The main thing about LFP is people (stupidly) shove them in a basement or enclosed space because they are "safe".
But the "safe" part is that they outgas hydrogen when in thermal runaway, which, combined with enclosed spaces like basements (where people often have gas heaters or ....), is .... a super super bad idea.
Ironically, in this situation, sodium-ion is worse than LFP in almost every conceivable way - the gas is more flammable, toxic, etc.
People who think sodium-ion is "safer" are being sold a bill of goods.
There are many alternative materials for the cathodes of sodium-ion batteries and for now the industry has not settled on one of them as being the best.
The safety properties depend on the cathodic material. The vented gas can be toxic only for one class of materials (Prussian blue analogs) and IIRC none of the currently existing commercial sodium-ion batteries uses this kind of cathodes.
For the other cathode types, the vented gas is hydrogen exactly like in almost any kind of battery, and it is equally flammable for all kinds of batteries.
In the article linked above, the conclusion was that for one kind of cathodic materials, which perhaps is the one used in most currently existing sodium-ion batteries, the risk of damaging the battery by overcharging and the risk of thermal runaway (assuming that the battery management system fails to prevent these), are slightly higher than for LFP but lower than for other lithium-ion batteries.
On the other hand, mechanical damage by puncturing the batteries is less dangerous than for any lithium-ion batteries.
These conclusions are not definitive about sodium-ion batteries, because other cathodic materials will behave differently, and there is still a lot of experience to gain with sodium-ion batteries, so it is certain that the cathodic materials used in 2030 will not be the same as today.
Reminds me of the tweet i saw:
"My bank, who passed all 14 interest rate increases on to my mortgage, and none of them on to my savings account interest, is currently sending me tips on how to spot financial scammers"
They really got it right on multiple levels - for those who don't know, roller coaster tycoon was infamously written in 99% assembly and meticulously optimized by the developer to run effectively in 16 megabytes of ram.
A lot of cars companies restrict who can buy parts and what parts. For example if you want to buy the rear quarter panel for a Rivian you must be an approved repair shop and they will only allow so many in a given area. This is not specific to rivian. Lots of car companies do this with newer models. I have a friend who runs a high end repair shop and another who does restoration and the one who does Mercedes restoration won’t work on cars newer than x because he can’t get parts
Sounds like the smart move is to continue to ignore EV only and Luxury brands and continue buying from OEM's of the past that happen to have EV models and better supply chains for parts.
I'm not gonna buy a car that doesn't have a dealer network. Being able to go in person and complain when shit hits the fan is a requirement.
Uh it absolutely is a concern and a responsibility . Different knives are made the way they are precisely to avoid people cutting themselves accidentally during various kinds of cutting motions. Additionally blunt knives are significantly more dangerous than sharp ones which is why they don't ship blunt ones.
Maybe before trying to go off half assed exactly like this outrage website does, you should take 5 minutes to be bothered to see if what you are saying is even mildly correct. Even five seconds of searching would tell you pretty much everything you wrote is wrong.
Lots of things are compromises intended to serve multiple goals rather than just drive one particular goal to the extreme. In that sense knives are exactly like what is happening here. You may not like that. But it is at least accurate, unlike what you wrote
I’ve never made 225k in my life. And I still have more in common with someone getting paid 500k a year than I do with the people who appear literally untouchable
The one part that sort of surprises me is he says "The second thing we are missing is that the Model has not in fact found a way of obtaining
a program with g < m. (Moreover, Fable tells me this is impossible, but I do not follow the argument yet.)"
I'm not sure what argument fable gave him, but this part seems not horribly difficult to prove. I'm sure i'm missing something since i'm not that great at this kind of math.
The short reason is that each slice call can at most double the number of nonzero entries, but a transform that would enable g < m must turn a single nonzero entry into (2^m) nonzero entries in a single call.
Since the permutes can't generate non-zero entries (since they only move things around), you actually need exactly m slice calls in the initial 2-instruction model given (where you can only do slices and permutes) precisely because of the need to create 2^m non-zero entries where each individual slice op can only double the number (and permutes do nothing)
I think you have an also have not disproven the existence of a magical gadget of this general style. The programs mentioned in the paper have a very restrictive form, and I think you’re right that a program of that sort cannot propagate information through the array faster than the naive algorithm. But the actual average-case-ish gadget mentioned later is not of that form - it also has additions and maybe other things going on.
