Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

100x defect tolerance: How we solved the yield problem

cerebras.ai

Post reply on HN
Print view View original

161–170 of 186 posts

Re: 100x defect tolerance: How we solved the yield problem

#161

Post by Shorel »

immibis wrote: post #158
Shorel wrote: post #148

Earlier quoted context omitted.

By analogy with human brains: Because our own brains are far more than the Broca's areas in them. Evolution selects for efficiency. If token prediction could work for everything, our brains would also do nothing else but token prediction. Even the brains of fishes and insects would work like that. The human brain has dedicated clusters of neurons for several different cognitive abilities, including face recognition,…

What do our brains do that isn't token prediction? They receive information about photons and air vibrations and control muscles, okay. If a human brain was hooked up the way ChatGPT was, only to text input and output, would that make it not intelligent?

> What do our brains do that isn't token prediction?

I am planning a masters and phd on that question, so give me a few years to answer.

Re: 100x defect tolerance: How we solved the yield problem

#162

Post by Shorel »

myrmidon wrote: post #156
Shorel wrote: post #148

Earlier quoted context omitted.

By analogy with human brains: Because our own brains are far more than the Broca's areas in them. Evolution selects for efficiency. If token prediction could work for everything, our brains would also do nothing else but token prediction. Even the brains of fishes and insects would work like that. The human brain has dedicated clusters of neurons for several different cognitive abilities, including face recognition,…

> Evolution selects for efficiency. I think this is a poor argument here. From an evolutionary point of view, our brains are optimized to: - Provide fine-motor control to craft weapons and tools (enhancing adaptibility and enabling us to hunt way outside our weight class) - Communicate/coordinate effectively in small groups - Do sensor processing and the above with a low energy budget Our brains are *not* selected to…

Just because there are some parts of our brains that are not needed for an AGI...

Doesn't mean that there aren't some part of our brains that are needed for an AGI, and are not present in the current crop of LLM.

Re: 100x defect tolerance: How we solved the yield problem

#163

Post by Thews »

Retric wrote: post #160
idiotsecant wrote: post #151

Earlier quoted context omitted.

>it's gonna rewire the way everything else operates too. This is the part that I think a lot of very tech literate people don't seem to get. I see people all the time essentially saying 'AI is just autocomplete' or pointing out that some vaporware ai company is a scam so surely everyone is. A lot of it is scams and flash in the pan. But a few of them are going to transform our lives in ways we probably don't even ant…

I’m not so sure it’s going to even do that much. People are currently happy to use LLM’s, but the outputs aren’t accurate and don’t seem to be improving quickly. A YouTuber watch regularly includes questions they asked Chat GPT and very single time there’s a detailed response in the comments showing how the output is wildly wrong from multiple mistakes. I suspect the backlash from disgruntled users is going to hit th…

Using function calls for correct answer lookup already practically eliminates this, it's not wide spread yet, but the ease of doing it is already practical for many.

New models aren't being trained specifically on single answers which will only help.

The expense for the larger models is something to be concerned about. Small models with function calls is already great, especially if you narrow down what they are being used for. Not seeing their utility is just a lack of imagination.

Re: 100x defect tolerance: How we solved the yield problem

#164

Post by stonemetal12 »

emkee wrote: post #155
kragen wrote: post #145

Earlier quoted context omitted.

I did eventually get an LLM to produce what seems to be a correct diagram of a sentence it had never seen, but it took about ten tries. Grammatical analysis seems to have happened correctly every time, but getting to a usable diagram was difficult. (I know that it's generally rude to include LLM output in HN comments, but in this case I think it's essential supporting material to elevate the discussion of LLM capabil…

Having only taken one syntax class for fun in college, I find this pretty impressive. Generating syntax trees was never a trivial task for me (but I was just a CS major who needed a credit). Slightly related, but I have also never had ChatGPT successfully generate ASCII art, even with extensive conversation.

If you ask it to draw a dinosaur it does an okay brontosaurs.

