I just flamed out exactly as you describe. It's my second time also. If anyone has any tips for some kind of counseling or coaching or anything else, would be much appreciated.
Get a few responsibilities in the outside world, or someone who'll make demands of you. Get a dog.
Half of group therapy is just being in a group. Groups that require you to be rested and fit and provide a local social network orthogonal to everything else in your life include intramural sports, hiking clubs, radio repeating, crafts and markets... and they're all comfortable with nervous individuals showing up.
It's worth being very honest with yourself about what you're feeling. If going into the office fills a void, you should realize that you are now handing a load bearing portion of your life and psyche to your employer. If you are laid off, you will be devastated. Doing the work to get your personal life to a place where you want to work from home because you enjoy your own company, friends, family, pets, environment... is very much well worth it in the long run.
That’s not how technology works. We went from first flight in 1903 to a man on the moon in 1969. Then we basically did almost nothing in aerospace for the next 50 years. We never got past the supersonic barrier for feasible commercial air travel. Apart from somewhat worse fuel efficiency, a 747 from 1969 can achieve basically the same trips in the same amount of time as a 787 or A350.
What looks like exponential curves turn into S-shaped curves everywhere. In the last century, coal power plants have gotten less than twice as efficient. A professor at law school was giving a lecture on technological singularity back in 2011. And I pointed out that Moore’s law was already dead. And in fact in the decade since then improvements came far slower than they did from 1990-2000.
It allows people with half the available money to do what others could do before. That's more than twice as many people.
It allows you to move freight really fast at half the cost. That's a lot of things that can move around the world 100X faster than they could on the next-fastest available mode: ships and trains.
It allows you to move freight really fast at half the cost
That's leaving out fuel prices. The planes may use half the fuel, but aviation fuel has increased 7x in price since 1990. I couldn't find prices for 1969, but if you take the price of crude since then you're looking at a CPI-adjusted increase close to 100x.
Great point. Even more value to halving the fuel cost, then, but fewer products able to take advantage of it.
Of course, inflation should be canceled out before evaluating the overall benefit. If you can self the goods for 7X then the fuel savings is back to 2X.
When things are spread out over time people are able to adjust to the point that society, as a whole, might not even notice there was a problem, even if for some segment of that society it might feel like the end of the world - or at least an incomparable inflection point. The only real problem you have is when things come on hard and fast, like a tsunami.
It can move to any place within the search space but it can't move outside of it and it can't move in between the 'pixels'. Human thought can, as human thought has created the search space.
I'm not sure if that's what you were going for, but I read it as if it were written by The Board in the game Control, and found myself with the appropriate level of existential dread.
Because there's been nothing to discuss since their announcement. Their API access immediately closed due to overwhelming demand and they didn't fab newer models than Llama3 yet.
Probably they will make bank selling to HFT for a while.
Funnily enough, pasting your comment straight into Jimmy leads to a... Funnily suboptimal answer that does not answer the question.
As someone else already contributed, this is driven by a Canadian startup taalas that basically makes chips that are llms, so everything is very fast but also, baked into the chip. Once this kind of stuff is a commodity in like 10 years, our world will be very, very different.
Taalas HC1 AI uses Llama 3.1 8B, but takes up a massive 53B transistors and 815mm2 on TSMC N6 (nearly at the reticle limit of 858mm2). N2 is a little less than 3x as dense (110MTr/mm2 vs 313MTr/mm2).
This chip would still be 272mm2 on N2 which is an eye-watering $30k/wafer and bigger than a 9950x or Nvidia 5070.
This just isn't feasible. Some of the latest-gen LLMs seem to have 5-10T parameters or about 1000x more. I don't know that taping out just one chip makes economic sense let alone the 300-1000 chips required for a cutting-edge model. Things like continuing education so your model knows about the latest NPM packages or world news is super important, but seems like it would require new chips.
There are a TON of uses for an 8B parameter models on the edge, but this is WAY too big to put on the edge of anything. Something like a 10mm2 100m parameter voice model might be feasible on the edge, but only for expensive devices, but most of those are TSMC 28nm (up to 29MTr/mm2) or GF FDX22 (up to 40MTR/mm2) which would increase the AI chip to the point where it would absolutely dominate the BOM.
> Things like continuing education so your model knows about the latest NPM packages or world news is super important, but seems like it would require new chips.
They probably have a few ideas around that. Me, personally, I'd have one main expensive chip (replaced every 10 years, or whatever), with a secondary cheap chip in front of it that gets replaced every year or so.
The secondary chip could act the way RAG does, or perhaps both chips together can act as LoRA.
Either way, 99.999% of the knowledge is static, you just need to fine-tune the weights with that remaining 0.001% knowledge, which can be done using RAG or LoRA on a much smaller (thus cheaper) disposable chip.
The better solution would be making part of the chip cluster use something like FPGA which can be reprogrammed.
Text to speech or diagnostics equipment where the core model is relatively small and never changes seems like the ideal application. You might be able to fit something in the 25-30B range in 2nm to 14A, but it would need a way to update.
Large models are simply out of the question in my opinion. If you need 400+ different chip designs, it’ll be billions of dollars to tape out before you even make the first chip.
> The better solution would be making part of the chip cluster use something like FPGA which can be reprogrammed.
I'm not sure I follow (It's late, I am tired and I haven't had my dinner yet. That's my stupid trifecta!)
The original chip has the weights, so it's literally just a bunch of on-die (read-only) memory cells. The FPGA, while you could use it for the memory cells, would be way too expensive to use as pure memory. Typically one would hook up (read-only) storage to it, so you still need that read-only chip anyway.
