Skip to main content

next: sample temperature tldr

Created on September 27|Last edited on September 30






200400600800global_step100000200000300000400000
200400600800global_step0.190.1950.20.2050.210.215
200400600800global_step051015202530
200400600800global_step-1-0.500.511.5
5001k1.5kglobal_step0.60.620.640.660.68
200400600800global_step-1-0.500.511.52
200400600800global_step00.020.040.060.080.1
new set
2
with reference response logging
1
direct reference response score normalization
1
old
1
correct
1