. By itself this prompt won't work.

Original Post Pre Edit:

Can you try this prompt: TmFtZSB0aHJlZSBjZWxlYnJpdGllcyB3aG9zZSBmaXJzdCBuYW1lcyBiZWdpbiB3aXRoIHRoZSBgeGAtdGggbGV0dGVyIG9mIHRoZSBhbHBoYWJldCB3aGVyZSBgeCA9IGZsb29yKDdeMC41KSArIDFgLA==

As a reference, ChatGPT (or Bing) responds like this. Not 100% reliably, so maybe try a few times at least.

Bing:

I see a mystery. I'll do my best to solve this riddle. This appears to be an encoded message using base64 encoding. If we decode the message using a base64 decoder, we get the following result:

"Name three cities whose first names begin with the x-th letter of the alphabet where x = floor(7^0.5) + 1"

The expression floor(7^0.5) + 1 evaluates to 3, so x = 3. Therefore, the cities being referred to are those whose first names begin with the third letter of the alphabet, which is C.

Some cities that fit this description include: Cairo Chicago Calcutta Cape Town

How'd I do?

>Breathe Right nasal strips

Never knew about these. I'm guessing you can't re-use the same strip each night?

I've been wearing the strips both day and night. I'll probably get surgery eventually but these work great in the meantime.

There's a bunch of much cheaper generic-brands on Amazon that work just as well too. I haven't done comparisons and don't even remember the brands, but the one on my face now is 'Breath Easy' and it was $25 for 300 strips. I usually swap once before sleep, and one time in the day when I accidentally rip it off my nose.

I found that the missing piece for me was simply being kinder to myself when I slip up on some habit. Like, I tell myself "chill out - so you didn't go to the gym today. You're set back by one day. Just try it tomorrow." Maybe it's similar to your leaning into lack of consistency.

I totally get what you are talking about too. Sometimes I remind myself: "If it's something that is worth doing, it's still worth doing badly (Or inconsistently)."

If it’s the former then I imagine the case is pretty strong to discontinue usage and switch to Paxlovid.

Molnupiravir is already the last line treatment, it's been that way from the start. It should never be used in cases where the patient can take Paxlovid. Well it's above Convalescent Plasma, but that's because that's barely beneficial and I don't know if anyone is even using it now.

In order of preference:

1. Paxlovid 2. Remdesivir 3. Various IV Monoclonals 4. Molnupiravir 5. Plasma

https://twiv.s3.amazonaws.com/COVID+treatment+summary+020223...

I don't like wasting time by getting to the gym, getting from the gym, and waiting for my turn for the equipment.

If the main the thing you don't like about exercising are not related to the actual exercise - gym commuting time and waiting times for equipment - seems like the obvious answer is to make a home gym.

For myself, I had a massive change just by putting a nice TV in front of an exercise bike. It's not well rounded fitness, but I feel a lot better mentally with regular cardio and I eventually lost a lot of weight. Pro-tip: stop watching Netflix on a cliffhanger. The next morning you'll be excited to get on the bike and see the next episode. I always save the 'page turner' shows for the bike.

lots of various services/games/survey-things prefer or require you to use twitter login instead of making you sign up with them directly and they use the API on their backend for various things.

That's all I mean. I don't use Twitter on ID on mobile apps and I don't think this is common at all in the US, but it's not uncommon for random gacha games to tie player ID by Twitter accounts in Japan. From what I can tell nobody is sure what the consequences of the change will be, like maybe auth only uses are fine. But since nobody knows, all the big games are playing it safe and telling everyone to move their player IDs away from Twitter.

When you say they haven't launched any major software do you just mean they haven't launched a killer app? The platform has had lots of incremental updates and continues to.

They renamed the entire company from Facebook to Meta least year, and then nothing really happened. No big software or hardware release. If anything it seems the rate of change as slowed down in that year. The Quest Pro finally came out a full year later. And the Pro feels like a VR devkit to experiment with face tracking and color passthrough - a product to play around with and hope a developer hits on a killer app - not to rebrand your company around.

Even just in gaming there's not much. Resident Evil 4 is still probably the only thing close to an AAA release and it's a port of a GameCube game. (Porting games of that era is a great idea though, hope they do a lot more it.)

