4 ms·

I'm trying to tackle this problem more head-on, by outfitting LLMs with lambda calculus, stacks, queues, etc. directly in their internals, operating over their

I'm trying to tackle this problem more head-on, by outfitting LLMs with lambda calculus, stacks, queues, etc. directly in their internals, operating over their latent space. [1] I'll read your paper, but, LLMs famously fail horribly at "multi jump" reasoning, which to me means they can't reason at all. They can merely output a reflection of the human reasoning that was baked into the training data, and they can also recombine it combinatorially. Eager to see if you've solved this! [1] https://github.com/neurallambda/neurallambda https://github.com/neurallambda/neurallambda

eru 3y ago
Humans also fail much of the time at 'multi jump' reasoning. You have to prod them.
- littlestymaar 3y ago
  Except no human (non-colorblind at least) past three years old thinks bananas have the same color as the sky (see the example given in the repo, that's a mistake literally no human could make)
  - nielsbot 3y ago
    Maybe not no human :) But probably 99.99% of them.
    - littlestymaar 3y ago
      You're technically right, my two and a half had only be familiar with colors for six months, but I think it's fine to say that toddlers aren't reaching standard level of human intelligence ;)
  - mewpmewp2 3y ago
    For what it's worth, I tried it on ChatGPT and this was its response: "The color of the daytime sky is commonly blue. The common household fruit that is also blue would be blueberries. Blueberries typically grow in acidic soil. The pH of the soil they grow in is usually between 4.5 and 5.5."
    - neurallambda 3y ago
      It can get this simple example right if it does chain of thought. If you ask it to just output the answer without answering other bullet points, it will very likely not get it right. Chain of thought is duct tape to actual reasoning, and the errors/hallucinations compound exponentially. Try to get chatgpt to reason about concurrent state issues in programming. If it's not a well worn issue that it already memorized, it'll be useless. It's also perfect, for instance at Advent of Codes that it's memorized, and near 0% accuracy on new Advents.
      - mewpmewp2 3y ago
        But as a human being I couldn't ever do it without chain of thoughts. First I would have to bruteforce come up with fruits that have the required colour. Otherwise I am just random guessing. Also this problem I first tried to solve myself and I couldn't because I imagined sky to be light blue and blueberries are very dark if blue at all to me.
        eru 3y ago
        You could also go with grey or white etc for the sky.
still_grokking 3y ago
May I ask why there are NNs in your project at all? Just to heat up the planet and make Nvidia share holders even more happy? :-) I mean, what I've seen in the Readme makes sense. But doing basic computer stuff with NNs just makes the resource usage go brr by astronomical factors, for imho no reason, while making the results brittle, random, and often just completely made up. Also: Do you know about the (already 40 year old!) project Cyc? https://en.wikipedia.org/wiki/Cyc https://en.wikipedia.org/wiki/Cyc This software can indeed "reason". And it does not hallucinate; because it's not based on NNs.
- optimalsolver 3y ago
  Well with 40 years of work, I’m sure Cyc have had some really mind blowing results.
- mdekkers 3y ago
  +1 for cyc, genuinely awesome and overlooked
- neurallambda 3y ago
  "AI heats the planet"... really? You mean marginally? I'll assume you're asking in good faith. Using NNs allows this project to stand on the shoulders of giants: philosophically, mathematically, programmatically, but also I expect this to plug in to OSS LLMs, and leverage their knowledge, similarly to how a human child learns in a Pavlovian/intuitive response, and only later starts to learn to reason. Wrt inefficiency, training will be inefficient, but the programs can be extracted to CPU instructions / CUDA kernels during inference. Also, I'm interested in using straight through estimators in the forward pass of training, to do this conversion in training too. Cyc looks cool, but from my cursory glance, is it capable of learning, or is its knowledge graph largely hand coded? Neurallambda is at least as scalable as an RNN, both in data and compute utilization.
  - still_grokking 3y ago
    > Wrt inefficiency, training will be inefficient That's the "heating the planet part" I was referring to. :-) > but the programs can be extracted to CPU instructions / CUDA kernels during inference This just makes my original question more pressing: What are the NNs good for if the result will be normal computer programs? (Just created with astronomical overhead!) > Cyc looks cool, but from my cursory glance, is it capable of learning, or is its knowledge graph largely hand coded? The whole point is that it can infer new knowledge from known facts through a logical reasoning process. This inference process was run since 40 years. The result is the most comprehensible "world knowledge" archive ever created. Of course this wouldn't be possible to create "by hand". And in contrast to NN hallucinations there is real logical reasoning behind, and everything is explainable. I still don't get how some "dreamed up" programs from your project are supposed to work. Formal reasoning and NNs don't go well with each other. (One could even say they're opposites). Imho it's "real reasoning" OR "dreamed up stuff". How "dreamed up stuff" could improve "real reasoning"? Especially as the "dreamed up stuff" won't be included in the end results anyway, where only the formal things remain. To what effect are the NNs included in your project? (I mean besides the effect that the HW and energy demands will go through the roof, ending up billion times higher than just doing some lambda calculus directly…) And yes, these are genuine questions. I just don't get it. It looks for me like "let's do things maximally inefficiently, but at least we can put a 'works with AI' rubber stamp on it"; which is maybe good to collect VC money, but else? What do I overlook here?
CuriouslyC 3y ago
Chain of thought is basically reasoning as humans do it, the only difference is that unlike humans the model can't see that its output is wrong, abandon a line of reasoning and re-prompt itself (yet).
- skenderbeu 3y ago
  Multi agent LLMs talking to each other can already do this. It's just not cost feasible yet because it can lead to infinite loops and no solutions
- naasking 3y ago
  Various attempts at feeding their output back in to check itself have shown marked improvements in accuracy.
lamontcg 3y ago
> they can't reason at all. I'm pretty sure this is it. They don't understand even negation to begin with. In this thread someone asked an AI to output images of stereotypical American soccer fans and it drew all the Seattle fans with umbrellas: https://old.reddit.com/r/MLS/comments/1b10t68/meme_asked_ai_to_generate_stereotypical_insert/ https://old.reddit.com/r/MLS/comments/1b10t68/meme_asked_ai_... Seattle is pretty famously known for everyone not wearing umbrellas.

