3 ms·
Figure 3.8 in the GPT-3 paper (SuperGLUE Performance) pretty clearly shows a plateau in performance, with the increase in performance from 13B to 175B parameter
by KhoomeiK 5y ago
Figure 3.8 in the GPT-3 paper (SuperGLUE Performance) pretty clearly shows a plateau in performance, with the increase in performance from 13B to 175B parameters (13.5x params, +6 score) not being significantly different from the increase between 6.7B to 13B (1.9x params, +4 score).