@Jefflau
Pinned post
I was asked a few times this past week why I bought a super expensive machine for local inference. Why not rent?View quoted post on X
One side of the answer is that I want to automate some things for Dark Bio, like accounting and invoicing. I refuse to send customer data to 3rd party providers. So that's a practical necessity I can more easily rationalize.
The other side, however, is a hedge against LLM regulations. Restricting access to model weights themselves is impossible; software piracy existed forever and one more law against file sharing won't make a dent in curbing that. That said, restricting access to hardware is a *very* effective way to regulate things. These machines are not cheap things that fly under the radar; they trigger red flags across all payment and shipping flows, so one single law passed and all flows stop instantaneously. And already today, you need to fill out forms to be allowed to buy a DGX Spark.
My concern is that with LLMs advancing and various hacks happening, some regulator will make the unfortunate knee-jerk reaction to block hardware sales. And even if it takes weeks or months to pass, it might be too late to obtain the required local inference capability with the current 3-4 month entry queues.
Is this paranoia? Probably. Is this a waste of money? Possibly. But potentially having access cut off to both hosted frontier models and local inference overnight is a pretty nasty setback with no meaningful way to paddle out of.
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 🪄
To enter:
🟢 Share this post
🟢 Comment #RTXON
Terms and Conditions: nvidia.com/en-us/geforce/…View quoted post on X
GLM 5.3 Flash on 2x RTX PRO 6000sView quoted post on X
Yesterday, an anonymous black hat group reached out with an insane RL request.View quoted post on X
They offered an astronomical amount of money to post-train open-weight models on exploit data.
The amount of attack data they claim to hold is unlike anything I have ever seen. If someone told me this was an undercover FBI dataset, I would have believed it.
For context, a white-hat project I worked on using GLM-5.3 used less than one percent of that volume, and it already outperformed frontier models by a huge margin.
Of course, I turned it down. But what genuinely scares me is knowing someone else will eventually take the deal.
