Total Variation Distance
A statistical metric measuring the difference between two probability distributions, defined as half the sum of the absolute differences in probabilities of the two distributions over the same sample space, with values ranging from 0 to 1.
Timeline (last 90 days)
Total variation distance (TVD) is a standard method for measuring the difference between two probability distributions, defined as half the sum of the absolute differences in probability over the same set of events for a pair of distributions, with a range from 0 to 1.
In GA-GRPO, the magnitude of bias is controlled by the guidance divergence δ_G, i.e., the total variation (TV) distance between the guidance-augmented sampling distribution and the model's own distribution.
All Facts (2)
Total variation distance (TVD) is a standard method for measuring the difference between two probability distributions, defined as half the sum of the absolute differences in probability over the same set of events for a pair of distributions, with a range from 0 to 1.
50%UnverifiedIn GA-GRPO, the magnitude of bias is controlled by the guidance divergence δ_G, i.e., the total variation (TV) distance between the guidance-augmented sampling distribution and the model's own distribution.
50%