Top
Best
New

Posted by frenchmajesty 9 hours ago

Energy-Based Reasoners: 385x more efficient inference(energy-based-reasoners.github.io)
258 points | 41 commentspage 2
saithound 8 hours ago|
385x? Them's rookie numbers. Classical energies learn to reason 1000x faster than looped models. [1] Of course, all for a very generous definition of reasoning, that has almost nothing in common with the thing most people associate with the term nowadays. [2]

[1] https://arxiv.org/abs/2611.17309

[2] https://developers.openai.com/api/docs/guides/reasoning?api-...

yorwba 7 hours ago||
The common definition of "reasoning" for all three is "performing a sequence of intermediate steps with multiple passes through the model before outputting an answer." The main difference is that the reasoning models in TFA and [1] are puzzle-solving models, not language models.
saithound 6 hours ago||
Well, yes, that's the very generous definition of reasoning I was referring to, under which almost any iterative computation could count as reasoning.
ricardobeat 7 hours ago||
Did those two papers get published six days apart?
cobychapple 4 hours ago||
title says 385x and the page says 345x, I guess the compression is lossy?
netbioserror 6 hours ago||
The more abstractly reasoning can be modeled, the better. It means knowledge can be externalized to discrete knowledge bases and queried when needed, and models can be run locally with lower and lower energy hardware.

The question is, how integral to reasoning is compressing vast quantities of knowledge to the latent space?

serhii-a8 1 hour ago||
[flagged]
cnczeric 3 hours ago||
[flagged]
amajid012 3 hours ago||
[flagged]
Akseleu_omirbay 6 hours ago||
[flagged]
seventuning 4 hours ago|
[flagged]