Nokia devices never had capacitive touchscreens prior to the release of the iPhone; it took them two years to produce a response. In contrast, the LG Prada (the first phone-esque device to have a capacitive touchscreen) was announced four weeks before the iPhone.
There’s either a mistake or this is a bad metric, perhaps.
Currently the most powerful AMD iGPU is Radeon 780M (found 7840U/HS and 8700G CPUs). Judging by Notebookcheck‘s results, M2 Max GPU has up 2× the fps in Borderlands, 2.5× the fps in Witcher 3, and 3× the fps in Shadow of The Tomb Raider.
As for the benchmarks, the M2 Max GPU has 4–6× the fps in GFXBench, compared to Radeon 780M. And the RDNA3-based 780M has twice the raw compute performance, compared to Steam Deck’s RDNA2 GPU.
Unfortunately, GPUs in handhelds are always severely underpowered.
at least historically handhelds can't be too expensive if they want to archive wide success
M2 Max isn't available on the free market (as part) but if it where it likely alone would cost noticable more then what a full handheld can afford to cost
AFAIK Whisper still can't handle multi-language content. If the audio has two languages (different narrators, for example), Whisper transcribes both of them during the first minute or so, and then either entirely skips one of the languages, or translates the foreign language to English, for the rest of the audio.
So, the value proposition of a subtitle-generating wrapper for Whisper would be to have an option to split audio into ~1 minute segments, transcribe them separately, and to somehow accurately join them. And I don’t think this one does such a thing.
Not OP, but… I don’t consider myself a “power user”, but as it turns out, I use search for nearly everything: reviews, code snippets, spelling, cafes, stores, apps, news, products, files… But, perhaps, the most common type of search is when you read a text or hear something and think: “Hey, what’s that?”
And so I easily make 50–100 searches per day when I don’t even feel I’m actively searching for anything. And if I'm in a research mode, the count goes beyond that.
I noticed that only after I’ve subscribed to Kagi. After realising that I’ll burn throw the plan’s allowance very quickly, I tried doing most of my “dumb” searches via Google, and more important stuff — via Kagi. But that introduces a cognitive load, and kills the point of using a paid search engine.
Perhaps, I should switch to the $25 plan, but I can’t justify paying $250–$300 annually for a search engine, even though I have nothing against the idea of paying for one. Unfortunately, it’s too much for me. Nevertheless, I’m rooting for Kagi because it ticks all the boxes for me, and provides value. I hope that operational costs go down, and more people could use it.
IIUC the criticism in your post comes down to this:
1) Neither ChatGPT, nor Bing can access URLs when you ask them to.
2) However, similarly to Perplexity.ai and Phind.com, Bing infers a search query from your message, does a search, and then summarises the first 3–5 results. ChatGPT doesn't yet offer such a functionality.
3) Bing Chat has a much more restrictive system prompt, which results in hallucinations and lies happening less often.
4) The summary of gpt-3.5-turbo-based ChatGPT was less creative than then summary of the Bing Chat GPT-4 instance.
If I understood the points correctly, the comparison is… flawed, in my opinion.
> Neither ChatGPT, nor Bing can access URLs when you ask them to.
Bing used cached version and knew what's in the text when presented with an URL. In one of three modes Bing stated that it cannot access web but in two other modes did a good job and summarized the text in question. Again: it reliably informed the user when it was unable to do something.
ChatGPT states that it can access web, hallucinates and elaborately lies.
> ChatGPT doesn't yet offer such a functionality.
But it LIES THAT IT CAN DO SO. That's the problem I pointed out in the article. Also: Would it stated "I don't know and I cannot crawl the web", it'd be a perfectly fine response for me.
> Bing Chat has a much more restrictive system prompt, which results in hallucinations and lies happening less often
The prompt was simple and there was nothing to restrict. Bing did a good job, ChatGPT hallucinated and lied. Same simple prompt, no jailbreaking which I pointed out in the post.
> The summary of gpt-3.5-turbo-based ChatGPT was less creative than then summary of the Bing Chat GPT-4 instance.
Your point? As I stated, I had no preference whether the output should be editorialized or not. Both GPT-3- and GPT-4-based ChatGPT and GPT-4-based Bing got the same task. My post is about RELIABILITY of these solutions. ChatGPT failed miserably.
> If I understood the points correctly, the comparison is… flawed, in my opinion.
Yet in my PoV, you have not presented arguments to back this opinion. Among 9 compared tools[1], I have pointed out that OpenAI's ChatGPT and OpenAI's partial-owner's Bing Chat - supposedly using the same tech - rendered, respectively, unreliable and reliable answers. Have I lied or used any half-truths? What could I improve in my next article?
