5 ms·

I don't see why not. It's not taking a single answer from a database no, it's taking several based on probability and merging them into what it thinks we're loo

I don't see why not. It's not taking a single answer from a database no, it's taking several based on probability and merging them into what it thinks we're looking for. If you learn to multiply with code to perform one task, you can then apply that knowledge for a completely different task. It may look like solving a completely new problem but the LLM doesn't even see the difference. When you use the term "custom library" that might be you over-complicating the task. It's still just looking up function to do x, function to do y and applying it to the output. Don't get me wrong it's impressive where we're at but there's no need to exaggerate it as magic.

IanCal 3y ago
> . It's still just looking up function to do x, function to do y and applying it to the output. I mean no, no it isn't. I'm giving it info on how to construct data models with a custom library, so interacting with that is not using anything previously stored, and then giving it businesses/tasks to model as simple human descriptions. If you tell me that something which * Takes a human description of a problem * Describes back to me the overall structure and components required to solve it with a hierarchy * Converts that into code, correctly identifying where it makes sense for an address to be contained within a model or distinct and referenced * Correctly reuses previously created classes that are obviously not in its original dataset has no understanding or reasoning and it just regurgitating things it's seen before simply mashed together, I don't know what to say. Frankly > it's taking several based on probability and merging them into what it thinks we're looking for. I Sounds pretty much like understanding and reasoning to me. > but there's no need to exaggerate it as magic. I'm absolutely not saying there's magic. Humans aren't magic and they can do reasoning. I'm saying it's not just looking up text and regurgitating it. I think this is supported by things like othello-gpt, which builds an internal world model and outputs based on that.
- worksonmine 3y ago
  It's difficult for me to assess how original your library is without examples, maybe I could find the exact implementation on github within 30 minutes. But I've yet to see anything that isn't just mashing together stackoverflow and git repositories to save time. I get the same answers with less wordy fluff from a simple search, but I also know where to look. It's impressive that it knows the difference between "how many are 5 more apples than 10" compared to "how many percent are 5 apples of 10" (I don't know if it does, just assuming). But the first release also tried to reason why the weight of 1 pound of nails depends with the simple prompt "how much do 1 pound of nails weigh". That's most likely a perfect example of it mashing the classic "what weighs more, 1 pound of nails or 1 pound of feathers". It IS just looking in a database, and mashing it with some fluff. I'm happy to be proven wrong but I need more than your word for it. My experience is that as the topic gets more niche (less data in the training set) the worse the answers I get and it starts making things up based on probability. It doesn't reason in the sense I assume you're expecting.
  - IanCal 3y ago
    Have you had a look at othello gpt? https://thegradient.pub/othello/ https://thegradient.pub/othello/ It's a nice constrained example of a transformer learning a world model, not just looking up responses. > It's impressive that it knows the difference between "how many are 5 more apples than 10" compared to "how many percent are 5 apples of 10" (I don't know if it does, just assuming). But the first release also tried to reason why the weight of 1 pound of nails depends with the simple prompt "how much do 1 pound of nails weigh". That's most likely a perfect example of it mashing the classic "what weighs more, 1 pound of nails or 1 pound of feathers". Is there a formulation here that would get to a point where you'd think it's not just mashing things together? Are there elements of a simple question that would be required? Here's a slightly trickier one for it "Which weighs more, a pound of feathers or balloons made from one pound of rubber then filled with 100g of helium?" https://chat.openai.com/share/b841c96f-e46c-4adf-8ec3-8778ffee051b https://chat.openai.com/share/b841c96f-e46c-4adf-8ec3-8778ff...
    - worksonmine 3y ago
      Very impressive, but is it any more original than classic search engines' old trick of regular expressions to figure out if I mean the currency or weight when I ask "1 pound =" with the contexts USD or kg after "="? Does it understand the input, or are there just enough discussions in the training data to make it look like it is? I'm not convinced it's not the latter. It uses context to figure out we're trying to convert something to something else. Then it adds all those numbers up. Taking helium into consideration is no doubt interesting, but they've also polished that task since that was the common critique they got so very wrong with the first release (which I mentioned they had fixed). I'm not qualified to assess this part of the answer; > "If the balloons displace more than 100g of air when filled with helium, then they would effectively weigh less than if they were left empty. If they displace exactly 100g of air, then the balloons would have the same weight as if they were left empty." I don't know enough to understand how much 100g of helium is and how it behaves. And it doesn't try to explain it to me, it mentions it then takes the easy route assuming it's a trick question. What does that tell you? I guess there are similar discussions around and it gives me the summary. Why doesn't it tell me how much air it displaces under what circumstances? Temperature etc, it should be easy if it's not just a simple discussion on a random forum. A conversion regex could do it. This comment[1] has a very impressive example. But anything I'm qualified to assess has mostly been meh. If the fix is better training data does that mean it's reasoning or regurgitating? The mistakes it makes are what tells me how it works, not when it tricks me that it's correct. To me it's a very well polished search engine summary. [1]: https://news.ycombinator.com/item?id=37219351 https://news.ycombinator.com/item?id=37219351

