Last week, two of the biggest and oldest companies in the burgeoning AI drug discovery space announced a surprise merger. In a roughly 75-25 deal, Recursion Bio and Exscientia will (if the merger goes ...
How Recursion is leading a new era of AI-driven drug discovery AI drug discovery is not a new phenomenon – but it is evolving at an almost dizzying pace. Recursion is one of the earliest innovators in ...
2025年的最后一天, MIT CSAIL提交了一份具有分量的工作。当整个业界都在疯狂卷模型上下文窗口(Context ...
其核心思想是借鉴计算机科学中的“外存算法” (Out-of-core algorithms),不要把长 Prompt 直接喂给神经网络,而是把它放在一个外部的 Python REPL 环境里。 Prompt 被加载为环境中的变量context 。模型不再像以前那样生吞整个字符串,而是通过编写代码(如len (context) , context.split , re.search )来按需读取数据 ...
2026年的AI范式或许已悄然发生了剧变。在过去的一年里,人工智能领域的长文本处理能力经历了前所未有的飞速发展。从128k到1M再到10M的上下文窗口(Context Window)内卷化过程,表面上看似技术的进步,但实际上却隐藏着深层次的问题。基于Transformer注意力机制的线性扩张,正面临着物理与效果的双重边际递减,这让我们不得不重新思考AI的未来。
Unwind by hunting for words on the day’s themed list — they’re hidden horizontally, vertically, diagonally and backward. Come back daily for a new theme or to browse the recent archive. Jump in to ...