5 ms·

> If you simply scaled it up it would handle that fine. (it's costly to scale up this architecture). What compelling reason do we have to believe this is true,

> If you simply scaled it up it would handle that fine. (it's costly to scale up this architecture). What compelling reason do we have to believe this is true, when this is not how animal or human brains work? We humans don't get to "scale up" our architecture with billion-fold increases in our training set. We have more sophisticated hardware, sure, but we don't actually have more data. I think we've just pursued this path because it's the one we can do. We have more compute, we have more data, so we use it. And that's fine. But it still baffles me that the field isn't more interested in trying to move in the direction of less data, more sophistication in structures, when we already know that works at some level because that's how we work.

catalogia 6y ago
From what I understand, the human brain runs off somewhere around 20 watts. Maybe our silicon is really really bad, but I'm more inclined to think we're missing part of the puzzle still.
- FeepingCreature 6y ago
  Keep in mind that nature started out with access to a molecular nanoassembler, and then iterated on that technology for billions of years.
  - catalogia 6y ago
    Even so, that's massive discrepancy in efficiency if it's really just a matter of the hardware. I think we're missing something still.
nullc 6y ago
> What compelling reason do we have to believe this is true, when this is not how animal or human brains work? The problem of GPT having zero idea about far-past text isn't some spooky emergent problem: It literally has zero access to the far-enough past. GPT is completely memoryless: it has access to the past 1024 symbols. It predicts the next symbol given the last 1024-- and that's it. When it goes to predict the next symbol, it has literally zero access to what came before except for the effect it had on the text still in the window. (symbols are usually words, but can also be letters, due to the compressed input). The result is that when text ends up too far back it will totally "forget it". If you scale it up to more symbols that effect will go away. Maybe some other issues arise at some scale, but it seems unlikely that they'd be all that similar to that totally-forgetting effect. I'm not arguing that just scaling it up the best approach-- but just that faulting its forgetting effect is not the most compelling criticism because that one almost certainly can be addressed by scaling. An architecture that could give it memory would be interesting-- people have shown impressive results by making it "show its work" effectively making the conversation into external memory-- but it's less obvious how to train it to use memory.
- Barrin92 6y ago
  >I'm not arguing that just scaling it up the best approach-- but just that faulting its forgetting effect is not the most compelling criticism because that one almost certainly can be addressed by scaling. it can't because the size of the system obviously completely explodes to practically infinity if you try to learn the past just by absorbing random factoids. This is clearly not how human memory works. If I ask you "did you float through your bedroom on march 2nd 2015 at 11 pm?", you don't consult some sort of history you have burned in a neural net, you make an inference using the laws of physics to conclude that you didn't because that's impossible. These neural nets don't reason this way, they don't have high level understanding of ontologies or laws and they can't make inferences like this, I haven't tested GPT-3 but I assume it can't even reliably solve basic algebra that isn't found in the dataset. So memory is a function of combining high level understanding of the world with data, not shoving history into an encoder.
  - nullc 6y ago
    > I assume it can't even reliably solve basic algebra that isn't found in the dataset. ::sigh:: It can, even though there are specific technical reasons that working with both numbers and exact answers is hard for it. (The way the input text is encoded combines digits in an inconsistent and formatting specific way, and the way text is generated isn't just to take the models most likely output but to randomly sample among the top couple most likely outputs, which is an obvious handicap when there is a single right answer.) https://pbs.twimg.com/media/EdHuMgsWsAEk-29?format=png&name=large https://pbs.twimg.com/media/EdHuMgsWsAEk-29?format=png&name=... https://pbs.twimg.com/media/EdII5jhXkAA8eKv?format=png&name=medium https://pbs.twimg.com/media/EdII5jhXkAA8eKv?format=png&name=... (lines starting with > are the human)
- abernard1 6y ago
  > It literally has zero access to the far-enough past. > GPT is completely memoryless: it has access to the past 1024 symbols This is the critique I have of that. This model is so huge, the "memory" is encoded in the trained model parameters. Since there is so much repetition and structure in comments in the corpus, I believe we are seeing a large scale social memory compressed in this {model parameters x symbol memory} structure. That space is gigantic. > If you scale it up to more symbols that effect will go away. This though still remains to be proven. Does any significant further correlation of note happen when you have {175B model parameters x 1M symbol memory} structure? No one knows quite yet. But we do for sure know that humans can learn things with an extraordinarily smaller "training" set.

catalogia 6y ago

From what I understand, the human brain runs off somewhere around 20 watts. Maybe our silicon is really really bad, but I'm more inclined to think we're missing part of the puzzle still.

