Keep on to blur preview images; turn off to show them clearly
I wonder if geopolitics + long-term competitiveness is enough to skew US research programs into less efficient but also less imitable methods. RL pivot was naturally incentivized. Hiding real thoughts behind gibberish Gemini summary is ≈cost-free. Can frontier labs do WORSE?
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
so latent space is ramping up a teeny bit haha
achieve ambition with intentionality, intensity, & integrity - @dxtipshq - @sveltesociety - @aidotengineer - @latentspacepod - @cognition + @smol_ai
Bruh. Jensen, it's one thing to lobby for selling a strategic resource to the geopolitical rival, understandable, it's a trillion-scale market. But betraying your fiduciary duty before shareholders? *That* is a cardinal sin in Freedom Land.
I wonder if geopolitics + long-term competitiveness is enough to skew US research programs into less efficient but also less imitable methods. RL pivot was naturally incentivized. Hiding real thoughts behind gibberish Gemini summary is ≈cost-free. Can frontier labs do WORSE?
every Wednesday is date night 🔥 waymo + hot pot live a little
building @YumeBank - chime backed by stablecoins / prev @magiceden @snapchat @square / built @bruh_bears @web3zer0 / react native OSS
当我真的开始做营销工具时我就发现 优秀的人会依靠信息进行碰撞与优质加工 但同时也滋生了无限自动改写 万物有阳就有阴 好的内容营销可以带来流量增量 差的内容营销会塞满网络垃圾 信息爆炸是一定的 所以我也在想 既然我无法阻止生产垃圾 那么我可能也需要构建一个优质过滤器以进行对抗 这样我就能对内容做评估策略 让agent针对不同人群生产相对优质的信息 尽量保证信息的有效性和价值传递
Believing is seeing
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