wizee

Ollama defaults to a context of 2048 regardless of model unless you override it with /set parameter num_ctx [your context length]. This is because long contexts make inference slower. In my experiments, QwQ tends to overthink and question itself a lot and generate massive chains of thought for even simple questions, so I'd recommend setting num_ctx to at least 32768.

In my experiments of a couple mechanical engineering problems, it did fairly well in final answers, correctly solving mechanical engineering problems that even DeepSeek r1 (full size) and GPT 4o did wrong in my tests. However, the chain of thought was absurdly long, convoluted, circular, and all over the place. This also made it very slow, maybe 30x slower than comparably sized non-thinking models.

I used a num_ctx of 32768, top_k of 30, temperature of 0.6, and top_p of 0.95. These parameters (other than context length) were recommended by the developers on Hugging Face.

There are a huge number of temporary residents in Canada, either as students, or on a post-graduation work permit, or as temporary foreign workers. By cutting back on these programs bringing in temporary residents, and also slashing the permanent resident intake, many of these temporary residents will need to leave. As such, the government is projecting a net decrease in population over the next two years.

Agreed 100%. Waterloo has grown substantially over the last 15 years, and is generally thriving. The condo boom and general increase in density has only increased walkability. Most of these condos have ground level retail too. The light rail is also a nice addition.

As RIM/Blackberry declined, a whole ecosystem of startups emerged started or staffed by ex-RIM folks. The universities have also grown substantially.

It seems to be a gross exaggeration to say most Python code from 5 years ago doesn't work on current Python versions. Python 3 was mainstream a decade ago, and almost all code written for Python 3.3 or 3.4 still works on Python 3.13. Maybe some libraries have had breaking changes, but at least for common libraries like Numpy, Scipy, and Matplotlib, most code from a decade ago still works fine.

What's subjective is the threshold of what is considered "dangerous". I do agree improving visibility would be a good thing, and incentivizing safer designs through taxation could be good. Insurance prices already do that to an extent. I do agree that SUVs and pickup trucks have become taller than necessary in recent years; pickups from the 1990s were appreciably lower and had better visibility without being any less useful for moving things around.

The CAFE exemptions for SUVs and trucks compared to cars don't make sense to me and encourage maximizing vehicle footprint, so changing those rules would make sense. What else do you consider "subsidies" for this size of pickup trucks? Not being subject to the "chicken tax"? (not directly relevant to me as I'm in Canada)

Many people have legitimate needs and uses for pickup trucks and SUVs, and the definition of "dangerously oversized" and unsafe is rather subjective. Should semi trucks be banned because they are "dangerously oversized"? What should people use to move large and heavy items? Should farmers be prevented from having the vehicles they need to do their job?

While I agree that large pickup trucks and SUVs pose an increased risk to pedestrians, particularly in urban areas, and should be discouraged as single person commuter vehicles, particularly in urban areas, there are a lot of other use cases. What should someone who needs to move bulky/heavy/rough things with regularity and can only afford to have one vehicle use?

Discouraging negative externalities through taxation makes sense, but setting taxes to be so punitive that they make it difficult for people to afford the vehicles they need to do what they need to do is also harmful.

As an example, I have a pickup truck and a regular car. Driving around town I use my regular car, but when I need to move large or heavy things I use my truck. It's much more convenient to be able to use my truck when I need it rather than having to rent one every time or hire a company to move things for me.

The pickup truck is large (RAM 1500), and some would argue it's dangerously oversized, but its size is needed when I need to do truck things with it. The truck being affordable means I can afford to have a regular sized vehicle for tasks that don't involve moving big/heavy/dirty things.

Mozilla is like Wikipedia, where the vast majority of the funds they receive go to causes unrelated to development or maintenance of their core product (web browser or encyclopedia).

The RK3588 does seem to comfortably beat the Xeon D-1700/D-2700 series at performance per watt and per dollar.

