EsoLang-Bench: Evaluating Genuine Reasoning in Large Language Models via Esoteric Programming Languages
arXiv:2603.09678v1 Announce Type: new Abstract: Large language models achieve near-ceiling performance on code generation benchmarks, yet these results increasingly reflect memorization rather than genuine reasoning. …
Aman Sharma, Paras Chopra
조회수 44회