eru 3y ago

Humans also fail much of the time at 'multi jump' reasoning. You have to prod them.

littlestymaar 3y ago
Except no human (non-colorblind at least) past three years old thinks bananas have the same color as the sky (see the example given in the repo, that's a mistake literally no human could make)
- nielsbot 3y ago
  Maybe not no human :) But probably 99.99% of them.
  - littlestymaar 3y ago
    You're technically right, my two and a half had only be familiar with colors for six months, but I think it's fine to say that toddlers aren't reaching standard level of human intelligence ;)
- mewpmewp2 3y ago
  For what it's worth, I tried it on ChatGPT and this was its response: "The color of the daytime sky is commonly blue. The common household fruit that is also blue would be blueberries. Blueberries typically grow in acidic soil. The pH of the soil they grow in is usually between 4.5 and 5.5."
  - neurallambda 3y ago
    It can get this simple example right if it does chain of thought. If you ask it to just output the answer without answering other bullet points, it will very likely not get it right. Chain of thought is duct tape to actual reasoning, and the errors/hallucinations compound exponentially. Try to get chatgpt to reason about concurrent state issues in programming. If it's not a well worn issue that it already memorized, it'll be useless. It's also perfect, for instance at Advent of Codes that it's memorized, and near 0% accuracy on new Advents.
    - mewpmewp2 3y ago
      But as a human being I couldn't ever do it without chain of thoughts. First I would have to bruteforce come up with fruits that have the required colour. Otherwise I am just random guessing. Also this problem I first tried to solve myself and I couldn't because I imagined sky to be light blue and blueberries are very dark if blue at all to me.
      - eru 3y ago
        You could also go with grey or white etc for the sky.

