[HN Gopher] Continuous Autoregressive Language Models
       ___________________________________________________________________
        
       Continuous Autoregressive Language Models
        
       Author : Anon84
       Score  : 103 points
       Date   : 2025-11-05 21:49 UTC (8 days ago)
        
 (HTM) web link (arxiv.org)
 (TXT) w3m dump (arxiv.org)
        
       | mentalgear wrote:
       | Very interesting. Also I find these training parameters quite
       | elegant:
       | 
       | - Diversity: This term encourages the model to generate a diverse
       | set of samples, preventing mode collapse. - Fidelity: This term
       | rewards the model for making predictions that are close to the
       | ground-truth
       | 
       | I'm wondering if a continuos next-vector generative approach also
       | increase innate "reasoning" capabilities of the model, since it
       | could potentially capture more of the semantics of the data vs
       | just tokens.
        
         | barrenko wrote:
         | And may be even more adapted to sorts of RL finetuning?
        
           | mike_hearn wrote:
           | They say this technique isn't compatible yet with RL because
           | you can't adjust the logits. So no GRPO I guess, which is
           | going to be the biggest issue. An LLM with no RL applied
           | isn't going to be that useful.
        
       | suddenlybananas wrote:
       | The technique of compressing tokens down reminds me a bit of byte
       | latent transformers
        
       | killerstorm wrote:
       | Would be interesting to combine it with Reasoning In the Latent
       | Space: feed the vector from the output layer of transformer back
       | to input.
       | 
       | Obviously, you can't do it in pre-training. But you can add it
       | later as an optional 'extra' vector, I think. E.g.
       | `input_embedding + MLP(prev_output) * alpha`. Alpha is zero
       | during pre-training.
        
         | vessenes wrote:
         | I like this plan, but don't you already have this from the
         | input vector in the prompt, at least if the inference is 'chunk
         | wise' - generating a latent space vector, decoding it,
         | outputting it, doing the next one.
         | 
         | What if you trained a separate thinking phase using the auto
         | encoder, though? Might be more efficient, and then you've got
         | it using neuralese internally.
         | 
         | Actually, reading the (summary) paper - they tried your idea
         | and had trouble with it for a different reason:
         | > Once the generative head predicts the next vector , a natural
         | next step would be to feed it directly as input to the
         | Transformer for predicting . However, we found that the model
         | struggles to unpack the semantic information from such a
         | compact representation. Instead, we ground the autoregressive
         | process back in the more structured discrete space, where the
         | predicted  is passed through the autoencoder to reconstruct the
         | K tokens.
        
       | notrealyme123 wrote:
       | Congratulations for the authors, but damit, there goes a good
       | idea ^^
        
       | vatsachak wrote:
       | K being fixed here seems like it will eventually be done away
       | with
       | 
       | When I'm thinking about math proofs, sometimes I can have a
       | single idea which can be unfolded into a hundred lines of proof
       | 
       | Maybe I'm getting the wrong analogy here, but if vectors = ideas
       | then K should depend on the vector
        
       | mike_hearn wrote:
       | If they can reinvent RL so it works with this then I guess the
       | big labs will be all over it, as ~halving inference costs would
       | be huge (especially if Ed Zitron's leaked OpenAI inf costs are
       | accurate). Potentially the difference between inferencing being
       | profitable and loss making. It's an elegant approach.
       | 
       | I also wonder how far they can push K if other aspects are
       | tweaked. The approach of just doubling each parameter each time
       | leaves a lot of space between the chosen value and the next value
       | known to not work.
        
       ___________________________________________________________________
       (page generated 2025-11-13 23:01 UTC)