Amnesic probing results for the masked representations. Properties statistics, word-prediction accuracy and DKL results for the different properties inspected in this work. We report the vanilla word prediction accuracy and the Amnesic scores, as well as the Rand and 1-Hot controls which shows minimal information loss and high selectivity (except for the dep property which all information was removed). The DKL is also reported for all properties in the last rows which show similar trends as the accuracy performance.
. | . | dep . | f-pos . | c-pos . | ner . | phrase start . | phrase end . |
---|---|---|---|---|---|---|---|
Properties | N. dir | 820 | 675 | 240 | 95 | 35 | 52 |
N. classes | 41 | 45 | 12 | 19 | 2 | 2 | |
Majority | 11.44 | 13.22 | 31.76 | 86.09 | 59.25 | 58.51 | |
Probing | Vanilla | 71.19 | 78.32 | 84.40 | 90.68 | 85.53 | 83.21 |
LM-Acc | Vanilla | 56.98 | 56.98 | 56.98 | 57.71 | 57.71 | 57.71 |
Rand | 4.67 | 24.69 | 54.55 | 56.88 | 57.46 | 57.27 | |
Selectivity | 20.46 | 59.51 | 66.49 | 60.35 | 60.97 | 60.80 | |
Amnesic | 4.67 | 6.01 | 33.28 | 48.39 | 56.89 | 56.19 | |
LM-DKL | Rand | 7.77 | 6.10 | 0.45 | 0.10 | 0.02 | 0.04 |
Amnesic | 7.77 | 7.26 | 3.36 | 1.39 | 0.06 | 0.13 |
. | . | dep . | f-pos . | c-pos . | ner . | phrase start . | phrase end . |
---|---|---|---|---|---|---|---|
Properties | N. dir | 820 | 675 | 240 | 95 | 35 | 52 |
N. classes | 41 | 45 | 12 | 19 | 2 | 2 | |
Majority | 11.44 | 13.22 | 31.76 | 86.09 | 59.25 | 58.51 | |
Probing | Vanilla | 71.19 | 78.32 | 84.40 | 90.68 | 85.53 | 83.21 |
LM-Acc | Vanilla | 56.98 | 56.98 | 56.98 | 57.71 | 57.71 | 57.71 |
Rand | 4.67 | 24.69 | 54.55 | 56.88 | 57.46 | 57.27 | |
Selectivity | 20.46 | 59.51 | 66.49 | 60.35 | 60.97 | 60.80 | |
Amnesic | 4.67 | 6.01 | 33.28 | 48.39 | 56.89 | 56.19 | |
LM-DKL | Rand | 7.77 | 6.10 | 0.45 | 0.10 | 0.02 | 0.04 |
Amnesic | 7.77 | 7.26 | 3.36 | 1.39 | 0.06 | 0.13 |