JaRail

The downclock would only be on the core using the instruction.

Alright, I'll admit you made me laugh :D

> Just don't buy it if you don't like it.

You say that about a security issue that could blow California's power grid in minutes.

Not having any limits on that invert setting is definitely a security/safety bug.

The password is a level of incompetence beyond what you'd call a bug. It's clearly following a design decision that was imposed by someone with zero security experience.

This is an amazing lack of security best practices. To me, this screams outsourced. Given how many people hate Tesla, they need to be taking this seriously. This truly blows me away. This is "people should be fired" levels of organizational incompetence. There's no way some of these issues haven't already been noticed and put in the issue tracker. They're just not taking it seriously. It reminds me of Boeing to be perfectly honest.

Gemini man is a better example of where this is heading.

https://www.youtube.com/watch?v=k5y4kxhZIBA

A DF has a pretty specific meaning if you are diving into the technical aspects. It's an approach that has a lot of specific limitations that put a practical ceiling on how good it can be. For example, it's a 2D system that won't capture lighting differences with good accuracy. The limitations become more apparent as resolution increases.

Newer systems use source material to generate 3D facial models. The face in the target video is also 'motion captured.' The target video's lighting is estimated and a resulting face image is rendered and composited into the target. That's a lot closer to the pipeline of movies such as Gemini man.

So while DF isn't a technique used in movies, machine learning is heavily involved. It's built into model generation, motion capture, animation, etc. Artists will start with captured/generated assets and hand-tweak them to get the desired result.

It'd be interesting if they went with additional cameras for this. For example, the rear-view mirror is a video feed already. It could stream from a camera you attach to the back of the trailer instead.

I don't have a lot of personal experience with trailers, a few thousand miles maybe. But I've never liked clip on side mirrors. They bounce around/vibrate. A camera system that's optically stabilized and patches into existing displays would feel a lot more natural for someone like me with minimal experience.

You're completely ignoring higher center of gravity and rollovers.

Why would you paint it? It's not painted in the first place.

It's not an easy feature to test. A lot of magazines have used really fake looking inflatable or cardboard people. Radar systems are going to see those really well. Machine vision solutions may correctly classify the objects as non-human. You don't want your emergency breaks to go off every time a plastic bag floats in-front of your car.

Then you have far more complex real-world situations like this one. The emergency breaking works correctly. https://www.youtube.com/watch?v=X3rmevP6XVY

I'm not sure what statistics are available for pedestrian collisions. Probably nothing short of deaths are reported reliably at the national level.

$30,000 cheaper you mean. The mid-tier dual-motor is $20,000 cheaper.

Not that it's a guarantee but at least tesla's emergency breaking detects pedestrians already. We're still a couple years from release. I'd like to think the long-term solution is to just stop hitting pedestrians with cars.

Maybe you buy a lot of subscriptions and currency? I bet they avoid annoying their most profitable users. While it might have been coincidental, I never got ads until I unsubbed from the vast majority of streamers I follow. Same for other things like gifted subs. I used to receive a ton of gifted subs until I started lowering my sub count. They are obsessed with performance metrics. Big spenders have a very different experience.

Now, whenever I get an ad, I immediately close twitch and find something on YT/Mixer to watch. I'm training their algorithms to leave me alone.

Generally small particles like micro-plastics are too small to be filtered by treatment facilities. My assumption here would be no. It could be that a chemical/heating process in a treatment facility could accelerate the breakdown. That's just pure speculation though since we don't know enough to judge.

> 3. Regarding the environmental friendliness of our coatings: Our product is based on silicone. It has been shown in the scientific literature that silicone can be decomposed naturally in soil within 4-7 days. It breaks down in silica, water, and carbon dioxide.

"Based on silicone" does not mean "degrades like silicone." It also doesn't mean your coating will behave like the silicone "in the scientific literature." If I toss a silicone spoon in the ocean, it won't be gone in a week. You can make plastic out of wood. That doesn't make it biodegradable. There are different kinds of silicone. Some degrade quickly, some don't. Some is toxic as it degrades. It can accumulate in fish like plastic, etc.

Without knowing how the coating wears off (in chunks when toilet is cleaned, etc), it's hard to compare. Also, if it's not actually silicone, we shouldn't be pretending it acts like it is.

