• Essay: “Securing Model Weights” considered pointless | Lessons learned from (failed) protection of the Unix source code, et al…

  2025/01/30

• GitHub: “Try, compare, and implement this model in your code for free…” | great unwashed hordes of common users are going to be using AI models as interchangable consumables, like lightbulbs…

  2025/01/30

• Run a 670 Billion parameter LLM at home for $6000 with no GPU required

  2025/01/28

• What DeepSeek r1 Means—and What It Doesn’t | Lawfare | Yes, but…

  2025/01/28

• Playing Around With Deepseek (NSFW) | www.bentasker.co.uk

  2025/01/27

• DeepSeek vs: Winnie-The-Pooh: the winner is Chinese Censorship — unless you know how to circumvent it

  2025/01/24

• Dan Brickley on Bluesky: “Testing DeepSeek R1 censorship UI (mobile web), since it will get heavy use worldwide” | …thread with multiple screenshots

  2025/01/24

• “deepseek r1 has an existential crisis” | …more fuel for the upcoming “Open Source LLMs on Raspberry Pi’s are a National Security issue” crisis

  2025/01/24

• If what @BrianRoemmele claims is delivered, running a Chinese LLM on a Raspberry Pi at home should become trivial

  2025/01/23

• Why AI Works

  2025/01/13

• Media is flooded with complaints about AIs being trained on images or text & roughly reproducing them; meanwhile in the (human) music industry for the past 100 years…

  2024/12/24

• Sometimes I imagine that there is an alien race watching and discussing us, and they’re saying “…these humans aren’t *really* intelligent, it’s all just pattern matching and stochastic parroting”

  2024/12/04