← Back to context Comment by hgoel 15 hours ago Except compute 3 comments hgoel Reply ReptileMan 15 hours ago Do you think we are near the theoretical maximums of how much model we cram into 20B? hgoel 13 hours ago I don't consider myself qualified to make that call, though I believe there's plenty of room for improvement. But I don't mean inference compute, which is fairly accessible for well-off enthusiasts even now. Training is a lot more expensive though. ReptileMan 9 hours ago Back in the day we had seti@home, prime95 and fold@home. We will make it work.
ReptileMan 15 hours ago Do you think we are near the theoretical maximums of how much model we cram into 20B? hgoel 13 hours ago I don't consider myself qualified to make that call, though I believe there's plenty of room for improvement. But I don't mean inference compute, which is fairly accessible for well-off enthusiasts even now. Training is a lot more expensive though. ReptileMan 9 hours ago Back in the day we had seti@home, prime95 and fold@home. We will make it work.
hgoel 13 hours ago I don't consider myself qualified to make that call, though I believe there's plenty of room for improvement. But I don't mean inference compute, which is fairly accessible for well-off enthusiasts even now. Training is a lot more expensive though. ReptileMan 9 hours ago Back in the day we had seti@home, prime95 and fold@home. We will make it work.
ReptileMan 9 hours ago Back in the day we had seti@home, prime95 and fold@home. We will make it work.
Do you think we are near the theoretical maximums of how much model we cram into 20B?
I don't consider myself qualified to make that call, though I believe there's plenty of room for improvement. But I don't mean inference compute, which is fairly accessible for well-off enthusiasts even now. Training is a lot more expensive though.
Back in the day we had seti@home, prime95 and fold@home. We will make it work.