Bad Case Report

This page summarizes the output difference between analysis_model and baseline_model after calling auto_find_bad_case.

Statistics for all 20 case and Top3 bad case statistics are now grouped by output.

output_0 summary

Metric All Average All Worst Top3 Average Top3 Worst
L1 0.058722 0.178026 0.151333 0.178026
COSINE 0.932461 0.791521 0.832904 0.791521
ATOL 1.227576 1.972829 1.952205 1.972829

output_0 details

All cases error distribution

L1
99th percentile 99.9th percentile 99.99th percentile
0.177755 0.177999 0.178023
COSINE
99th percentile 99.9th percentile 99.99th percentile
0.999677 0.999857 0.999875
ATOL
99th percentile 99.9th percentile 99.99th percentile
1.967733 1.97232 1.972778

output_1 summary

Metric All Average All Worst Top3 Average Top3 Worst
L1 0.003957 0.004128 0.004103 0.004128
COSINE 0.999993 0.999993 0.999993 0.999993
ATOL 0.013574 0.014755 0.014676 0.014755

output_1 details

All cases error distribution

L1
99th percentile 99.9th percentile 99.99th percentile
0.004125 0.004128 0.004128
COSINE
99th percentile 99.9th percentile 99.99th percentile
0.999994 0.999994 0.999994
ATOL
99th percentile 99.9th percentile 99.99th percentile
0.014736 0.014753 0.014754

Bad case tables

The bad case input index of each output

Name/Index L1 COSINE ATOL
0 [11, 8, 13] [8, 11, 1] [1, 15, 4]
1 [14, 18, 1] [18, 14, 5] [11, 12, 3]

The metric results of each badcase

Name/Index L1 COSINE ATOL
0 [ 0.178026, 0.176601, 0.099371 ] [ 0.791521, 0.810388, 0.896802 ] [ 1.972829, 1.946006, 1.937779 ]
1 [ 0.004128, 0.004112, 0.004067 ] [ 0.999993, 0.999993, 0.999993 ] [ 0.014755, 0.014656, 0.014618 ]

The bad case input index of the worst output

Name/Index L1 COSINE ATOL
0 11 8 1
1 14 18 11