FeepingCreature 6y ago
Keep in mind that nature started out with access to a molecular nanoassembler, and then iterated on that technology for billions of years.
- catalogia 6y ago
  Even so, that's massive discrepancy in efficiency if it's really just a matter of the hardware. I think we're missing something still.

nullc 6y ago

> What compelling reason do we have to believe this is true, when this is not how animal or human brains work? The problem of GPT having zero idea about far-past text isn't some spooky emergent problem: It literally has zero access to the far-enough past. GPT is completely memoryless: it has access to the past 1024 symbols. It predicts the next symbol given the last 1024-- and that's it. When it goes to predict the next symbol, it has literally zero access to what came before except for the effect it had on the text still in the window. (symbols are usually words, but can also be letters, due to the compressed input). The result is that when text ends up too far back it will totally "forget it". If you scale it up to more symbols that effect will go away. Maybe some other issues arise at some scale, but it seems unlikely that they'd be all that similar to that totally-forgetting effect. I'm not arguing that just scaling it up the best approach-- but just that faulting its forgetting effect is not the most compelling criticism because that one almost certainly can be addressed by scaling. An architecture that could give it memory would be interesting-- people have shown impressive results by making it "show its work" effectively making the conversation into external memory-- but it's less obvious how to train it to use memory.

Barrin92 6y ago
>I'm not arguing that just scaling it up the best approach-- but just that faulting its forgetting effect is not the most compelling criticism because that one almost certainly can be addressed by scaling. it can't because the size of the system obviously completely explodes to practically infinity if you try to learn the past just by absorbing random factoids. This is clearly not how human memory works. If I ask you "did you float through your bedroom on march 2nd 2015 at 11 pm?", you don't consult some sort of history you have burned in a neural net, you make an inference using the laws of physics to conclude that you didn't because that's impossible. These neural nets don't reason this way, they don't have high level understanding of ontologies or laws and they can't make inferences like this, I haven't tested GPT-3 but I assume it can't even reliably solve basic algebra that isn't found in the dataset. So memory is a function of combining high level understanding of the world with data, not shoving history into an encoder.
- nullc 6y ago
  > I assume it can't even reliably solve basic algebra that isn't found in the dataset. ::sigh:: It can, even though there are specific technical reasons that working with both numbers and exact answers is hard for it. (The way the input text is encoded combines digits in an inconsistent and formatting specific way, and the way text is generated isn't just to take the models most likely output but to randomly sample among the top couple most likely outputs, which is an obvious handicap when there is a single right answer.) https://pbs.twimg.com/media/EdHuMgsWsAEk-29?format=png&name=large https://pbs.twimg.com/media/EdHuMgsWsAEk-29?format=png&name=... https://pbs.twimg.com/media/EdII5jhXkAA8eKv?format=png&name=medium https://pbs.twimg.com/media/EdII5jhXkAA8eKv?format=png&name=... (lines starting with > are the human)
abernard1 6y ago
> It literally has zero access to the far-enough past. > GPT is completely memoryless: it has access to the past 1024 symbols This is the critique I have of that. This model is so huge, the "memory" is encoded in the trained model parameters. Since there is so much repetition and structure in comments in the corpus, I believe we are seeing a large scale social memory compressed in this {model parameters x symbol memory} structure. That space is gigantic. > If you scale it up to more symbols that effect will go away. This though still remains to be proven. Does any significant further correlation of note happen when you have {175B model parameters x 1M symbol memory} structure? No one knows quite yet. But we do for sure know that humans can learn things with an extraordinarily smaller "training" set.