名义收入高 vs 隐藏扣除多： 美国的图表中并没有包含美国打工人通常还要扣除的 401(k)（企业补充养老金，通常个人扣 5%-10% 的税前收入）和 Health Insurance Premium（每月几百美元不等的医疗保险费）。如果把这两项加上，湾区程序员的综合扣除比例会轻松突破 45% - 50%。税收倒逼的“润州（Relocation）”潮： 这也就是为什么疫情后远程办公兴起时，大量硅谷程序员搬去了德州、佛罗里达或内华达州。拿着硅谷的 Base，交着 0% 的州税。中美“封顶”机制的区别： 中国的高薪程序员（超3.4万月薪）由于五险一金封顶，在高收入段的主要负担是单纯的个税；而美国程序员不仅个税极高，加州的高额州税也是没有封顶的（赚得越多比例越高），属于真正的“劫富”。

湾区20万刀贫困线？——图解美国程序员的“税后真相”

“月薪两万”是个分水岭：此时五险一金是扣费绝对主力，个税占比不高。这时候不要心疼五险一金，尤其是其中的公积金（12%个人+12%企业），这实质上是免税的个人买房储蓄，是你实打实的隐形财富。“月薪三万五”是机制拐点：社保公积金封顶，此后的收入增长将完全暴露在个税的“枪口”下。“高薪阶层”需重视专项附加扣除：当年薪达到50万至百万级别时，高昂的累进税率会让每次涨薪的到手率骤降。这时候，足额填报“子女教育、赡养老人、住房贷款利息”等专项附加扣除，或者利用个人养老金（每年12000元额度）进行税前抵扣，才是最有效的“合法加薪”手段。

月薪过万到年薪百万：在北京，你的工资去哪儿了？——图解税前与税后收入的秘密

Woodpecker: Hallucination Correction for Multimodal Large Language Models

Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies

RECOMP: Improving Retrieval-Augmented LMs with Compression and Selective Augmentation

Review of Agentbench

Review of Learning to Retrieve In-Context Examples for Large Language Models

Review of WizardLM

LLM for Science — An Attempt to Solve Partial Differential Equations

Review of ScienceQA

Life Lessons on Two Wheels: The Power of Slowing Down

Review of Universal and Transferable Adversarial Attacks on Aligned Language Models

Review of Measuring Faithfulness in Chain-of-Thought Reasoning

Review of A survey on evaluation of large language models

A Comprehensive Review of Adversarially Robust Vision Transformer

CNS 审美配色

SCI 斜体词(latin)汇总

Determined 实用小技巧

ChatGPT Plugins 使用记录

code sample

test long article

这是示例的文章摘要，摘要内容可被用做搜索，预览。文章列表默认展示概要；若要在列表直接展示文章内容，可在blog.config.js中配置。

🧯Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies

🌀RECOMP: Improving Retrieval-Augmented LMs with Compression and Selective Augmentation

🎫Review of Agentbench

🛷Review of Learning to Retrieve In-Context Examples for Large Language Models

🛠️Review of WizardLM

🚜Review of Universal and Transferable Adversarial Attacks on Aligned Language Models

🏸Review of Measuring Faithfulness in Chain-of-Thought Reasoning

📱Review of A survey on evaluation of large language models

📺A Comprehensive Review of Adversarially Robust Vision Transformer