IanCal 3y ago

> . It's still just looking up function to do x, function to do y and applying it to the output. I mean no, no it isn't. I'm giving it info on how to construct data models with a custom library, so interacting with that is not using anything previously stored, and then giving it businesses/tasks to model as simple human descriptions. If you tell me that something which * Takes a human description of a problem * Describes back to me the overall structure and components required to solve it with a hierarchy * Converts that into code, correctly identifying where it makes sense for an address to be contained within a model or distinct and referenced * Correctly reuses previously created classes that are obviously not in its original dataset has no understanding or reasoning and it just regurgitating things it's seen before simply mashed together, I don't know what to say. Frankly > it's taking several based on probability and merging them into what it thinks we're looking for. I Sounds pretty much like understanding and reasoning to me. > but there's no need to exaggerate it as magic. I'm absolutely not saying there's magic. Humans aren't magic and they can do reasoning. I'm saying it's not just looking up text and regurgitating it. I think this is supported by things like othello-gpt, which builds an internal world model and outputs based on that.

worksonmine 3y ago
It's difficult for me to assess how original your library is without examples, maybe I could find the exact implementation on github within 30 minutes. But I've yet to see anything that isn't just mashing together stackoverflow and git repositories to save time. I get the same answers with less wordy fluff from a simple search, but I also know where to look. It's impressive that it knows the difference between "how many are 5 more apples than 10" compared to "how many percent are 5 apples of 10" (I don't know if it does, just assuming). But the first release also tried to reason why the weight of 1 pound of nails depends with the simple prompt "how much do 1 pound of nails weigh". That's most likely a perfect example of it mashing the classic "what weighs more, 1 pound of nails or 1 pound of feathers". It IS just looking in a database, and mashing it with some fluff. I'm happy to be proven wrong but I need more than your word for it. My experience is that as the topic gets more niche (less data in the training set) the worse the answers I get and it starts making things up based on probability. It doesn't reason in the sense I assume you're expecting.
- IanCal 3y ago
  Have you had a look at othello gpt? https://thegradient.pub/othello/ https://thegradient.pub/othello/ It's a nice constrained example of a transformer learning a world model, not just looking up responses. > It's impressive that it knows the difference between "how many are 5 more apples than 10" compared to "how many percent are 5 apples of 10" (I don't know if it does, just assuming). But the first release also tried to reason why the weight of 1 pound of nails depends with the simple prompt "how much do 1 pound of nails weigh". That's most likely a perfect example of it mashing the classic "what weighs more, 1 pound of nails or 1 pound of feathers". Is there a formulation here that would get to a point where you'd think it's not just mashing things together? Are there elements of a simple question that would be required? Here's a slightly trickier one for it "Which weighs more, a pound of feathers or balloons made from one pound of rubber then filled with 100g of helium?" https://chat.openai.com/share/b841c96f-e46c-4adf-8ec3-8778ffee051b https://chat.openai.com/share/b841c96f-e46c-4adf-8ec3-8778ff...
  - worksonmine 3y ago
    Very impressive, but is it any more original than classic search engines' old trick of regular expressions to figure out if I mean the currency or weight when I ask "1 pound =" with the contexts USD or kg after "="? Does it understand the input, or are there just enough discussions in the training data to make it look like it is? I'm not convinced it's not the latter. It uses context to figure out we're trying to convert something to something else. Then it adds all those numbers up. Taking helium into consideration is no doubt interesting, but they've also polished that task since that was the common critique they got so very wrong with the first release (which I mentioned they had fixed). I'm not qualified to assess this part of the answer; > "If the balloons displace more than 100g of air when filled with helium, then they would effectively weigh less than if they were left empty. If they displace exactly 100g of air, then the balloons would have the same weight as if they were left empty." I don't know enough to understand how much 100g of helium is and how it behaves. And it doesn't try to explain it to me, it mentions it then takes the easy route assuming it's a trick question. What does that tell you? I guess there are similar discussions around and it gives me the summary. Why doesn't it tell me how much air it displaces under what circumstances? Temperature etc, it should be easy if it's not just a simple discussion on a random forum. A conversion regex could do it. This comment[1] has a very impressive example. But anything I'm qualified to assess has mostly been meh. If the fix is better training data does that mean it's reasoning or regurgitating? The mistakes it makes are what tells me how it works, not when it tricks me that it's correct. To me it's a very well polished search engine summary. [1]: https://news.ycombinator.com/item?id=37219351 https://news.ycombinator.com/item?id=37219351