Keep on to blur preview images; turn off to show them clearly
buried: Nemotron is not just a Mamba2 hybrid now, they've used LatentMoE. «By shifting routed expert computation and all-to-all traffic to the latent space, both per-expert weight loads and communication payloads are reduced by a factor of 𝑑/ℓ» ideologically similar to MLA
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
This is absolutely ridiculous behaviour by Norwegian immigration. Investors go through all this just to be treated as second class citizens.
🌏 RCBI advisor & offshore services for HNWI, business owners with a focus on:🇨🇭🇲🇾🇰🇳🇵🇾🇻🇺🇳🇷🇵🇦🇱🇻🇦🇪🇭🇰 | Geopolitics | Healthy lifestyle
I'm a big fan and DAU of Granola - IMO, it's the easiest AI notetaker to start using "out of the box," and it gets better with more meetings on it. ...and this is day 10 of @venturetwins and I highlighting our favorite consumer AI products of the year! Follow along for more.
Partner @a16z and twin to @venturetwins | Investor in @gammaapp, @happyrobot_ai, @krea_ai, @tomaauto, @partiful, Salient, @scribenoteinc & more
Recipes is the most underrated feature on Granola. You can automate action across one (or several) transcripts. Some of my favorite use cases: - Daily/weekly list of promised to-dos - Follow-ups customized w/ topics discussed - Ideas for content based on what you've said
I'm a big fan and DAU of Granola - IMO, it's the easiest AI notetaker to start using "out of the box," and it gets better with more meetings on it. ...and this is day 10 of @venturetwins and I highlighting our favorite consumer AI products of the year! Follow along for more.
RT @arctotherium42: I feel the need to point out tge difference between birth rate and TFR. Japanese TFR is low, but not that low. Japanese…
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
现在有一种流派玩法 前期靠运营n8n验证 验证成功后工程师vibe coding+hitp去反转n8n workflow成代码产品化 不过n8n的workflow有很都不太能直接投产使用 比如就拿youtube视频转中文来举例吧 这种东西卡点很多 ytb视频下载就是一个非常大的卡点 之前还可以无限下 现在一个cookie只能下一次 然后时间戳校对也是个卡点 这些卡点都拼在一起后，多数情况下光靠工作流想做到比较好的效果还是非常困难的 这种需求还是建议找靠谱的开源项目，比如videolingo 简单总结一下我对n8n的看法： 1、产品/运营专家能用来验证流程，想量产还是要代码化，如果会vibecoding，直接写很多时候比n8n快 2、个体户/机构可以用来唬smb，没啥使用规模，业务几近定制，n8n切这里是有很多市场空间的，一方面可以自己外包赚钱，另外一方面可以靠这个验证需求，后面再想办法产品化 3、做培训的/搞自媒体的可以用来搭玩具，做一些看起来有用的情绪价值很大的东西，很容易传播 4、不要对n8n抱有太高预期，也不应该太低估它，很适合个体户/小业务做运营验证和初步自动化，也适合搞社群卖课，不适合产品化规模化，上体量追求性能还是需要换换
Believing is seeing