[HN Gopher] Efficient Reasoning with Hidden Thinking
       ___________________________________________________________________
        
       Efficient Reasoning with Hidden Thinking
        
       Author : fofoz
       Score  : 117 points
       Date   : 2025-02-03 16:06 UTC (6 hours ago)
        
 (HTM) web link (arxiv.org)
 (TXT) w3m dump (arxiv.org)
        
       | moolimon wrote:
       | I feel like this is the obvious next step for chain of thought
       | reasoning. Excited to see work on models that try and transform
       | the intermediate thinking space tokens, down to language.
       | Allowing us to still try and see what's happening inside the
       | "mind" of the LLM, if that process is even possible to map to
       | language anymore. I also wonder what the implications of this
       | research are on chain of thought reasoning with reinforcement
       | learning, since from my understanding many of the reward
       | mechanisms set up during reinforcement learning are around the
       | structure of thought process.
        
         | Davidzheng wrote:
         | WRT last sentence: I think the recent breakthroughs are
         | precisely not caring at all about the cot itself and evaluating
         | only the end product, allowing the model to develop a method of
         | reasoning which is not necessarily procured by human data
         | distribution (has the benefit of allowing it to collapse to a
         | "personalized" reasoning pattern)
        
         | knowaveragejoe wrote:
         | I'm a noob hobbyist, but in theory couldn't SAEs or similar MI
         | constructs learn to decode the "thinking" tokens into something
         | more resembling the CoT they originally encoded? Or am I
         | completely off the mark?
        
       | scribu wrote:
       | Would be curious to know how this stacks up against Coconut [1]
       | which also uses latent space for reasoning.
       | 
       | [1] https://arxiv.org/abs/2412.06769
        
         | kevmo314 wrote:
         | Definitely curious, this looks very similar to Coconut, even
         | down to the CoT encoding process in Figure 2. They go into a
         | lot more detail though, seems like parallel innovation.
        
         | esafak wrote:
         | I'm behind on reading but don't all models use continuous
         | embeddings to represent reasoning?
        
           | winwang wrote:
           | I believe the "continuous" in Coconut means that the CoT is
           | in the continuous latent space, instead of being on output
           | tokens (see Fig. 1).
        
       | jakobschwich wrote:
       | Seems like a promising next step.
        
       | aradox66 wrote:
       | Could someone ELI5? It sounds like they generate a compressed
       | token which represents a whole "thought" rather than elaborating
       | the entire "thought" in actual language. Is that right?
        
         | ipunchghosts wrote:
         | Currently, when AI models solve problems, they write out long
         | chains of thoughts (like showing their work in math). While
         | helpful, this takes up a lot of computing power.
         | 
         | Heima does something clever - instead of writing out long
         | explanations, it compresses each step of thinking into a single
         | "thinking token." Think of it like using a shorthand symbol
         | instead of writing out a full sentence.
        
           | Ancapistani wrote:
           | I've been doing a lot of introspection lately about how _I_
           | think. I lack the terminology here unfortunately, but your
           | description here sounds super familiar:
           | 
           | > instead of writing out long explanations, it compresses
           | each step of thinking into a single "thinking token." Think
           | of it like using a shorthand symbol instead of writing out a
           | full sentence.
           | 
           | I have clear memories of how cognition worked for me before I
           | understood spoken language. I recall thinking in concepts -
           | kind of a weird mix of forms, motions, and intent. I know
           | this sounds metaphysical, but that's not my intent. I just
           | don't have the words to explain it.
           | 
           | I wish I did, though, because my very early memories of self-
           | awareness certainly seem to map well onto the current state
           | of AI development.
        
             | fzzzy wrote:
             | Plenty of people don't think with an internal monologue or
             | internal imagery.
        
       | Davidzheng wrote:
       | Probably not needed in the end to reason in latent space. Unless
       | constrained by human preference/SFT data, RL spontaneously should
       | create new additions to language to help with new reasoning
       | methods/new concepts invented by the system.
        
         | numba888 wrote:
         | > RL spontaneously should create new additions to language to
         | help with
         | 
         | Yes, but it may take millions of years. One of the main reasons
         | of LLMs success is their amazing trainability. For every input
         | token it produces predictable output. I.e. loss. While most RL
         | techniques go one by one 'state'. For not tokenized output we
         | cannot predict what it should be. Thus it can be trained only
         | through the next tokens. Which makes it probably unstable and
         | expensive to train, limiting the length of 'continuous' part.
         | But looks like it's still a good idea to have.
        
         | pishpash wrote:
         | Can definitely create new math concepts, for example.
         | 
         | "Let two dhdud and three otincjf be called a Uhehjfj"
        
       | antirez wrote:
       | Cool, but isn't this encoding a potentially very long thinking
       | process into a fixed embedding? Intuitively should not work as
       | well.
        
       | byschii wrote:
       | isn't this dangerous? isn't the efficiency given at the expense
       | of safety and interpretability?
       | 
       | https://arxiv.org/abs/2412.14093 (Alignment faking in large
       | language models)
       | 
       | https://joecarlsmith.com/2024/12/18/takes-on-alignment-fakin...
       | 
       | PS I m definitely not an expert
        
         | achierius wrote:
         | Yes, but what do you think matters more: - Safety and (in the
         | long run) human lives - More papers ?
        
