|
|
|
|
|
|
by zbshqoa
498 days ago
|
|
|
It's 80/20, but in some tasks it's much better (e.g. translation) Nonetheless the fact that you can just change a bit the prompt to instruct the model to do what you want makes everything much faster. Yes the trade-off is that you need GPUs to make it run, but that's why we have cloud |
|
|