HUGE AI NEWS : MAJOR BREAKTHROUGH!, 2x Faster Inference Than GROQ, 3 NEW GEMINI Models!

FULL OF FRAUD VIDEO

A 100 million token context window is nothing short of jaw-dropping. But, Until we get some solid details or a nod from someone who isn't on a hype train, I'll keep the confetti in the bag. If this turns out to be legit and it can be efficiently used, it's amazing, but huge claims require convincing data …

It’s somebody that has done sales for decades. This will only work on the low end and as a support mechanism. If your AI is trying to sell somebody all you’re telling people is that they’re not worth a human interacting with them.
Remember in sales, the customer only cares when he thinks you care.

Drink, every time this guy says like… I'm hammered 😂

This channel is insufferable. The titles and thumbnails are always so overly dramatic they make a mockery of the subject matter. It could be a great channel if the subject was treated seriously, but instead we get tabloid level titles that make the whole thing a waste of time. Added to the "don't recommend" list!

Is THIS music video the FIRST made with STRAWBERRY A.I. ???

Peachy da WHUUPI: 🎶 “Strawberry ! And the BUBBLES ?” On You Tube now .

Oh Boy ! Surely looks we're going to have a party. What do YOU think ?

What is going on ?

And did you know the add was going to show when you were explaining the 100m context windows or was it AI that linked it up so that when the Ad finished you mentioned being interrupted 😂😂😂😂😂😂😂 I wonder if anyone else has realised your an ai yet ❤❤❤❤❤❤❤ lol

That demo with Bland AI was highly edited, more than Carmella Harris’s CNN interview. I went to the website and the conversation was terrible, talking with ChatGpT is way way better. I do agree this is the future, but these guys are not even close with their current model.

Please, everyone, reply to this comment with ACTUALLY good AI/ML youtube resources who don't claim that the industry is being broken several times a week and that AGI is like totally like around the corner u guys. God, I'm so fckng tired of opportunistic sensational bullshit.

cerebras is fast as hell but it makes tons of error and seems much less able to iteratively cure those mistakes. coding is a good example, but there are many others. the model behind it doesnt work as well in my albeit not exhaustive testing. that said I have found many examples where it doesn't perform near as well. the key will be when all these micro innovations get combined into a single powerful model . surely companies with the biggest budgets will be including all these baby steps, or leaps, into their base models. progress is fast nonetheless.

For a guy that doesn't make any jokes, you say "genuinely" a lot

ok, 100 Million context window is insane. I've been really glad that we have claude with 200k context so far. It enabled me to generate complex apps with just a prompt

Cerabas is fake and going against H100s and it was done with a 7b param model