nahnahno

My prediction is that this becomes permanent. There is no good reason to be much more expensive than opus. At $4/$20 they are roughly at parity.

Making 2/10 permanent would be a killer move and make a strong argument against open-weight. For the sake of the open weight ecosystem I hope they do not.

how does this compare to Arrow IPC / Feather v2?

The fact that GPT-4.1 was the judge does not convince of the validity of the bench.

Sad. This guy was such a force of nature and a uniquely wonderful, creative personality.

This is not true in my experience. Cranking out code is obviously the bottleneck, unless you have the luxury of working on a very narrow problem. The author describes a multi-modal project that does not afford this luxury.

Complete nonsense. Intel 18A, were yields good enough, is competitive with TSMC N2.

Based on their heavily biased view of the Gaza conflict, based on their Arabic affiliations and the Hamas-run Gaza government’s reporting.

LoRA and QLoRA are still fine tuning I thought? Just updating a subset of parameters. You are still training a base model that was pre-trained (and possibly fine tuned after).

79% accuracy. Useless

Yes, likewise.

You can’t possibly be naive enough to believe that Nvidia’s Titan class cards were designed exclusively for gamers.

Moving away from color and race blind hiring to discriminatory hiring was not woke, by your definition.

Yes there should be. We don’t want to pay literal 10x markup because the card is suddenly “enterprise”.

> Writing naive algorithms

Depends on the circumstance, and how difficult an appropriate algorithm is to write, but in my experience, if code performance is important, this tends to yield large, painful rewrites down the road.

2.5-3 micrograms per deciliter is nothing. Wouldn’t even come up as elevated by current childhood screening guidelines. I very much doubt 2-3 IQ point difference.

This is not what premature optimization is the root of all evil means. It’s a tautological indictment of doing unnecessary things. It’s not in support of making obviously naive algorithms. And if it were it wouldn’t be a statement worth focusing on.

As the point of the article is to see if Claude can write better code from further prompting so it is completely appropriate to “optimize” a single implementation.

I read it as “most people don’t connect put 2 and 2 together - the doctor could be the boys father”. It is poorly worded, but not strictly wrong I think?

Looks great

Correct.

An open source project that doesn’t receive in kind development cost reduction benefit from free contributors should not stay open source.

It makes sense to anyone who has tried to build open source software as a sustainable business.

It doesn’t make sense to people who contribute small hobby projects or work as researchers / are government funded, which means ultimately paid for by businesses and consumers thereof.

In reality open source is very difficult to build a business around, which means that software can’t exist long term. It’s not about not wanting to be open source, it’s about realising that you and your employees livelihoods are being abused by people who see open source and take it to mean they shouldn’t pay. Especially egregious when your competitors take your work and build a closed sourced business around it.

Then think of what it means if you have a person in their 40's that is really investing times in startups. Either they're stupid or they are committed in a way that 20 year olds likely aren't.

Have you read the paywalled content? Can you summarize?

This is explained in the article:

“ That does happen, of course, but research shows that it is not the full story. A study of more than 17,600 overweight and obese patients from 41 countries who took semaglutide found that participants lost about 10% of their body weight and had a 20% reduction in serious adverse coronary events, strokes, heart attacks and all-cause mortality. Crucially, these cardiovascular improvements long preceded any meaningful weight loss. ”

I don’t think they have a plan B. Architectures take half a decade of work. Porting from risc-v to arm is not a matter of a backup plan, it’s that of a very costly pivot.

Yeah the comment reads as originating from a person who has never tried to sell something. You need to ask to get attention

Agreed. The "purists" are dickwads.

This is so stupid.

If you want to not be fat, exercise and eat fewer calories. Full stop.

The fact that a certain group of individuals don’t have that self control is just evidence that education and public health have a place. Drugs won’t solve that.

These drugs are needed for people with metabolic disorders caused by years of food abuse or poor genetics. It’s not a population wide solution.

Batching helps, and if you can avoid syncing after every write that helps too. In practice I get great throughout, but then I’m OK with losing a small amount of data if my server dies during writes, as long as that doesn’t lead to corruption, which it doesn’t with MDB_NOSYNC and kernel guided fsyncing

the review guide had everyone on 24H2; there were some issues with one of the updates that messed up performance for lunar lake pre-release, but appears to have been fixed in time for release.

I’d expect lunar lake’s position to improve a bit in coming months as they tweak scheduling, but AMD should be good at this point.

Edit: around 16 mark, https://youtu.be/5OGogMfH5pU?si=ILhVwWFEJlcA3HLO. The laptops came with 24H2