yogthos

does google not exist at the troll farm you're working at? https://www.iqair.com/ca/world-air-quality-ranking

just wait for the bubble to pop, and I'm sure we'll see a lot of affordable GPUs flood the market from the abandoned data centres :)

Except that he does exactly that by stating that he gets his values from religion. Perhsaps you don't understand the difference between characterization and quoting?

He's literally quoting the bible at a religious event, in his post as a Canadian official, and saying that these are his guiding values. The fact that you're trying to spin this as anything else is frankly incredible.

Correct, the article published in The Star covers the specific national prayer breakfast event that Carney attended. The CBC article does not cover this event, the parts it does cover match what The Star reports. Here's the video of the event itself. Are you going to claim that the video is fabricated now?

https://www.youtube.com/watch?v=MISUFPHP-po

It doesn't claim that the PM did or say something he never said or did. Seems that you didn't actually bother reading the article in the submission. The 'less biased' version Omgpwnies linked basically says the same thing as the article I linked. Hence it's just pearl clutching.

Historically, this has been a feature and not a bug since forever wars provided a steady demand for the military industry. The difference here is that Iran controls escalation dominance, and the US can't just decide when to dial things up or down as convenient.

That's part of the idea with the whole mixture of experts (MoE) approach in newer models actually.

Rather than using a single neural net that's say 512 wide, you split it into eight channels/experts of 64. If the neural net can pick the correct channel for each inference, then you only have to run 1/8th of the neurons on every forward pass. Of course, once you have your 8 channels/experts in parallel, you now need to decide which expert/channel to use for each token you want to process. This is called a router which takes in an input and decides which expert/channel to send it to. The router itself is a tiny neural network. It is a matrix that converts the input vectors to a router choice. And the router itself has a small set of trainable weights that gets trained together with the MoE.

Oh I see we have a commenter from a genocide supporting instance. So brave.

This is where identifying the biases of the author is useful. People tend to cherry pick facts that support their existing beliefs and desires as a rule.

It's true, the numbers for the US regime are likely grossly undercounted just as they were for COVID https://gizmodo.com/updated-death-toll-shows-the-us-grossly-undercounted-covid-deaths-early-on-2000734587

I agree, the US wants to prevent a Eurasian bloc from forming at all costs. Iran and Afghanistan are a big part of the BRI which connect China with West Asia. And that's why the US has been destabilizing the region obsessively.