sonium

TLDR: There are two versions of the Parquet file format, but adoption of Version 2 is slow due to limited compatibility in major engines and tools. While Version 2 offers improvements (smaller file sizes, faster write/read times), these gains are modest, and ecosystem support remains fragmented. If full control over the data pipeline is possible, using Version 2 can be worthwhile; otherwise, compatibility concerns with third-party integrations may outweigh the benefits. Parquet remains dominant, and its utility far surpasses these challenges

Or you simply use the pytorch.powerplant_no_blow_up operator [1]

[1] https://www.youtube.com/watch?v=vXsT6lBf0X4

But I still need pip to install uv, right? Or download it using a one-liner alternatively.

Yes, but the criticism is that using mathematics is a really bad guide, as this approach keeps failing

I’d like the option to automatically choose the LEAST privacy conserving option, because

1. I don’t care

2. It should work better since it aligns with the goal of the site

The fact that OpenAI is releasing a fancy UI instead of an improved model says something. I'm afraid GPT-5 won't be there any time soon.

Literally every color grading example shows log footage as the "before". Of course this lacks contrast and vibrancy because it's not meant to be watched "as is". Please show me regular footage as a baseline so it's a fair comparison.

I just watched the demo with the Apollo 11 transcript. (sidenote: maybe Gemini is named after the space program?).

Wouldn't the transcript or at least a timeline of Apollo 11 be part of the training corpus? So even without the 400 pages in the context window just given the drawing I would assume a prompt like "In the context of Apoll 11, what moment does the drawing refer to?" would yield the same result.

You probably should not use UUIDs to start with in your database at least not as an ID. UUIDv7 aims solve some of the issues of UUIDv4 that are even less suitable in for databases. 99% of times using BigInt for an ID is better.

Aren't there already tons of apps answering that specific question? I think the strength of this approach is answering the non-obvious questions.

I did something similar (but not for documents) but I’m struggling with selecting the optimal number of clusters.

yes, you are right I meant to say "and a Nobel laureate to explain it"

I have a PhD in material science and worked with superconductors, here are two things that is often misunderstood:

1) it takes a undergrad to make and measure a superconductor but a Nobel laureate to explain it.

2) in theory if you understand the mechanism of something you can improve it. In material science however almost always all you need is a trial and error approach to end up with a good result.

Conclusion: the effect is probably true, the explanation not

edit: typo

There is also a JSON analog which if found quite handy sometimes to extract some values deeply nested in an API response. Its also natively supported by some databases that have JSON datatypes like mysql.

AFAIK it’s not actually about cookies but a website tracking you by any technical means. Could be your IP or user agent as well

I’d argue they are not more generating the value than a printing press is generating the value of money. IMHO luxury brands are generating their products value by means of marketing.

I'm not sure why this would be more reliable. But it would probably fit, but at the cost of additional complexity.

That's what I thought as well, but now I do have some long-running jobs that exceed GCF's 60min limit. So I'm stuck with docker on Compute Engine, where GCP treats you like a 2nd class citizen as the OP found out.

If you are in a tsunami advisory area;

* A tsunami with strong waves and currents is possible.

I found this blog a gold-mine of information. Is there any book recommendation that goes in the same direction?

It’s the “arrangement” of elements in the simplest terms, like changes in crystal structure.

The stoichiometry stays exactly the same, just shred the battery and feed the result back to the beginning of the e.g cobalt mining operation operation. Better however if you somehow manage to roughly separate it into lithium, cobalt and so on and use the result instead of the respective ore.

Thank you for answering everything I ever wondered about SiC. Are there any other semiconductors that would be even better than SiC in all four points?

The whole point of the die on silicone seems to be that this maximizes the interface bandwidth and minimize latency between the dies. If this true the next step would be to bring the multi die modules as close as possible in three dimensions to ultimately build a borg-cube like structure in zero-g with a power source at its core.

If renewable energy is as cheap as he claims, why do I pay more than 20ct/kWh for it in Germany with a ~50% renewables mix?

This is nothing compared to how the car industry measures computational power: kDMIPS, kilo-Dhrystone-Mega-instructions-per-second

I always feel a bit puzzled when one thing is generalized to a the behavior of a Country .

It probably come down to the decision of one guy at Bosch who ultimately made the decision to go ahead with this, and this sounds like it was the democratic decision of an entire population of a country - who mostly don’t care what Bosch is up to.

I guess they could have added the RPi WiFi chip for about USD 1.5 in BOM cost (assuming common industry prices based on distributor price/quantity matrix)

Step down converters are commonly used in car ECUs. Similar input and output range, power and everything goes on a single PCB. So there must be a different reason.

I am always wondering that there seems to be demand for a PoE hat. I have never heard of anyone using it. So who uses it?

Also how large is Raspberry Pi’s share in commercial applications?