Skip to main content

Rope Implementation Comparison

Comparison between Ben's simplified RoPE implementation and the original one
Created on April 18|Last edited on April 18

Section 1


Select runs that logged samples/sec
to visualize data in this line chart.
Select runs that logged lm loss
to visualize data in this line chart.
Select runs that logged flops/s/gpu
to visualize data in this line chart.
Run set
0