           | jononor wrote:
           | Turns out we are the main paperclip optimizers...
        
             | anticensor wrote:
             | or goat compressors:
             | https://x.com/GZilgalvis/status/1883107575010619649
        
         | patcon wrote:
         | Yeah, agreed. The limits of human minds constrain language. To
         | allow these things to reason outside words is in my intuitions
         | a tactic with more abundant paths toward super intelligence,
         | and the exact sort of path we'll have a harder time monitoring
         | (we'll need fancy tools to introspect instead of just watching
         | it think)
         | 
         | My current thinking is that I would support a ban on this style
         | of research. Really hard to set lines for regulation, but this
         | feels like an easy and intuitive place to exercise caution
        
         | winwang wrote:
         | Depends on if we can interpret the final hidden layer. It's
         | plausible we evolve models to _have_ interpretable
         | (final/reasoning) hidden layers, just that they aren't
         | constrained to the (same representation of) input/output
         | domains (i.e. tokens).
        
         | IshKebab wrote:
         | I don't see how it is any more dangerous than the already
         | existing black-box nature of DNNs.
        
         | nowittyusername wrote:
         | the hidden tokens can be decoded to English language if the
         | user wants to see the thinking process.
        
         | numba888 wrote:
         | > isn't this dangerous? isn't the efficiency given at the
         | expense of safety and interpretability?
         | 
         | Final text is only a small part of model's thinking. It's
         | produced from embeddings which probably have much more in them.
         | Each next token depends not only on previous, but all the
         | intermediate values for all tokens. We don't know them, they
         | are actually important and represent inner 'thinking'. So, LLM
         | is still a black box. The result is usually A because of B.
         | Sort of explanation for A, but where B came from we can only
         | guess.
        
         | swagmoney1606 wrote:
         | We should always be able to clearly understand and interpret
         | all of the thinking leading to an action done by an AI. What
         | would the point be if we don't know what it's doing, just that
         | it is doing "something"
        
       | deoxykev wrote:
       | I don't think autoregressive models have a fundemental difference
       | in terms of reasoning capability in latent space vs token space.
       | Latent space enables abstract reasoning and pattern recognition,
       | while token space acts as both the discrete interface for
       | communication, and as a interaction medium to extend, refine and
       | synthesize high order reasoning over latent space.
       | 
       | Intuively speaking, most people think of writing as a
       | communication tool. But actually it's also a thinking tool that
       | helps create deeper connections over discrete thoughts which can
       | only occupy a fixed slice of our attention at any given time.
       | Attentional capacity the primary limitation-- for humans and
       | LLMs. So use the token space as extended working memory. Besides,
       | even the Coconut paper got mediocre results. I don't think this
       | is the way.
        
         | bravura wrote:
         | I appreciate your argument, but add the following nuance:
         | 
         | Latent space reasoning can represent and manipulate UNCERTAINTY
         | more concisely and elegantly than token space reasoning.
        
       | another_poster wrote:
       | Is "multimodal reasoning" as big a deal as it sounds? Does this
       | technique mean LLMs can generate chains of thought that map to
       | other modalities, such as sound and images?
        
         | exclipy wrote:
         | It'd be cool to see its reasoning for solving visual puzzles,
         | as imagery.
        
       | gunalx wrote:
       | I would be interrested in seeing how a combined latent space and
       | traditional gpro cot could perform vs just one of either.
       | 
       | My intuition is still that latent space would be better at
       | emulating larger models with fewer params, and cot helping
       | refining the output after latent space.
       | 
       | Combined it would kinda being able to think about a problem.
       | Throw down a draft then refine it.
        
       | thom wrote:
       | Very importantly here they provide a ways of decoding the encoded
       | thought tokens, so you're not really losing explanatory power or
       | debuggability. As much as OpenAI want to present hidden chain of
       | thought as some sort of long term advantage or safety feature,
       | it's horrible when you want to understand how a model came to
       | some insane conclusion.
        
       | _KnighT_ wrote:
       | I'm new to this topic. Can someone help me understand this
       | sentence?
       | 
       | "Meanwhile, through the next-token prediction constraint, the
       | explicit textual symbols of the hidden representations for Heima
       | Encoder are aligned to the text of the corresponding special
       | tokens {<CoT>(k)} in vocabulary, while the hidden representations
       | contained in hidden states of thinking tokens remain distinct and
       | variable depending on the inputs"
       | 
       | I understand that they have fine-tuned the MLLM to produce, in
       | response to each query and image input, the CoT "thinking tokens"
       | in addition to the answer.
       | 
       | How does that establish an association between the thinking
       | tokens and the original plain-English CoT statements?
       | 
       | The second clause seems to say that the thinking tokens encode
       | information that is "distinct and variable depending on the
       | inputs." Is my interpretation correct?
        
       ___________________________________________________________________
       (page generated 2025-02-03 23:00 UTC)