8 comments

  • fabsalvadori 17 minutes ago
    There is a useful engineering consequence here beyond terminology.

    If intermediate tokens are not a faithful representation of the computation, then they are a pretty bad audit artifact too. We probably shouldn't be trying to make the model's internal narration more interpretable., but rather the computation around it more reproducible.

    Record the actual inputs, model/version/configuration, tool observations and outputs, then make the execution replayable enough that differences between runs can be isolated.

    In other words, don't ask the model to explain what it thought, and instead make the system able to show what actually happened.

  • smugtrain 11 hours ago
    Strong dislike for papers that tell me what to do in the title, especially when even the paper admits a loose correlation of the intermediate tokens compared to solution correctness. My solutions work and they speak for themselves.
  • tokai 2 minutes ago
    Pretty wild dressing a blog post up as a scientific paper.
  • florianherrengt 23 hours ago
    > While a human may say “aha” to indicate exactly a sudden internal state change, this interpretation is unwarranted for models which do not have any such internal state, and which on the next forward pass will only differ from the pre-aha pass by the inclusion of that single token in their context. Interpreting the “aha” moment as meaningful exemplifies the long-neglected assumption about long CoT models – the false idea that derivational traces are semantically meaningful, either in resemblance to algorithm traces or to human reasoning.

    This paper addresses something that has always bothered me about LLMs. You read their reasoning, see something like “Wait, that’s wrong” and then watch them make the exact mistake they just identified.

    • Jeff_Brown 22 hours ago
      By itself, "aha" carries no insight, but the insight is probably stated immediately after it. In that case the aha is semantically useful, by identifying the insight it is near.
      • deaton 32 minutes ago
        It really isn't useful though, unless it is a summary. At best it is a semantic trick to tell the next iteration to come up with something smart.
      • paimapi 20 hours ago
        it's a rhetorical heuristic that a writer should know to use when directing a reader to a declarative that they want them to pay attention to, usually because it's a non-obvious or roundabout insight

        when utilized by AI, it's a probabilistic output and it's variable whether or not that rhetorical trick is useful. it also pushes a non-skeptical reader to focus too much on the following text or even to believe that they, themselves, derived some insight. this is effectively a kind of persuasive sophistry which is not helpful - adding rules around it prevents people from deluding themselves with AI

        • ghostpepper 29 minutes ago
          Did not read the paper so apologies if this is covered but isn't it possible that there is some recognizable semantic pattern in the training data where an "aha" is often followed by a subtle semantic shift that proves closer to the original premise in some critical way, and by emitting the "aha" token the model causes itself to produce such a subtle semantic shift that pushes the subsequent reasoning closer to the desired response?
        • abitmoa 17 hours ago
          It amounts to noise overall, but it has further unwanted and potentially misleading 'properties'. I think it's rather sobering to see how much bandwidth is still being wasted.
      • wizzwizz4 20 hours ago
        > but the insight is probably stated immediately after it.

        If the intermediate tokens represent reasoning or thought, you would expect "aha" to occur after the thoughts that led to the realisation, including the thoughts encoding the explanation: they don't have any other state. There is no reason to draw the conclusion you've drawn. Furthermore, what LLMs are doing isn't thought.

  • basedpolymer 20 hours ago
    The anthropomorphization of LLMs should be discouraged as much as possible. It perpetuates bad practices and encourages the use of these bots for tasks they are not intended for (particularly as chatbots).

    Thinking traces should be treated as black boxes. There is no point in reading them. Only the LLMs’ conclusions are relevant. This is particularly true of Opus 5, which employs reasoning that seems highly questionable but very often reaches excellent conclusions (compared to its peers)

    • FloorEgg 18 hours ago
      Sometimes I monitor thinking traces for misunderstandings (missing context / bad assumptions). If it's going to go off on a ~20 min task and I can catch it's going in the wrong direction in the first minute I save a lot of tokens and wasted time. I don't monitor the whole thing, mostly just the first bit to see if there was a gap or misalignment in intention.

      As an aside, anthropomorphization has nothing to do with my motivations.

    • qarl2 3 hours ago
      > The anthropomorphization of LLMs should be discouraged as much as possible.

      And yet, they have extensive human-like behavior. If you treat them nicely or encourage them, they perform better.

      Ignoring that human-like behavior is wrong headed.

      • fedpost 42 minutes ago
        I think you're ending that train of thought too early. Why does this occur?

        Well... We can hypothesize that these things are largely trained on internet dialogue so there's probably some correlation between threads where people are not flaming each other and the quality of the replies. They're just statistical engines so anything you can do to raise the odds of a helpful next token...

        I'm essentially just making shit up here, maybe it's right, maybe it isn't, but rather than saying "it's human and we should treat it so" we're trying to get to the ground truth of how it works.

      • thaanpaa 44 minutes ago
        That's not a consequence of an LLM. It's a consequence of the training data. In fact, I would argue that the latest models aren't nearly as sensitive to the tone of input anymore. It's an issue that has been addressed by better curating training data.
      • andai 45 minutes ago
        A while back I made an "OpenClaw in 50 lines" by just wrapping Claude Code in a Telegram bot.

        I asked it for the weather. "I don't know that. I'm just a programmer."

        I added "believe in yourself, you can do anything" to sysprompt, suddenly it had the confidence to Google the weather...

  • clhodapp 22 hours ago
    Seems like they are closer to scratch than reasoning... Generating some scratch to draw from helps make it easier to compute the real answer.
    • forgotTheLast 2 hours ago
      That's my personal theory too. The model is stuffing its own context with vaguely related tokens, which helps the attention heads retrieve the right tokens.
    • cyanydeez 20 hours ago
      I assume theyre searching the local gradient to see if theres a better descent before proceeding.
      • eigenspace 8 hours ago
        LLMs dont do gradient descent to generate tokens.

        They are trained by gradient descent, but inference doesnt involve it.

      • c0_0p_ 12 hours ago
        I don't think there's anything like that going on. They just word vomit into a secondary area, and then there is an internal prompt that says "clean this up and summarize for the user".
  • Terr_ 23 hours ago
    I've been calling them film noir internal monologues, within the documents being generated by the LLM which happen to look like movie scripts.

    In other words, it isn't qualitatively different from character dialogue. "Keep cheese on your pizza by using glue" is the same problem regardless of whether the script calls for the character to speak it out-loud or not.

  • porridgeraisin 20 hours ago
    Related:

    Poster side dialogue and Q&A about this work at ICML.

    https://news.ycombinator.com/item?id=49277303