开启时会模糊预览图，关闭后正常显示
embedding based retrieval is back in vogue….ready for all the rediscoveries :) my bet: someone renames query augmentation next week
building agents and harnesses, prev @awscloud, phd cs @ temple
“be surgical in every file edit” what does it mean? no one knows but it gets the agent going
building agents and harnesses, prev @awscloud, phd cs @ temple
a fun behavior i'm trying to prompt into my agent builder harness...get the agent to interview me: - figure out roughly what i want, and then dive into the details (planning) - share thoughts and gets my takes/feedback (iterative edits) - help me define what a successful agent looks like (evals/test cases) the approach is the same but it's the nature/form factor of the convo that makes it more fun
building agents and harnesses, prev @awscloud, phd cs @ temple
who’s gonna be the first big player to disavow mcp without beating around the bush 👀
building agents and harnesses, prev @awscloud, phd cs @ temple
when all those PhD hours spent on optimizing tf out of representations for retrieval might come in handy 🥹 everything’s a cycle, there’s a lot of nice tricks and ideas in Information Retrieval but it’s incredibly hard to beat compute: - bigger model for embedding - multivector representations - test-time augmentation - etc the list continues and this doesn’t discount other great ideas (ex: I worked a lot on training that matched the distribution of queries and databases during train time so it would do better at test time), but good ideas + compute has long been a winning ML strategy (in IR it’s no different)
building agents and harnesses, prev @awscloud, phd cs @ temple
interested in a potential future where the thing that determined success for an Agent Task was largely the search and context preparation step that precluded it (ie. any model of sufficient intelligence would have solved the task because of the alley-oop it was thrown) makes sense that Cursor and Cognition are focusing here bc for many tasks, great search + good context organization means you don't need that last bit of frontier model reasoning juice bitter lesson again? more compute...on search
building agents and harnesses, prev @awscloud, phd cs @ temple