"How far back in time can you understand English?", a post that tells a story starting with the English of 2000 AD and ending with the English of 1000 AD has gone viral, and gotten a lot of people interested in older forms of English.
Opens in a new window
，这一点在雷电模拟器官方版本下载中也有详细论述
Гангстер одним ударом расправился с туристом в Таиланде и попал на видео18:08
Even though my dataset is very small, I think it's sufficient to conclude that LLMs can't consistently reason. Also their reasoning performance gets worse as the SAT instance grows, which may be due to the context window becoming too large as the model reasoning progresses, and it gets harder to remember original clauses at the top of the context. A friend of mine made an observation that how complex SAT instances are similar to working with many rules in large codebases. As we add more rules, it gets more and more likely for LLMs to forget some of them, which can be insidious. Of course that doesn't mean LLMs are useless. They can be definitely useful without being able to reason, but due to lack of reasoning, we can't just write down the rules and expect that LLMs will always follow them. For critical requirements there needs to be some other process in place to ensure that these are met.
有趣的是，报告通过构建「2028 年宏观假想模型」，详细拆解了这一死循环的传导路径。