Compare the RK3588 to the Xeon D-1718T as an example:

RK3588: https://browser.geekbench.com/v6/cpu/5757475

Xeon D-1718T: https://browser.geekbench.com/v6/cpu/945

This Xeon is a 46W TDP part with a $274 USD unit price for the CPU alone. The RK3588 has a ~11W TDP and entire Orange Pi 5 boards can be bought for around $120 USB. The RK3588 can get around 62% of the multicore performance for around a quarter the power budget and vastly cheaper SBC prices compared to Xeon D series servers.

The BLE MAC address consists of a random periodically changing part and a deterministically generated part based off encrypting the random part with a device-specific secret key (the Identity Resolving Key). This allows paired devices to identify the BLE advertiser since they know the IRK, while random devices like signs would not know the IRK and thus would not be able to deanonymize the MAC address. This feature is called Resolvable Private Addresses (RPA).

>Sure, but a lot of cars don't have them, and that's just the speedometer, not the speed limit or navigation directions (which are usually what I'm glancing at my phone's map app to check)

Most automotive HUDs show speed limits and navigation directions alongside speed.

I routinely wash my iPhone 13 mini in water, putting it under a running tap and washing all over. I've been doing it for more than two years now, no issues. I've fully submerged it a few times too without issue.

BMW has had some reliability misses in the past (especially with past V8 engines), though their current B48/B58 engines are very reliable. Porsche has a very extensive vehicle testing process before their models come to market, and they have always been known for good reliability (though they have had some misses in the past, like M96 engine IMS bearings and bore scoring from the early-mid 2000s, and 991.1 GT3 engine failures).

My old W126 was an amazingly well built car. Very durable and reliable, extremely long lasting, very comfy, interesting to drive, and amazingly safe compared to other cars of that era. Mine was previously owned by a couple who just did bare-minimum maintenance and drove it 120k km over their 6 years of ownership. The car was 24 years old when they got it, and 30 years old when I bought it. I drove it many more years, and the only repairs I needed to do were fixing age related leaks, and replacing one relay.

The W126 was a fairly simple car though, compared to their modern offerings, and it was built under a philosophy of building the best car regardless of cost.

My W140 was a also a well made car, great quality in materials and mechanical fundamentals, but it was too complicated and quite a bit more problematic for me.

Modern Mercedes-Benz cars don't seem as distinguished or well built as their offerings from the 1980s. With that said, my C216 CL-class is a very well built car in terms of materials and general attention to detail; it's just too complicated with hurts its overall reliability (though the powertrain is decently solid).

The B58 and B48 are quite impeccably good engines, both in terms of reliability and also performance/response. I had an excellent experience with my B58-powered M240i. I've also never heard anyone needing to do anything in practice with timing chains on these engines, even at high mileage. It seems the whole car would be worn out before the timing chain wears out, provided you use decent quality oil and don't neglect oil changes.

The older N52 engine is also very solid and reliable, my high mileage 2006 330i (E90) has been rock solid reliable in its powertrain.

My mechanic told me the N20 engine is terrible though, he has encountered an absurd number of them failing. The N54 is solid enough mechanically, though they get HPFP issues and need occasional cleaning of soot/gunk from valves.

People have been repairing their own cars for a century, and the world functioned just fine with it. This locking down is a recent phenomenon. Your car is not going to mow down people just because someone other than the dealer serviced it.

Such keys sold in large numbers could be detected and blacklisted though.

There were a lot of people avoiding medical treatment or diagnosis out of fear of Covid at hospitals and medical clinics, and this led to extensive delays in elective surgeries and delayed diagnosis of medical conditions such as cancer. Thus, fear of Covid resulted in numerous excess deaths. There were other factors too, like increased traffic deaths due to increased reckless driving due to emptier roads due to pandemic lockdowns.

It's also disingenuous to say that all deaths of people who had Covid were due to Covid. Many of them who caught Covid while in hospital for serious conditions would have died regardless of Covid.

