You don't have to use the most recent bleeding edge model to succeed. A local FOSS coding agent coupled with a reasonably priced LLM could yield the optimal ROI.
As an Israeli, this is an inexcusable crime by IDF soldiers. Appallingly, I expect them to receive no punishment. My country's government is criminally racist.
It's hard to claim ""bad apples"" when the top brass acquits soldiers who get caught on camera sexually abusing a prisoner and instead prosecutes the whistleblower for leaking evidence of the crime.
I mean, almost entire society participates in (what they believe is the "most moral") military service. It is scary to think folks with murderous tendencies might not be minority.
But at what stage are we asking for that RAM? if it's the inference stage then doesn't that belong to the GPU<>Memory which has nothing to do with the CPU?
I did see they have the unified CPU/GPU memory which may reduce the cost of host/kernel transactions especially now that we're probably lifting more and more memory with longer context tasks.
What the heck is agentic inference and how is it supposed to be different from LLM inference? That's a rhetorical question. Screw marketing and screw hype.
No, we just outsourced car maintenance to professional shop services. Both because mechanical aspects have become reliable enough to last a year without maintenance and because electronic/computer aspects are mind-bogglingly complicated.
> because electronic/computer aspects are mind-bogglingly complicated
And because it's software, it happens to be a perfect way for the manufacturer to extract rent (er, "recurring revenue") from car repair business. It's not complexity that's shaping how end-user repair experience looks like, but the fact that you often need proprietary connector, proprietary software, and a valid license key to interface with the car's computer.
And because plenty of engineering goes into designing subsystems with the explicit but unstated purpose of making them close to impossible to repair without ultimately resorting to help from the manufacturer.
Software is just the latest layer on the cake. Non-repairable designs, special tools, unavailable parts, unavailable instructions, fragile and error prone procedures, encryption, and more. They're all occasionally used to with the main purpose of blocking any attempt to easily repair without generating revenue for the manufacturer and their network.
Source: I have family working for two large car manufacturers both in engineering and management, who have personally experienced explicit demands to make things hard to repair by the owner but make them in a way where a reasonable explanation can be used for plausible deniability.
Lost me right about in the middle when he started chirping AI AI AI like a parrot. AI and trust do not go hand in hand. Focus on privacy, transparency and simplicity because instead. Good luck.
That's just irresponsible advice. There is so little actual evidence of this technology being able to produce high quality maintainable code that asking us to trust it blindly is borderline snake-oil peddling.
reply