I signed up for Koodo for a few months. It went like this: I signed up for 5g and the first month was ~200mbs with very good signal. After a month it dropped to practically nil, terrible signal.

I waited a bit and noticed they now offered a plan with “full 5g speed” which is what I should have had. So I cancelled. That was a bit of a pain in the ass. A month later I got another bill for a whole other month. Not only that, they said payment was late a whole month. I was so pissed, I was about to lose it. I paid to avoid too much immediate hassle. Thankfully two days later they sent a cheque back for that amount.

Needless to say. Not going back.

They don’t see either as people

Or in American dollars, about three fiddy

Drunk scientist man-pickle

“Economics” aren’t political?

It was a joke. He famously does his own stunts.

Like Tom Cruise and his stunt double?

The 80s: clear your throat in too high of a pitch? Get followed to the bathroom and the shit kicked out of you.

Just the highly centralised power structure and the single party consisting entirely of nepotism.

This one was good.

https://open.spotify.com/episode/1xj51Tr4n4lPRvDoxeg8aV

I’m pretty sure it’s the follow-up though

Some downtown big cities had the buildings interconnected.

Oh, that part is. But the splitting tech is built into llama.cpp

With modern methods sometimes running a larger model split between GPU/CPU can be fast enough. Here’s an example https://dev.to/maximsaplin/llamacpp-cpu-vs-gpu-shared-vram-and-inference-speed-3jpl

fp8 would probably be fine, though the method used to make the quant would greatly influence that.

I don’t know exactly how Ollama works but a more ideal model I would think would be one of these quants

https://huggingface.co/bartowski/Qwen2.5-Coder-1.5B-Instruct-GGUF

A GGUF model would also allow some overflow into system ram if ollama has that capability like some other inference backends.

The technology for quantisation has improved a lot this past year making very small quants viable for some uses. I think the general consensus is that an 8bit quant will be nearly identical to a full model. Though a 6bit quant can feel so close that you may not even notice any loss of quality.

Going smaller than that is where the real trade off occurs. 2-3 bit quants of much larger models can absolutely surprise you, though they will probably be inconsistent.

So it comes down to the task you’re trying to accomplish. If it’s programming related, 6bit and up for consistency with whatever the largest coding model you can fit. If it’s creative writing or something a much lower quant with a larger model is the way to go in my opinion.

Another odd Canadian one. It has been codified that a suspect saying the words “I’m sorry” cannot be used as proof of guilt. Since in Canada especially, it leans a bit more into meaning “pardon” or “excuse me” rather than how an American might interpret it more as an apology.

The name of the accused can’t usually be reported on in Canada. Though there seems to be many exceptions. Also, released offenders get a lot of protection. It’s pretty controversial, especially when it’s someone famous like this case.

Yikes

Must be like that. Watched a televised rally with a guy and after a few minutes he looks at me and says “So, what do you think?” And I’m like “what even was that incoherent nonsense?”. Trump said three goddamn conflicting things in one sentence. Complete horshit. I couldn’t fucking believe this guy.

Same for Canada. Libs are red, cons blue.