still_grokking 3y ago

May I ask why there are NNs in your project at all? Just to heat up the planet and make Nvidia share holders even more happy? :-) I mean, what I've seen in the Readme makes sense. But doing basic computer stuff with NNs just makes the resource usage go brr by astronomical factors, for imho no reason, while making the results brittle, random, and often just completely made up. Also: Do you know about the (already 40 year old!) project Cyc? https://en.wikipedia.org/wiki/Cyc https://en.wikipedia.org/wiki/Cyc This software can indeed "reason". And it does not hallucinate; because it's not based on NNs.

optimalsolver 3y ago
Well with 40 years of work, I’m sure Cyc have had some really mind blowing results.
mdekkers 3y ago
+1 for cyc, genuinely awesome and overlooked
neurallambda 3y ago
"AI heats the planet"... really? You mean marginally? I'll assume you're asking in good faith. Using NNs allows this project to stand on the shoulders of giants: philosophically, mathematically, programmatically, but also I expect this to plug in to OSS LLMs, and leverage their knowledge, similarly to how a human child learns in a Pavlovian/intuitive response, and only later starts to learn to reason. Wrt inefficiency, training will be inefficient, but the programs can be extracted to CPU instructions / CUDA kernels during inference. Also, I'm interested in using straight through estimators in the forward pass of training, to do this conversion in training too. Cyc looks cool, but from my cursory glance, is it capable of learning, or is its knowledge graph largely hand coded? Neurallambda is at least as scalable as an RNN, both in data and compute utilization.
- still_grokking 3y ago
  > Wrt inefficiency, training will be inefficient That's the "heating the planet part" I was referring to. :-) > but the programs can be extracted to CPU instructions / CUDA kernels during inference This just makes my original question more pressing: What are the NNs good for if the result will be normal computer programs? (Just created with astronomical overhead!) > Cyc looks cool, but from my cursory glance, is it capable of learning, or is its knowledge graph largely hand coded? The whole point is that it can infer new knowledge from known facts through a logical reasoning process. This inference process was run since 40 years. The result is the most comprehensible "world knowledge" archive ever created. Of course this wouldn't be possible to create "by hand". And in contrast to NN hallucinations there is real logical reasoning behind, and everything is explainable. I still don't get how some "dreamed up" programs from your project are supposed to work. Formal reasoning and NNs don't go well with each other. (One could even say they're opposites). Imho it's "real reasoning" OR "dreamed up stuff". How "dreamed up stuff" could improve "real reasoning"? Especially as the "dreamed up stuff" won't be included in the end results anyway, where only the formal things remain. To what effect are the NNs included in your project? (I mean besides the effect that the HW and energy demands will go through the roof, ending up billion times higher than just doing some lambda calculus directly…) And yes, these are genuine questions. I just don't get it. It looks for me like "let's do things maximally inefficiently, but at least we can put a 'works with AI' rubber stamp on it"; which is maybe good to collect VC money, but else? What do I overlook here?

CuriouslyC 3y ago

Chain of thought is basically reasoning as humans do it, the only difference is that unlike humans the model can't see that its output is wrong, abandon a line of reasoning and re-prompt itself (yet).

skenderbeu 3y ago
Multi agent LLMs talking to each other can already do this. It's just not cost feasible yet because it can lead to infinite loops and no solutions
naasking 3y ago
Various attempts at feeding their output back in to check itself have shown marked improvements in accuracy.

lamontcg 3y ago

> they can't reason at all. I'm pretty sure this is it. They don't understand even negation to begin with. In this thread someone asked an AI to output images of stereotypical American soccer fans and it drew all the Seattle fans with umbrellas: https://old.reddit.com/r/MLS/comments/1b10t68/meme_asked_ai_to_generate_stereotypical_insert/ https://old.reddit.com/r/MLS/comments/1b10t68/meme_asked_ai_... Seattle is pretty famously known for everyone not wearing umbrellas.