Testing LLM reasoning abilities with SAT is not an original idea; there is a recent research that did a thorough testing with models such as GPT-4o and found that for hard enough problems, every model degrades to random guessing. But I couldn't find any research that used newer models like I used. It would be nice to see a more thorough testing done again with newer models.
I then added a few more personal preferences and suggested tools from my previous failures working with agents in Python: use uv and .venv instead of the base Python installation, use polars instead of pandas for data manipulation, only store secrets/API keys/passwords in .env while ensuring .env is in .gitignore, etc. Most of these constraints don’t tell the agent what to do, but how to do it. In general, adding a rule to my AGENTS.md whenever I encounter a fundamental behavior I don’t like has been very effective. For example, agents love using unnecessary emoji which I hate, so I added a rule:
According to deHoop, the celebration was born as a direct response to Trump’s joke on a phone call to the victorious men’s team that he would also have to invite the women to the White House or face impeachment.，推荐阅读夫子获取更多信息
По данным источников, он обеспечивал оплату этих работ, а также, как утверждается, помогал подрядчикам избегать штрафов и неустоек за нарушение сроков. Таким образом, считает следствие, создавалась схема покровительства коммерческим структурам.
，更多细节参见快连下载安装
“It is always best to err on the side of caution until you are very clear on the purpose and culture of the group,” Wesson said.
阿里千问将发布多款 AI 硬件。搜狗输入法2026对此有专业解读