Large Language Models as Optimizers · HackerPapers discussions

Large Language Models as Optimizers — mentioned by cyphar

Fri, 16 Oct 2026 21:14:30 GMT

> 2. Whats with all the "hmm..." in the internal reasoning process log? Do models really do that? And if so, why? Some weird emergent language thing, or just mimicry? Assuming that is actually the real reasoning trace (open weight model reasoning traces usually do things like switch languages or talk in non-standard English because of model drift, so it's a bit surprising that Claude wouldn't do the same) there was a study a few years ago that found base models produce better answers if you put "take a deep breath" in the prompt[1]. Maybe in the RL training stage stuff like that gets…

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

Large Language Models as Optimizers — mentioned by gs17

Thu, 04 Jan 2024 01:36:24 GMT

It's from this paper (albeit not applied to ChatGPT?): https://arxiv.org/abs/2309.03409

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

Large Language Models as Optimizers — mentioned by benlivengood

Sun, 17 Sep 2023 17:25:54 GMT

DeepMind has https://arxiv.org/abs/2309.03409 so as long as you can specify an objective scoring function you can get closer than manual prompt engineering.

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

Large Language Models as Optimizers — mentioned by falcor84

Fri, 15 Sep 2023 23:26:37 GMT

The original paper on arxiv: https://arxiv.org/abs/2309.03409

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

Large Language Models as Optimizers — HN submission

Fri, 08 Sep 2023 14:37:30 GMT

Large Language Models as Optimizers. +50% on Big Bench Hard

95 points; 33 discussion replies on HN. Paper and source history. A submission is not an endorsement.