slagfart

Sage thoughts, trash88

GLM5.2 isn't even on this benchmark

Sorry if I missed this, but is there an example instance we can play with?

The Google Play app seems to be missing from the store? I get 'app not found'. I'm in Croatia if that makes a difference

Let's say I have about 100k novel-length texts I want to import into a model like this (I want to build a book recommendation system).

How should I approach it? I think this product can't deal with that much text yet?

As a couterpoint, I've never been happier than at my current role where I'm expected to put in 40 hours per week, and not a minute more. Sure I do timesheets, but they cap at 40 hours.

Prior jobs always had the implicit expectation that I could be working a little harder, or clocking off early on a Friday was evil. Now, 40 hours a week can be done any time, and that flexibility is hugely liberating. I honestly think they get more out of me as well.

What would the side effects of lower gun ownership be?

Perhaps not strictly on-topic, but is there any equivalent FS/program in Windows that will allow users to have read-only access to files that are deduplicated in some way?

My use case is the MAME console archives, which are now full of copies of games from different localisations with 99% identical content. 7Z will compress them together and deduplicate, but breaks once the archive exceeds a few gigs.

These archives are already compressed (CHD format, which is 7Z + FLAC for ISOs), but it's deduplication that needs to happen on top of these already compressed files that I'm struggling with.

Sorry for the off-topic ask!

I’d like to understand what tooling people use to govern process flows in these architectures. The concept is strong and totally makes sense, but with so many services running, how do we automatically monitor the metrics of each one? Not just in uptime, but in queue length, output quality, server cost, time taken, etc?

I used to run a TF2 server that ran some stupid hack I found that played shitty music nonstop (GBS.fm) through a widget that loaded on the welcome screen. Lots of people hated it, but everyone kept coming back for how fun the vibe was.

So fun. I miss it too.

In a small enough community, the Boy Who Cried Wolf remains a pertinent tale.

If you had the reputation as a decent and polite poster, the sudden use of a vile cuss, judiciously applied, yielded far greater aggregate results than someone who used them all the time. At the very least, you would save it for someone who really did deserve it.

Perhaps the sanding off of the web’s edges is simply a product of its size. Maybe we should go back to Usenet.

pls help me get online, nerd

Seriouspost: Agree with all that’s been said. Corporations have turned the web into an unpalatable wasteland. The prospect of someone finding out my name and job has taken all the joy out of calling someone a fuckhead online. I want to find it again.

I will start searching, but the good communities I’ve found have been invite only.

Credit is honestly cheap enough that, if you really wanted to invest money that is a liability from the second you earn it, you could.

All other things being equal, if the government earns the money upfront instead of having to wait, tax rates can decrease slightly for the same amount of income. I live in a PAYG country, and given higher rates of tax-default and tax-stress in non-PAYG countries, I prefer PAYG.

My understanding is that IOTA has achieved scale by sacrificing the durability to DDOS/network spam attacks. See: https://www.reddit.com/r/Iota/comments/7egqk7/is_iota_really...

There's no such thing as a free lunch in distributed system design! If IOTA had solved for scalability without sacrificing other attributes, the technology would probably be integrated into other coins very quickly.

Just to highlight - this is in no way related to the existing Apache Phoenix project, or to the ten other tech projects also called Phoenix. (https://en.wikipedia.org/wiki/Phoenix#Computing) This is a new one.

How? Even with the password for this user, you could still only gain access to the read-only schema.

Something I should have spelled out - the read-only schema has only the data that the charts need (heavily aggregated views). We basically build with the assumption that the schema will be compromised, but only that one schema.

This happily runs off a view that's got an aggregated subset of your data - so instead of sending customer IDs, consider sending a count of customer IDs against each metric instead.

I think, with Postgres, if you have:

1. A dedicated read-only schema

2. A dedicated user, with only CONNECT to the read-only schema

3. A unique password

4. A dedicated read-only replica DB

you should be safe against pretty much everything.

I'd actually like to be corrected if I'm wrong - this is how I've built numerous externally-facing services.

6 hours was just a ballpark to compensate for weak sunlight early/late in the day, bad weather, some panels getting damaged, loss due to dust, etc.

Not sure what the real rate of generation would be, and I'm happy if someone with actual solar experience can clarify my estimate.

I got my pricing from AliBaba - commercial rates seem to be $0.3/Watt to $0.4/Watt - so roughly $100USD for a 250W panel

Also, on snowy days, even the 'cold' contents are heated (from below freezing to just above)

Sorry - I must be missing something.

300 MW = 300,000 kW

6 hours of generating capacity per day, 365 days in the year, gives 2190 hours of generation per year. 2,190 hours * 300,000kW * 0.02USD gives an annual cost to Saudi Arabia of $13.14M a year, for 300MW of installed capacity.

A modern panel is around 250W, so they are installing and running 1.2M panels, for ~$11/year per panel, including grid connection and transformation costs? Insane!

Gobblin (https://gobblin.apache.org/), which looks like it does something similar, can be packaged up as a single .jar, and run on AWS Lambda.

Anyone had working experience with either?

Don't forget that the performance improvement of hash over b-tree gets better for wide keys - the wider the key, the better the improvement, as the key width is irrelevant to the hash index.

Based on the width of the key, the row count and the amount of memory available, there might be instances where a disk-hit gets replaced with an in-memory cache, which is awesome for spinning-disk based systems.

There's significant value, when developing a product, in knowing which other products your customers are likely to have, so that you can prioritise your different integrations.

Similar data could probably be compiled from Google Trends, but this clean and authoritative view is something that can be trusted, and I think the result makes the open-source ecosystem stronger.

This is why you won't make it as a politician, sadly.

You need initial popular buy-in to be elected, which is why populist policies will always do well.

"Unparalleled productivity" sounds like hyperbole, especially considering the posts below, showing how exactly the same thing can be done trivially in Postgres.

Do you have an example?

Does anyone know the % interest rate on these? It seems to be blanked out.

How can financial markets function efficiently when such basic information is suppressed?

You don't need a tool for this. If you store base data in one schema, and functions/views in another, you can update this second schema willy nilly without worrying about damaging your data.

Just start the second file with DROP SCHEMA IF EXISTS schemaname CASCADE; CREATE SCHEMA schemaname;. Then, make sure any functions/views/permissions built on top of this schema are also contained within this file.