I love how this line of thinking completely avoids the issue re. improvements in local models.
I suppose if you are desperate to justify a large investment this what you would do - frame the story in a particular way.
I suppose if you are desperate to justify a large investment this what you would do - frame the story in a particular way.
Once computer constraints ease up, you will see much larger models. The reason LLM seems to have stalled a bit is because there just not enough compute.
You have more people using AI which requires more compute, and you want to build larger models which requires more compute and you have limited compute. What do you do?
" The reason LLM seems to have stalled a bit is because there just not enough compute."
lol okay mate.
and yet now we have far bigger rooms with far bigger computers anyway
Hardware may improve exponentially, but demand for compute increases double-exponentially. we'll always need more, bigger computers
There is literally not enough compute dude. If we had infinite compute we would see everyone building Mythos size models.
"Right.. and computers were once the size of a large room vs now fit into a pocket."
Yeah because we offload much of the compute to off device. You think we would have these small computers in our pockets if the internet didn't exist?