Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens · HackerPapers discussions

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — mentioned by ForHackernews

Tue, 06 Oct 2026 18:33:53 GMT

I'm not sure anyone meaningfully understands it: "Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens" https://arxiv.org/html/2505.13775v3 > More interestingly, our experiments also show that models trained on corrupted traces, whose intermediate reasoning steps bear no relation to the problem they accompany, achieve performance largely comparable to those trained on correct traces.

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

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — mentioned by ds_opseeker

Thu, 27 Aug 2026 16:29:43 GMT

Technically it is not a reasoning machine. If it was a reasoning machine then we would not see results like this: > To systematically investigate the role of end-user semantics of derivational traces, we set up a controlled study where we train transformer models from scratch on formally verifiable reasoning traces and the solutions they lead to. We notice that, despite gains over the solution-only baseline, models trained on entirely correct traces can still produce invalid reasoning traces even when arriving at correct solutions. More interestingly, our experiments also show that models…

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

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — mentioned by nyrikki

Thu, 17 Jul 2025 14:20:52 GMT

Subbarao Kambhampati and his team has several papers strongly suggesting that the intermediate tokens are of no symantic import.[0] My hunch is that training explicitly for human acceptance of the intermediate/scratch tokens will end up being expensive. What you need is to converge to the correct answer, and scratch space dramatically improves expressivity[1], but the relationship to learning and retrieval is still open unless I missed something. Remember even access to a pure random Oracle dramatically improves compute time on some problems. [0] https://arxiv.org/abs/2505.13775 [1]…

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

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — mentioned by emp17344

Mon, 26 May 2025 18:52:55 GMT

This paper makes the claim that the “reasoning” output these models produce isn’t a necessary component of these models, and is likely unrelated to the actual final output: https://arxiv.org/abs/2505.13775 Just because these models are good at outputting text that appears to be a series of thoughts, doesn’t mean that LLMs are thinking.

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

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — HN submission

Fri, 23 May 2025 16:13:43 GMT

Beyond Semantics: Unreasonable Effectiveness of Reasonless Intermediate Tokens

138 points; 66 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — HN submission

Thu, 22 May 2025 12:05:41 GMT

Beyond Semantics: Unreasonable Effectiveness of Reasonless Intermediate Tokens

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

Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens — HN submission

Wed, 21 May 2025 15:44:14 GMT

The Unreasonable Effectiveness of Reasonless Intermediate Tokens

4 points; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.