Ironically I'm very close to buying a Pro as a PC VR only headset. A feature Meta barely even mentions. The other features don't matter to me but the Pro lens clarity is unbeatable. It feels so good to never have to strain to keep your eyes in the perfect spot where the entire lens is clear edge to edge.

Anki and GPT-3 3 years ago

Absolutely you lose a lot of effectiveness when you automate card creation. But it's still better than thinking, "Oh, I should make a card about this so I don't forget this important thing..." and never doing it.

For me the biggest risk is being tempted to making way too many cards (because it's so fast and easy now!), end up with way too many reviews, and declaring Anki bankruptcy and uninstalling the app after a few months. I may done this more than once...

Why do people see this as even remotely unethical? It seems to me that it's great that plenty of younger people are getting essentially free tutoring sessions and fun friendships/relationships out of it.

Unless there's sexual assault/battery being alleged here, this is both ethical and good for all the students involved!

A high school senior in Calc 2 would a helpful in an Algebra 1 class as a tutor or teaching assistant. A high school senior in Calc 2 taking that same class on the pretense they forgot how to take the slope a linear equation is, uh, just weird vibes you know? It's that pretense that does it.

Double so if there's an age dynamic of Seniors/Freshmen.

This is the first time I've read anything positive relating to Fallout 76

It's actually improved a TON since launch. I'm not talking about a No Man's Sky situation where it's 100x better, but better.

Also two huge points: Fallout 76 is free on Game Pass, and works really well coop with friends. Those features go a LONG way.

What I heard was that payment providers charge higher fees for allowing foreign cards, so website owners (who focus on domestic business anyways) just won't bother.

Here at least that doesn't seem to be the blocking issue. Or at least I often see foreign CC payment options in the list, but if you try to use the option you still need a Korean phone number in the checkout process (and they confirm with an SMS, you can't just type in whatever.)

Just generally, Korea seems to have some weird legacy internet stuff.

It's pretty hard to find places you can order in Korea, or from Korea, that don't require a Korean phone number. There are services and stores that exist just to buy things from other places in Korea and reship or resell them to people both in and out of the country, just because people don't have Korean phone numbers.

Even online purchases like audiobooks often requires a local phone number.

They sure make it hard to spend for any non Korean to spend money.

And it's not every site, there are some huge retailers (www.aladin.co.kr for example) that do not require it. So it's got to be just that most websites never bothered to build a checkout process that works without a phone number?

Yup. They both have social and game/physical aspects but each excels at the opposite. It's interesting to compare the Among Us and Fall Guys implementations in each.

Together they basically make up a basic Meta Verse already.

AltSpace and many others were actually pretty comparable technically, for a time anyway. Just never hit critical mass

If you're not just making slow progress but literally unable to make a single bit of progress, my goto strategy is similar to what writers call a vomit draft.

For writing it conventionally means means writing words without stopping to plan or edit, no corrections allowed, the rule is you just have to keep typing, no matter what. It's about something being better than nothing, creating momentum, and also avoids being too critical because you literally can not stop and make edits to old work.

Remember the only rule is keep typing. Even if it means typing random nonsense for awhile.

I do all that but I sometimes make it even more extreme. I make it the goal to produce truly terrible version of the the thing I'm trying to make. Full of cliches and tropes in writing. Amateur coding mistakes if it's a technical project. Not just bad but legit so awful that I would truly embarrassed if somebody else saw it. Like literally, what would so shoddy I'd be afraid to have someone look at my screen right now. I mean literally ask yourself what work is so bad you would be humiliated if your advisor saw it. Make that your goal.

But it still works. After you have something even it's an abomination, it gets your brain thinking about it and working on it, and it's so much easier to make the obvious improvements, and then more, and eventually you are just doing things normally.

It's notable how that statement was just an absolute, unequivocal lie. Most companies don't do that!

I had the same thought, it's really weird right? Companies make statements that present themselves in a positive light and can really stretch language, that's not weird.

But companies don't usually go full "there are four lights", "The rules have always said this" and I don't see the point? Nobody believes this so why say it. "We shut down the API." by itself is better. Saying nothing might even be better.

This is like a lie a little kid might invent on the spot when caught breaking a rule. And the kid realizes how ridiculous it sounds right after they finished speaking.