Have any independent environmental assessment been done? What happens when you drink it? What happens when you feed it to fish?

I'm just guessing but I'd imagine they'd take debit too.

> How do you define just fine? It's taken us many years to migrate EMCA versions only to have multiple incompatible runtimes.

ECMAScript versions are forward-compatible. Any valid ECMAScript 3 code is also valid 5.1, 2015, 2016, etc.

I'm not sure what migration you're talking about. If you mean using new language features before your runtime targets support them, that's kinda on you. Even so, the ecosystem has tons of robust solutions for supporting legacy interpreters. Most notably, Babel does a wonderful job of transpiling to lower language-version targets.

The headline makes no sense when they follow it up by saying

> Battery-grade lithium is used primarily in EV batteries, and many automakers have high purity standards. Much of the industry’s capacity to produce high-quality, battery-grade lithium is locked up until 2024

They're just putting two unrelated things in a headline to imply they're related. It's clickbait/FUD. If they're selling all the battery-grade lithium, why don't they tell us about why demand is down for low-grade lithium?

> how would the hospital obtain records from an unconscious person?

In most developed countries, they pull his ID out of his wallet and type it into the nearest computer. National healthcare has many advantages.

I'll take that as a no.

> There’s no investigation of the physical or natural world going on

I just quoted them describing their observational method! Do you just not believe psychology is a science?

> unless they really think they’re modeling how humans are able to talk

I've lost you. They're not generating birdsong. What do you think WaveNet does exactly?

> Progress in AI is due to data and computational power advances.

I think you'd be surprised how much progress is also being made outside those two factors. It's sort of like saying graphics only improve with more RAM and faster compute. We know there's more to it than that.

In many cases, the cutting edge of a few years ago is easily bested by today's tutorial samples and 30 seconds of training. We're doing better with less data and orders of magnitude less compute.

The biggest missing piece is WaveNet, which is Google's proprietary voice-synthesizer. With only the models trained for this paper, the best you could build is a voice-recognizer. As far as I know, Google only allows people to do TTS with one of their provided voices.

I don't expect them to open it up until other companies/academics have achieved similar results. It's too much of a competitive advantage right now. Alexa, Siri, etc all sound like robots compared to WaveNet (google assistant).

Did you read section 3 of the paper where they evaluate their system?

> We primarily rely on crowdsourced Mean Opinion Score (MOS) evaluations based on subjective listening tests. All our MOS evaluations are aligned to the Absolute Category Rating scale [14], with rating scores from 1 to 5 in 0.5 point increments. We use this framework to evaluate synthesized speech along two dimensions: its naturalness and similarity to real speech from the target speaker.

They're testing if the generated speech sounds natural with a well-defined and reproducible experiment. That's science.

I'm not really sure where you're getting this. It doesn't pick a specific voice from a database to use.

From their introduction: "Our approach is to decouple speaker modeling from speech synthesis by independently training a speaker-discriminative embedding network that captures the space of speaker characteristics and training a high quality TTS model on a smaller dataset conditioned on the representation learned by the first network."

Section 2 of the paper explains how it works. Two minute papers also goes through it if you'd prefer a video. Link: https://youtu.be/0sR1rU3gLzQ

There are lots of systems that let multiple users sign in and switch without re-entering credentials. Game consoles are a good example.

It's not the worst idea. If they had performance problems in multiple regions, their engineers would be chasing peak hours around the clock. If they focus on NA first, they can at least get some sleep at the end of the day.

It's a lot more disturbing if other parts of the government can arbitrarily target people they don't like for harassment. There's a big difference between 10 random searches and 10 targeted searches. Journalists aren't that common. If there's a large cluster of journalists getting targeted in a close period of time, it's worth figuring out why that's happening.

It depends on the entry point. If I go through US customs in the Toronto airport, it's on Canadian soil. If someone there is having an issue with the process, they can withdraw their entry request and leave. That's perfectly fine and they shouldn't be punished for it. You don't have that option if you use a land crossing as you're physically inside the US.

That doesn't really help in an emergency. You could be dead from a drug interaction before they even know you're in a hospital.

Don't forget gaming. They're going after the Twitch business model there. Combining that with traditional YouTube and Stadia offers a very robust social gaming platform. They're basically going after digital entertainment all across the board, looking for opportunities to leverage cloud/ai to leapfrog encumbants.