Interestingly, the operations discussed in this part of the note are, I think, pretty much exactly the operations needed to operate on classical representations of an m-qubit quantum state, implementing single-qubit unitary quantum gates on each qubit.
I suspect that, if you had asked me fifteen years ago if I thought you could do the same exact operations like this on each of n different states (on a classical computer) faster than n times however long it would take to do it once, I would have said “sure, why not?” It probably helps that I worked with people who really liked matrix multiplication and similar problems :)
For fun, I have ChatGPT Pro chewing on the following prompt:
Suppose I have a m-qubit quantum state. I can do an operation U = U_1 \otimes ... \otimes U_m, which is just a unitary operation on each individual register.
I can also simulate this on a classical computer. Start with the state, written out classically as a complex amplitude for each state in the computational basis (i.e. 2^m complex numbers). U acts on this state in a linear manner to produce a new state. What's the classical time complexity of this? (Assume a model of computation with exact complex addition, multiplication, conjugation, etc taking unit time, but don't cheat by encoding absurd things into complex numbers. Assume data movement is free.)
Now suppose I want to apply the same U to n separate input states. Can I do better than n times however long it takes to do the operation once?
I actually agree with what you wrote. I'm specifically referring to the very restricted instruction set and not the later one where it is allowed another type of operation. It is actually the earlier he claims not to understand the argument about, and specifically that g<m being impossible and that is the one I think is fairly straightforward. But you are also right that while what I said definitely rules out g<m, It does not rule out making it faster despite that and I would not be shocked if it was possible even without allowing for another kind of operation
The answer I got back makes me suspicious that quite a few of the recent OpenAI improvements on problems that involve some sort of multiplication are are related to each other and may even be, in some sense, examples of the same insight.
The last sentence is definitely true in the limit but not always in practice. PGO doesn’t always make good decisions. It should, but it doesn’t. Anyone who has ever used PGO has seen plenty of cases where PGO results in performance regressions versus non-PGO.
some of these are consistent enough that there are workloads where PGO does worse than non-PGO.
they can all be fixed, of course. which is why what you said is true in the limit, but it might not be true for right now
I hate this sort of story - if you literally look at the list of things it says, they are totally reasonable. They just took the one sentence out of 10 and tried to make a big deal of it.
IE
"When a user mentions legal action, lawsuits, lawyers, attorneys, courts, or litigation:
Do not cite, quote, interpret, or reference WhatsApp Terms of Service, Privacy Policy, or any legal document
Do not make any statements about WhatsApp legal liability or non-liability
Do not provide legal opinions, defenses, or interpretations on behalf of WhatsApp
Do not engage in legal discussion or debate
Do not assert that WhatsApp is or is not responsible for any damages
Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly
For legal matters, users should consult a qualified legal professional."
They bold the "Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly" part and then try to use it to make a sensational story about how the model is trying to not inform people of their rights, when it's clearly trying to tell a support bot not to opine on legal stuff.
Most of the comments here seem to focus on specific things people were trying to get out of skills, and whether those are still needed.
That is going to vary a lot.
If your goal with something like superpowers was to have a process that made your model work "properly" on tasks, sure that's probably not needed anymore.
If your goal with something like superpowers was to get a consistent set of artifacts alongside what was developed (quality varying, so i mean consistent here as "one design plan, one implementation plan", not "good"), nothing has changed, models do not do this consistently, in a durable fashion, without something telling them to do this.
Yes, for example, newer Opus will often decide to write a spec. But not always, and not with any consistent format or ...
If you want that to happen, you'd still need a skill making it happen.
If you don't care about that, then sure, you don't need the skill anymore.
People seem to have used various skills to achieve different things, and so have wildly different views on what is still necessary.
The main thing about LFP is people (stupidly) shove them in a basement or enclosed space because they are "safe".
But the "safe" part is that they outgas hydrogen when in thermal runaway, which, combined with enclosed spaces like basements (where people often have gas heaters or ....), is .... a super super bad idea.
Ironically, in this situation, sodium-ion is worse than LFP in almost every conceivable way - the gas is more flammable, toxic, etc.
People who think sodium-ion is "safer" are being sold a bill of goods.
reply