Capture One? That’s what I migrated to after using Apple Aperture for many years past its discontinuation. Perpetual licenses still exist for it.

True, all vehicles (ICE or EV) are fairly noisy at highway speeds outside, though a pickup truck with highway tires isn’t much noisier than say Corolla or a Nissan Leaf moving at the same speed. Off road tires, particularly mud terrain tires, can get noisy though, as those tires were optimized for grip in mud rather than noise or highway fuel efficiency.

Outside the vehicle, the engine and tires are pretty quiet too, not much different from any other car.

As for hummer level acceleration, I agree that’s unnecessary and would be dangerous to use in an urban area. 0-65 mph in 1.4 seconds sounds wrong though, it’s more like 4 seconds. Car and Driver measured 3.3 seconds for 0-60 excluding rollout time. Accelerating full blast in such a fast car in an urban area would be reckless on the driver’s part, but that doesn’t mean you can’t drive it safely and reasonably in urban areas.

The Hummer’s bad stopping distance is from a combination of heavy weight an off-road focused tires not optimized for dry pavement grip. With that said, its stopping distance is only around 15% worse than the average car.

As an example of stopping distances and cornering grip of a more typical pickup like my RAM 1500, on stock tires, it’s 187 ft from 70 mph, and cornering grip is 0.78 G. For comparison, a Toyota Corolla has a 174 ft stopping distance and 0.82 G grip on stock tires, so around 5% better but not drastically better than the RAM truck.

A RAM 1500 weighs around 4900-5500 lbs depending on spec (cab size, bed size, 4x4, engine choice etc.). That’s around the same weight as lots of EV sedans like the BMW i4, Mercedes-Benz EQE, Lucid Air, Porsche Taycan. A Mercedes-Benz EQS or Tesla Model X both weigh more than my crew cab V8 pickup.

I don’t know about Hummers in particular, but most modern trucks are very quiet at highway speeds, at least with highway tires. My RAM 1500 is quieter inside on the highway than my BMW 3 series or my Ford Escape or any Tesla I’ve driven. The V8 engine is also quieter at idle and at highway speeds than either of those aforementioned cars. My very topical truck handles pretty decently too (for its size at least).

Staffer means member of staff. For example, engineers are staffers.

Ah, yeah, there was a rule requiring apps to use a native GUI toolkit when the App Store first launched over a decade ago, but that rule is no longer there.

When was this rejection? They don’t have any rule like this currently, and the App Store is full of apps not using Cocoa/Carbon/SwiftUI.

What Apple policy bans Qt-based apps from the App Store?

Speaking of Fluoroquinolones and Cipro (Ciprofloxacin) specifically, I personally had bad experiences with it causing tendon damage and severe back pain, though fortunately the damage healed and the back pain cleared within a couple days after stopping my Cipro course early and switching to a different antibiotic.

You are not in a position to be giving your (ex?) wife and kids $3000 a month when you are currently unable to get a job paying even a low amount for subsistence existence of your own. Get whatever work you can so you can pay rent for some small basic room somewhere, and food for yourself, and petition the court to greatly lower your support obligations for now given your incapacity to earn. Only once you are back on your feet and able to support yourself can you consider supporting your family.

That comparison chart is not very accurate. The Core i7-1250U CPUMark score of 13453 is at 29W (TDP up), not 9W (TDP down), so the performance per watt ratio they're using is inaccurate. Also, if you look at the CPUMark score distributions, you see a bimodal distribution because it mixes the 10 W (MacBook air) and 15W (MacBook Pro) configurations.

The 10W (MacBook Air) variant of the M1 achieves a median CPUMark score of ~14500. That's a lot better than the Intel CPU's 13453 at 29W. Now, if you limit the Intel CPU to a lower TDP by underclocking, the performance per watt can improve substantially (since the efficiency plummets when going for peak clock speeds), but it's still considerably behind the M1.

Though it sounds like the normal automated account creation flow is also blocked from that ISP. That could be more problematic and discourage editor account creation.