|
|
|
|
|
|
by slevis
491 days ago
|
|
|
Looks like I might be the minority, but I disagree with this prediction.
Better models will also be better at abstracting and we have seen several examples (e.g. the paper LIMO: Less is More for Reasoning) that with a small amount of training data, models can outperform larger models. |
|
|