Self-playing Adversarial Language Game Enhances LLM Reasoning — mentioned by bubblyworld

Thu, 08 Aug 2024 12:47:49 GMT

The SPAG paper is an interesting example of true reinforcement learning using language models that improves their performance on a number of hard reasoning benchmarks. https://arxiv.org/abs/2404.10642 The part that is missing from Karpathy's rant is "at scale" (the researchers only ran 3 iterations of the algorithm on small language models) and in "open domains" (I could be wrong about this but IIRC they ran their games on a small number of common english words). But adversarial language games seem promising, at least.

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

Self-playing Adversarial Language Game Enhances LLM Reasoning — HN submission

Mon, 29 Apr 2024 14:28:57 GMT

Self-Playing Adversarial Language Game Enhances LLM Reasoning

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

Self-playing Adversarial Language Game Enhances LLM Reasoning · HackerPapers discussions

Self-playing Adversarial Language Game Enhances LLM Reasoning — mentioned by bubblyworld

Self-playing Adversarial Language Game Enhances LLM Reasoning — HN submission