The FPGA is just the compute bits, but this chip has on-die weights, not just compute.
I was proposing that the they have the base weights on a primary (permanent) chip, and have a secondary (replaceable) smaller chip with weights for a specific use-case, or for fine-tuning with new knowledge/updates to the model.
The matrices can be multiplied LoRA style, applying the matrix in the secondary chip to the primary chip, resulting in up-to-date weights through which the prompt is pushed.
I'm wondering about something different. FPGA seems ideal for an AI chip because you can simply flash the latest model. The downsides are low density and low clockspeed. It seems that you can only fit 100-300M parameters in even very large FPGA, but that seems like it would be enough for most finetuning.
I'm thinking of a situation where you do the initial model calculations in hardware on the Taalas chip then hand that off to the FPGA to do the LoRA subset of calculations in hardware that can be continuously re-tuned to keep the model up-to-date. This would probably reduce throughput (or at least increase latency), but would save tons of money by allowing you to use the chips longer.
Yeah, they're clearly just starting out and just shipped their very first proof of concept. But to me, their plans seem generally reasonable https://taalas.com/the-path-to-ubiquitous-ai/, and like I wrote, if this kind of thing succeeds and could become some kind of cheaply producible commodity component, I think there's huge value in that. Alas, maybe not as a frontier model replacement, but say 10 years from now you can drop a cheap raspberry pi like device in your Lan and have a fast local engine for things like text sentiment analysis, text summarisation, voice recognition, basic vision and things like that, that would be pretty exciting to me (but maybe as you outlined, impossible in practice)
There is a reasonable kernel of an idea here, but only if you dial expectations WAY back. The 10 years speculation is just wrong though. Even in 10 years, their 8B param model isn't going to be in consumer devices.
6nm is just 7nm++ and the process will be a decade old in a few months. In the decade since, we've only had a slightly less than 3x increase in transistor density and that's including EUV, BSPD, and GAAFET (which means progress is likely going to slow down even more).
Even if we hit another 3x increase, their 815mm2 design will still be a bit over 90mm2. For comparison, the entire M5 Pro/Max CPU die is just 61.7nm.
If our current progress somehow holds (not likely), even 20 years from now the 8B model would be 30mm2. You need 30 years of dead consistent progress to get it down to an includable 10mm2.
As you can see, this doesn't make sense to invest in. As to the stuff like voice recognition or basic vision, these can often fit within 100m parameter models which would be around 10mm2 on their current 6nm design. That's doable today in custom edge computing devices.
The other possible use is cheap fallback models for AI companies. Moving to N2 and shrinking chips to 600mm2 to improve yields a bit would give about 50B parameters with 3 chips plus another FPGA-ish programmable chip for continuing training and interconnects for everything. You'd need hundreds of thousands of chips produced for that exact AI model just to get costs below $100,000 per board.
That seems like a lot of money for the AI model you are essentially giving away, but maybe it still beats the power and price of GPU server racks.
The government isn't going to be making chip fabs go any faster which is the biggest limitation here.
The second big issue is that it takes months to fab chips meaning your hardware AI is months to maybe a year or more behind the times when it lands.
I do think it makes sense for something like a medical scanner where the model simply doesn't need constant updates, but that doesn't need government involvement to ship.
Sometimes I visualize a setup like this [0], based on 2D art by Simon Stålenhag. Someone has their home robot sitting on a desk connected to their old PC with thick cabling, dumping endless lines of each subsystem's <think> logs to diagnosis why it did something weird earlier in the day. Systems pushing 750+ tokens per second per subsystem might even be considered on the slow side for realtime tasks by then.
In the whole Christian tradition, God/Jesus generally does not go for organ or limb regeneration. Two counter examples are a healed ear in Luke (but this may well have been resumption of hearing? details are a little light), and then a single Spanish example in the 1600s.
For His own mysterious reasons, He simply doesn’t go in for that stuff, however much intercessionary prayer ends up in His inbox.
not an amputee, but he did heal a man with a withered or malformed hand in Matthew 12. he also healed plenty of cripples, so I don’t think it’s a stretch to think that some or many of them were missing limbs.
I think to claim that 2000 years ago there was one person who performed miracles and/or healed people that nobody else could, with no actual evidence it was done and nobody else has been able to do it since, you need a better response to someone questioning it than “oh were you there? prove it didn’t happen.”
even the Jews of the time that did not believe or follow Jesus wrote that he performed signs. They claimed it was sorcery.
in Folio 43a of Tractate Sanhedrin of the Babylonian Talmud as follows:
‘It is taught: On the eve of Passover they hung Yeshu and the crier went forth for forty days beforehand declaring that "[Yeshu] is going to be stoned for practicing witchcraft, for enticing and leading Israel astray.”’
The relevant portions of the Bible record that the Jews of the Sanhedrin acknowledged the signs and miracles but said it was by the power of Satan that he did these things.
They also have countless records proving hundreds of thousands of women were witches, with ironclad proofs such as “if we stab them they scream so they must be witches, off to the pire” or “throw them bound in the water, if they float it means they’re evil because water (baptism) rejects them so they’re witches -> burn.”
He's not unique though. Quite a few people are on record for performing miracles in front of large audiences. Seemed like it was all the rage around this time period.
Vespasian apparently healed blind people in Alexandria. Apollonius of Tyana had a very colourful life performing all kinds of magic. Honi the Circle Maker was bringing the rain over in Judea.
Must've been something in the water other than lead!
This is ironic. They didn’t say they believe. You offered your belief that you know something that happened long ago (extraordinary claim), and they are naturally curious how you could know that. If you’re a time traveler or whatever we’d be quite interested to hear more.
