Continuous Thought Machines · HackerPapers discussions

Continuous Thought Machines — mentioned by granitepail

Fri, 05 Jun 2026 19:21:45 GMT

A great deal of their research has been focused on zigging where others zag. Their paper "Continuous Thought Machines" (https://arxiv.org/abs/2505.05522, presented at NeurIPS) was posed specifically under the framing of there needing to be more fundamental research beyond squeezing as much as we can out of relatively vanilla transformer stacks. It is very biologically inspired and unique. Now that models are getting stronger at agentic work, it is very natural that many labs are chasing some form of auto-research.

Paper and discussion history. A mention is not an endorsement.

Continuous Thought Machines — mentioned by FuckButtons

Thu, 19 Mar 2026 17:08:14 GMT

You might be interested in this paper: https://arxiv.org/abs/2505.05522, in essence they demonstrate that a novel architecture that incorporates bounded convolution at the block level that has some variable time horizon on the number of iterations through the convolutional loop can be very effective and solves problems in a way that is much more similar to how humans do.

Paper and discussion history. A mention is not an endorsement.

Continuous Thought Machines — HN submission

Sun, 30 Nov 2025 03:10:49 GMT

Continuous Thought Machines

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.