· 方法论
涉及标的： $NVDA
嗯，这种方法需要花费数百万美元来构建定制化的金融数据管道（Financial Data Pipelines）接入到大语言模型（LLM）中（我不想花这笔钱）。 这大概是对 @demishassabis、@sama、@elonmusk、@DarioAmodei、@ylecun、@arthurmensch、@aidangomez 或者拥有资源的初创公司来说的一个挑战。 这绝对是一个价值十亿美元的问题：如何让散户也能通过提示词（Prompt）获得与 Jane Street 算法同等的阿尔法（Alpha）生成能力。 我可以开源一个轻量级版本，就像上面的 LLM 委员会那样，仅基于我个人的领域特定判断，但肯定无法做到深入。有趣的事实是，$NVDA 过去曾试图招募我领导他们的人工智能团队之一，所以也许如果我现在没忙于其他事情的话……
Uh this approach would cost millions into building custom financial data pipelines into the LLM (which I don't want to spend). Prob a challenge for @demishassabis, @sama, @elonmusk, @DarioAmodei, @ylecun, @arthurmensch, @aidangomez or a startup with the resources. Definitely a billion dollar question for alpha generation to make the same Jane Street algorithm capability available to retail to prompt. I could open source a lightweight version like the LLM council above just based on my own domain-specific judgment, but won't be anywhere close to being in-depth. Fun fact, $NVDA tried recruiting me to lead one of their AI teams in the past so mybe if i wasn't working on other things right now