Funny how my comment was raided and downvoted to oblivion after steady flow of upvotes.
The rise of cheap and abundant clothing basically finished the production of domestic knitting machines.
Toyota stopped production around 1990, Brother’s last machine was produced in ~1997, and Pfaff closed their business in 2001.
Just like the Juki, Brother, and Toyota, Silver Reed was a Japanese company (known primarily for their typewriters). They moved their knitting machine business in the early 1990s to China, just around the time the SK840 model was released. No newer Silver Reed models were designed since then. Silver Reed (the Japanese one) itself is defunct since the Fukushima earthquake, I believe.
The only new model of knitting machine that can be slightly considered “domestic” is Kniterate (https://www.kniterate.com/). However, at this size and price (€16k) it’s more of a machine for fashion designers, dressmaking studios, and the like.
Personally, this makes me very sad. There was a time when hand-knitting offered endless possibilities and machine-knitting sorta like a fast, by very limited shortcut.
With the arrival of industrial machines like the Shima Seiki MACH2X not only it is possible to knit the whole garment without seams (as compared to already good circular hand-knitting), it’s possible to have some texture, too. All thanks to four needle beds instead of the traditional two. “Traditional” for industrial machines and some domestic ones like Pfaff Duomatic — most domestic machines typically had only one bed.
Unfortunately, with the death of domestic knitting machines, such technology would never be available for the regular people.
It does sound about as bad as I thought. This "death of domestic knitting machines" as you put it has turned it all into some weird steampunk, because these old models are being sold (I can walk into a store and buy a new Silver Reed where I live), so probably someone still makes them, but the technology has been locked in to these models from 80's. Unlike with the sewing machines, where they are still trying to come up with new models, by adding every sort of useless gimmick.
Wouldn't touch Kniterate with a long pole, because of the long term repair and maintenance outlook. They are basically single product startup, and no idea who their OEM is.
I’m sorry, I should’ve been more clear in my response. The Chinese-owned “Silver Reed” company still makes the SK280 and the SK840 machines that were designed by the now defunct Japanese company back in the 80-90s. These are the machines sold in stores. Some people believe they are of inferior quality when compared to identical models of [used] machines manufactured in Japan. And quite often (depending on your location) you could find non-used or slightly used examples for less money. However, old machines have to be re-oiled (since the oil has oxidised long ago), and have the “sponge bar” replaced (polyurethane foam is likely to be disintegrated, too). Personally, I have no opinion on “Chinese vs Japanese”, since I’ve once played with different models of Passap (also known as Pfaff).
But yes, it’s somewhat eerie to observe women (it’s mostly them who enjoy knitting) discussing what kind of model to buy, and having absolutely no clue that the knitting machine age is over.
As for the sewing machines, I think that the best option for any domestic user is to buy a vintage sewing machine. Models made in the 1960s and earlier typically don’t have nylon gears and don’t tend to go out of alignment if you happen to hit a sewing pin with a needle.
If you’re in the US, the models to look for are 4×× and 6×× series Singers, although there are tons of different options. The German-made Singer 401G in particular is considered the Holy Grail of vintage sewing machines because it’s a free-arm machine (only common in Europe since ~40–50s, not common in the US till the 1980s), has almost all the functions a domestic user may need, and has no plastic gears.
However, pre-1980s sewing machines have one huge disadvantage: they all had a [resistive] rheostat foot controller, which made maintaining low RRM practically impossible. The ones that came with a Triac in the late 70s and early 80s sometimes boasted the word “Electronic” somewhere on the case (this is the case with some Berninas, like the 830). If the machine doesn't come with a Triac foot controller, it’s best to adapt a modern one or do one yourself.
Other options are:
1) To replace the motor with a servo,
2) To add an RPM-control (PID ro ARDC) circuit (either with or without the encoder). One Back-EMF based example with Dremel-like tools is https://hackaday.io/project/166492-grinder-speed-control-wit.... But it doesn't apply to sewing machines well enough, since it’s difficult to determine very low RPMs needed in sewing.
3) Use pulleys, if you’re willing to sacrifice high sewing speed https://www.youtube.com/watch?v=qKDSaNmB0rs
I do believe that some modern brands like Bernina and Pfaff make some advances, however I feel like most of them are related to quilting/embroidery, and not sewing. For example, there are no modern domestic sewing machines (besides the Sailrite) that can comfortably sew thick leather. Modern machines also have no real option for mechanical adjustment of presser foot pressure (important when you deal with vastly different thicknesses of materials).
