111–120 of 389 posts
Earlier quoted context omitted.
There is just so much downward pressure on token price, from every direction. We would need a completely new understanding of economics to explain why the price shouldn’t go down. Or market collusion/regulatory manipulation.
The price has been going down for ages, its not clear what you are pointing at
All of these words spilled and no mention of the ISA.
I hadn't seen the token/Joules comparison with human speech before. Humans are still 22x more efficient, which is not that far considering the rate of progress in this area.
The 20W number includes EVERYTHING else the brain does. The chips/models are literally only producing tokens. Let's see an LLM drive a robot harness and have the robot produce speech, as well as move through 3D space, keep track of metabolic needs, etc. etc. etc. before we compare efficiencies. That is even assuming the tokens are of equal quality. This comparison is currently Apples and Oranges.
I don’t see how tokens can’t produce speech or track metabolic needs. You can talk to chatgpt can’t you? Or do you mean literally talking? Because that’s not a brain function, that’s the mouth, vocal chords, and lungs.
I think they talked about this being general purpose chip but I would think that Anthropic/OpenAI are at the scale now they could bake LLM weights into chips themselves. For example, GPT Sol baked into a custom chip run for $100M that runs 10x as fast and 10x as cheap should pay for itself as long as the chip is useful for long enough. While 2 years ago nothing was useful more than 1 year long, there are many older m…
Probably! But not viable yet; the chips would be about a year behind SOTA. Note the ~16 months that the article quotes as being insanely fast to get this chip to tape-out (read: start producing). We'll have to bootstrap our way there: AI is actively being used to get us closer to viable lead times for this. Unfortunately, there's some real physical constraints: IIRC, manufacturing a wafer takes on the order of a mont…
The bigger issue seems to be that these chips can’t hold that many weights at the moment.
(I’m curious if chips with large weights in them would be more tolerant or less to yield issues. If you flip a few bits in the weights, does it really matter at scale?)
Competition is good for all of us, we will get better and faster chips. Or at least Nvidia GPUs will become slightly cheaper for regular consumers again
When people talk about the commodification of inferencing, they imagine a future where everyone has access to frontier models and can run them at the same cost, and what will actually happen is closer to the commodification of _oil_, where only a few companies have the scale to produce it at a competitive price, and advances like this are _why_. Once models are more or less interchangeable, the price of LLMs will dro…
With competition we will actually have the fair split, whatever that is, and thus much lower prices.
At the moment, to have a big AI firm, or really AI firm at all, you need to be blessed by NVIDIA, in the form of receiving circular financing for your compute. They know that their prices aren't fair, or competitive.
Commoditization of inference is the end of that. The end of the mega-premium on inference hardware, and it's good not only for people who like running their LLMs, but it's the first step towards commoditization of training.
Well Sam Altman finally has built a moat against Chinese open weight AI. Well done. But what will this mean for Cerebras? I remember when Tesla was building its own inference chips, and after about 2 years and billions spent, the whole effort was scuttled b/c they simply could not keep up with the iteration and R&D cycles of dedicated chip companies. I suspect the same will be the case with OpenAI vs Cerebras + Nvidi…
Cerebras is targeting a distinctly different point on the cost/latency curve. They are betting that there will be some high value applications where latency and not just throughput is super important.
Even in nvidia land rubin + LPU does a similar thing.
It has its downsides of course - if your traffic swings prefill heavy to decode heavy, you can't suddenly use your lpu for prefill. With GPUs they're totally interchangeable. Tradeoffs.
Earlier quoted context omitted.
This may just be a classic case of Jevons paradox: https://en.wikipedia.org/wiki/Jevons_paradox In short, better hardware will drive down token cost in the near-term, but will drive up the demand for tokens as it gets cheap enough for other sectors to start to use it heavily. It comes from steam engines where economists originally thought that coal demand would plummet with more efficient engines, but it actually jus…
[flagged]
And you can say of course, it's so obvious, how could a dumdum not see that! But then there are lots of examples of things where increased efficiency results in less usage overall, because demand is inelastic, etc. Jevon's paradox doesn't apply to everything.
I don't think we know yet what is going to happen as software development gets much cheaper. If in ten years we can produce software 1000x more cost effectively, will we need fewer software engineers, the same, or more? Guess we'll see!