4 ms·This seems like it should be attributed to better post training, not a bigger model.by zaptrem 2y agoThis seems like it should be attributed to better post training, not a bigger model.