Luckily, there's an alternative: images.yandex.com
It effortlessly does both reverse image search AND image objects search. Results are usually so good, it almost seems like Google decided to cripple their image search on purpose.
If there's one party I'd be even less comfortable sharing my data with than Google, it would be a Russian company. Not because I think Russians are evil or anything, but rather because there's even less accountability and oversight (if any at all) with what they do with that data than we have in the US/Europe.
I worked for a startup that had a similar business model: some car dealership has a fleet of cars depreciating in value (last year’s models, loaners, trade-in cars) and they provide their inventory to the rental startup to partially cover the losses.
And it seemed like a great idea, at first. Customers had an option to either pay 2 to 3 times less for a normal boring rental car, or… to pick up something nice that had a V8 and leather heated seats in it. The other part of the equation was the customer service: in order to compete with Hertz, Avis and others, it had to be exceptional. And it was.
There was only one problem. Renting out used cars was only a temporary and partial solution for the dealerships. Eventually, all these cars had to be sold. But, as it turned out, if you’re a dealership and if you’re renting out a car for 2/3 of the month, you’re constantly losing opportunities to sell it. Unsold used cars depreciate in value for evenn longer periods of time, take up more and more space… and those rental earnings no longer make financial sense.
Super good point. I've definitely run into that objection while trying to sign on used car dealerships--i.e. that they can't sell the car if it's being rented, so putting it on our platform might mean forgoing a sale.
But that's fine, because it basically means that dealerships typically only list cars that they have more than one of. So for example, if they have three Audi A4s in inventory, they might list only two of them with us. When they sell the other one, they'll take an A4 off Carve and put it back into sales inventory. In other cases, dealers will give us cars during periods of low sales and then putting them back into inventory when sales pick back up, thus minimizing the risk of a foregone sale.
But wouldn't a car dealer try to have a variety of cars models instead of multiple cars of the same model? Wouldn't it be better for them to be able to show a potential customer a variety of cars rather than just one model? For example, if they could only have 3 cars, wouldn't it make more sense for them to have an A4, a BMW 3, and an Acura TLX (for example) rather than 3 A4s?
Or maybe used car dealers have significantly more stock than I'm giving them credit for?
It seems to me that you could also rent to people who would wanna try the car before they buy it. Heck, put that in the app. "Get 10,000 airline mile on your credit reports right now of you buy this rental!"
Just to put this into perspective: when these dealers bring their vehicles to auction if they want to sell them, they'll literally bring them to the auction the morning of the sale, and then leave immediately when its done. It's a really big deal to have it off of the lot. In fact lots of the innovative work being done in the remarketing space is to decrease the amount of time these vehicles have to spend off the lot (mobile check-in, mobile CRs, etc.)
I've worked pretty deeply in this industry if you guys are interested in chatting, drop me a line. Email in profile.
For traditional dealerships, the cars not being on the lot does seem like a potential problem. But I'm seeing plenty of cars on the road with Carvana license plate frames these days, so there seems to be some willingness to buy a used car sight unseen over the internet (at least for relatively new cars in good shape).
If Carve could sign up Carvana as a provider, that could be a major source of cars, and Carvana's customers don't expect to be able to bring their car home today. Right now it looks like Carvana has a nationwide inventory of ~16000 cars, although I don't know where those cars live, relative to human population centers.
On the other hand, Carvana isn't Uncle Willy's Used Car Lot ... if this is such a good idea they've probably got the talent to implement it in-house.
Do you really need to have the car physically in order to sell it? Can't you just have a comparable display model that is used for test-drives? When sale progresses, then you can show them the actual vehicle and sell it to them for a discount.
If the dealership only sells new and last year’s models, it’s probably less of a problem, though there might be a situation, when there’s only one or two models with a specific level of trim (that the potential buyer is interested in) left in stock and they’re all rented out. Assuming that all of these cars have similar levels of wear, which is not always the case. And even then many potential buyers prefer to have a look a exactly the car they’re buying.
And if it’s a used car dealership or a dealership that accepts trade-ins, there’s going to be a handful of models in single quantity.
Is this for new or used cars? If I was buying from a dealer that was trying to sell me a new car with over 500 miles or so I would consider it a used car and expect used car pricing. Isn't the new car depreciation the biggest downside to buying a new car. If they are renting new cars don't they essentially take a huge 25% hit on the sale price of the car?
Also If I found out a dealer was doing this without disclosing it at the sale time my next stop would be the state attorney general office.
