Did you try using the the talaas chat? Something stupid like 18k tokens/second.

Think it's called Askjimmy or similar.

What is the model they're using there though? Interrogated, it claims it's a BERT variant and has capabilities around GPT-3 and below GPT-4.

(Not that I believe it, it writes too well for GPT-3.)

Hosted frontier models from two years ago would be much faster today, too.

They run Llama 3.1 8B.

Oh boy, thanks for sharing this, truly mind blowing. It was chatjimmy.ai

The big benefit is ROM cells require fewer components than DRAM. So the chips would be tiny, dense, cheap and consume far less power.

It's not even just that. If you just built the rom chips separately and swapped them for the RAM of a normal accelerator, it would not help at all.

The trick is that every compute element in their system has it's own small pool of ROM, instead of putting all the ram behind a common pipe. ROM is just used because it's the densest kind of memory that can be fabricated on the same process as their logic.

I thought DRAM was pretty dense already. Is mask ROM that much denser?

Yes, each rom bit can be a transistor or even a diode with a decoder circuit. Simplest Dram cell is capacitor+transistor - and you need a clock, refresh circuit etc.

Someday, I imagine model weights could even be encoded as analog resistors (memristors or similar) for even greater density

Hm. I wonder how many relays I'd need to make a physical MNIST classifier. That'd be dope