Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Scaling up test-time compute with latent reasoning: A recurrent depth approach

arxiv.org

Post reply on HN
Print view View original

31–40 of 45 posts

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#31

Post by pilooch »

janalsncm wrote: post #7

One of the benefits of using thinking tokens compared to “thinking in a latent” space is that you can directly observe the quality of the CoT. In R1 they saw it was mixing languages and fixed it with cold start data. It would be hard to SFT this because you can only SFT the final result not the latent space. I also notice the authors only had compute for a single full training run. It’s impressive they saw such good…

It could be argued that "thinking" / CoT in latent space abstracts away the language issue, and that in fact language in reasoning steps doesn't matter. Latent tokens could actually be decoded afterwards to any target language. Much more powerful IMO.

On a side note, there's decent research on how well bilingual humans do actually think in both language, and are actually better at decisive thinking outside of their mother tongue.

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#32

Post by nsikorr »

ckrapu wrote: post #9

My opinion is that opaque reasoning is a prerequisite for many of the worst possible AI outcomes. We should make reasoning fully visible in the output space.

The paper suggests that that is still possible with the proposed architecture if needed.

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#33

Post by WhitneyLand »

One of the hoped for benefits of this approach that’s described later in the paper. It’s not fully fleshed out what this will mean but the prospect is tantalizing.

"On a more philosophical note, we hope that latent reasoning captures facets of human reasoning that defy verbalization, such as spatial thinking, physical intuition or (motor) planning. Over many iterations of the recurrent process, reasoning in a high-dimensional vector space would enable the deep exploration of multiple directions simultaneously, instead of linear thinking, leading to a system capable of exhibiting novel and complex reasoning behavior."

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#34

Post by miven »

ckrapu wrote: post #9

My opinion is that opaque reasoning is a prerequisite for many of the worst possible AI outcomes. We should make reasoning fully visible in the output space.

Chain of thought isn't exactly transparent either, you shouldn't fall into the pitfall of believing that the final sequence of tokens thinking about the task is the only processing the model actually performs during CoT.

There might me a lot of other hidden computations happening within the model's latents which may not immediately influence the predicted tokens but be relevant for the model's internal processing. And even disregarding that, the model is under no formal obligation to stick to the chain of thought it produced for its final decisions.

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#35

Post by radarsat1 »

timbilt wrote: post #2

Twitter thread about this by the author: https://x.com/jonasgeiping/status/1888985929727037514

If you keep digging in that thread the author posts a gist containing information on how the recurrence works:

https://gist.github.com/JonasGeiping/65959599ca637d72d50c96c...

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#38

Post by thomasahle »

Tostino wrote: post #30
thomasahle wrote: post #6

Earlier quoted context omitted.

> Latent / embedding-space reasoning seems a step in the right direction Might be good for reasoning, but it's terrible for interpretation / AI-safety.

Why is it any different to do 4 recurrent passes than having a model that is 4x deeper?

I guess the most interpretable is to have as shallow a model as possible, but with longer cot. It would be quite interesting seeing the trade-off between the two. Though, unfortunately, deeper is probably better.

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#39

Post by janalsncm »

pilooch wrote: post #31
janalsncm wrote: post #7

One of the benefits of using thinking tokens compared to “thinking in a latent” space is that you can directly observe the quality of the CoT. In R1 they saw it was mixing languages and fixed it with cold start data. It would be hard to SFT this because you can only SFT the final result not the latent space. I also notice the authors only had compute for a single full training run. It’s impressive they saw such good…

It could be argued that "thinking" / CoT in latent space abstracts away the language issue, and that in fact language in reasoning steps doesn't matter. Latent tokens could actually be decoded afterwards to any target language. Much more powerful IMO. On a side note, there's decent research on how well bilingual humans do actually think in both language, and are actually better at decisive thinking outside of their m…

I think another argument is that the CoT is simply unrolling the recurrent loop that this method uses, and doing an unembedding -> embedding -> unembedding during the decoding process.

So at best, using a recurrent loop is only saving you from doing the embedding -> unembedding at each token which is relatively small compared with the height of the decoder blocks.

Re: Scaling up test-time compute with latent reasoning: A recurrent depth approach

#40

Post by Manabu-eo »

HarHarVeryFunny wrote: post #4

Latent / embedding-space reasoning seems a step in the right direction, but building recurrence into the model while still relying on gradient descent (i.e. BPTT) to train it seems to create more of a problem (training inefficiency) than it solves, especially since they still end up externally specifying the number of recurrent iterations (r=4, 8, etc) for a given inference. Ideally having recurrence internal to the…

While not the main focus, see Section 6.1 and Figure 10 for a simple adaptative exit strategy for inference.

I imagine that they choose a fixed number of recurrent iterations during training for parallelization purposes. Not depending on the previous step to train the next is the main revolution about transformers vs LSTM (plus the higher internal bandwidth). But I agree that it might not be the most efficient model to train due to all that redundant work at large r.

Post reply on HN