                 __
                / _)
       _.----._/ /
      /          /
   __/ (  | (  |
  /__.-'|_|--|_|
Asking for a Tyrannosaurus Rex gives you more or less the same brontosaurs:

                   __
                  / _)
         _.----._/ /
       /          /
   __/ (  | (  |
  /__.-'|_|--|_|

Re: 100x defect tolerance: How we solved the yield problem

#165

Post by CamperBob2 »

kragen wrote: post #145
ChuckMcM wrote: post #117

Earlier quoted context omitted.

It could diagram a sentence it had never seen.

I did eventually get an LLM to produce what seems to be a correct diagram of a sentence it had never seen, but it took about ten tries. Grammatical analysis seems to have happened correctly every time, but getting to a usable diagram was difficult. (I know that it's generally rude to include LLM output in HN comments, but in this case I think it's essential supporting material to elevate the discussion of LLM capabil…

(I know that it's generally rude to include LLM output in HN comments, but in this case I think it's essential supporting material to elevate the discussion of LLM capabilities above "yes it is", "no it isn't".)

You just have to be prepared to take a karma hit for it. The audience here does not consist largely of 'hackers', but seems to skew toward the sort of fearful, resentful reactionaries that hacker culture traditionally opposes.

I will say I wouldn't peg ChuckMcM as being one of the reactionaries, though. That would be an unpleasant surprise.

As far as the diagram goes, my guess is that sentence diagrams were underrepresented in the training corpus. Diagramming sentences was already out of fashion when I was in school in the 1980s -- in fact, I don't recall ever having done it. The model is struggling much the same way you'd expect a grade-school student (or me, I guess) to struggle upon being asked to perform the task for the first time.

Knowing when to say "I don't know how to do that" is still a foundational weakness of LLMs, but I don't expect it to remain unaddressed for long. We will see improvement in that area, sooner or later. The anklebiters will respond by moving their goalposts and hitting the downvote button as usual.

Re: 100x defect tolerance: How we solved the yield problem

#167

Post by aaroninsf »

The number of people ITT this thread who have absorbed the world-weary AI-is-a-bubble skepticism...

I'm just gonna say, with serene certainty,

the economic order we inhabit going through phase change is certain. From certain myopic perspectives we can shoehorn that into a narrative of cyclical patterns in the tech industry or financial markets etc etc.

This is not going to be that. No more than the transformation of American retail can be shoehorned to kind of look like it used if you don't know anything at all about what contemporary international trade and logistics and oligopoly actually mean in terms of what is coming into your home from where and why it is or isn't cheap.

Where we'll be in 10, 20, years is literally unimaginable today; and trying to navigate that wrt traditional landmarks... oof.

Re: 100x defect tolerance: How we solved the yield problem

#168

Post by CamperBob2 »

ChuckMcM wrote: post #117
CamperBob2 wrote: post #109

Earlier quoted context omitted.

it cannot ever get to an AGI level that you'd assume to be competitive to a human, even most animals. Suppose you turn out to be wrong. What would convince you?

It could diagram a sentence it had never seen.

A language model can generate valid sentences that have never existed before -- not just syntactically or statistically valid sentences in a Markovian sense, but semantically valid ones. How could it possibly do that without an implicit awareness of all aspects of language, including grammar and parts of speech?

That implicit awareness is indistinguishable from knowledge, at least to the extent it yields correct results.

And the fact that it was inferred from examples rather than taught explicitly is indistinguishable from intelligence.

Re: 100x defect tolerance: How we solved the yield problem

#169

Post by timerol »

Qwertious wrote: post #113
Scaevolus wrote: post #8

Earlier quoted context omitted.

Why does their chip have to be rectangular, anyways? Couldn't they cut out a (blocky) circle too?

You need a rectilinear polygon that tessellates, and has the fewest sides possible to minimize the number of cuts necessary. And it would probably help the cutting if the shape is entirely convex, so that cuts can overshoot a bit without damaging anything. That suggests a rectangle is the only possible shape.

Why does it need to tessellate if there's only one chip per wafer?
Post reply on HN