Task — fault_category_line

benchmark version 1.1.0 · held_out results v1.0.0 · splits v1.0/held_out, v1.1/multi_grid
updated 2026-09-29 13:18 UTC · git 7243f310ddbc

tier 2 · fault_category · line view · 3 classes · all cells measured

What this task is. Group the fault into its coarse category (e.g. single-phase-to-ground vs multi-phase) — the level of typing many protection schemes actually act on, between binary detection and the full nine-class vocabulary. Line view: the cubicles at both ends of one line (12 channels), the information basis of a line-differential scheme.

How it's scored

Headline metric balanced_accuracy (↑ higher is better); full metric set: balanced_accuracy · macro_f1 · accuracy. Metrics are a frozen contract (src/evemtbench/evaluation/metrics.py); label derivation: src/evemtbench/tasks/labels.

Example window

example waveform window

event flt_2phg_shc · grid double_line (adapt_grid/test split) · sample 84, window #1932, cubicle 3 (max-RMS) · 50 ms / 480 samples @ 9.6 kHz · top: 3-phase current, bottom: 3-phase voltage · figure provenance in assets/manifest.json

Results (held_out)

fault_category_line — every grid × baseline; click headers to sort
taskgridbaselineheadlinetestbenchmarkparamsfit/seedtrace
fault_category_linecigre_mvmajoritybalanced_accuracy ↑0.333 ± 0.000 (n=5)0.333 ± 0.000 (n=5)—0strace: output · eval-code · log · train-code · config · W&B
fault_category_linecigre_mvrandom_forestbalanced_accuracy ↑0.595 ± 0.007 (n=5)0.643 ± 0.006 (n=5)—12strace: output · eval-code · log · train-code · config · W&B
fault_category_linecigre_mvmlpbalanced_accuracy ↑0.649 ± 0.003 (n=5)0.662 ± 0.001 (n=5)3.1M39strace: output · eval-code · log · train-code · config · W&B
fault_category_linecigre_mvgrubalanced_accuracy ↑0.656 ± 0.015 (n=5)0.670 ± 0.012 (n=5)55k1.8mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linecigre_mvcnnbalanced_accuracy ↑0.670 ± 0.003 (n=5)0.673 ± 0.015 (n=5)52k1.5mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linecigre_mvresnetbalanced_accuracy ↑0.658 ± 0.027 (n=5)0.674 ± 0.032 (n=5)509k4.7mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_linemajoritybalanced_accuracy ↑0.500 ± 0.000 (n=5)0.500 ± 0.000 (n=5)—0strace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_linerandom_forestbalanced_accuracy ↑0.827 ± 0.007 (n=5)0.683 ± 0.094 (n=5)—14strace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_linemlpbalanced_accuracy ↑0.964 ± 0.000 (n=5)0.998 ± 0.001 (n=5)3.1M1.3mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_linegrubalanced_accuracy ↑0.967 ± 0.001 (n=5)1.000 ± 0.000 (n=5)55k1.4mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_linecnnbalanced_accuracy ↑0.964 ± 0.003 (n=5)0.999 ± 0.001 (n=5)52k1.5mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linedouble_lineresnetbalanced_accuracy ↑0.965 ± 0.004 (n=5)0.999 ± 0.001 (n=5)509k5.1mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvmajoritybalanced_accuracy ↑0.500 ± 0.000 (n=5)0.500 ± 0.000 (n=5)—0strace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvrandom_forestbalanced_accuracy ↑0.824 ± 0.005 (n=5)0.628 ± 0.004 (n=5)—14strace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvmlpbalanced_accuracy ↑0.955 ± 0.000 (n=5)0.995 ± 0.000 (n=5)3.1M59strace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvgrubalanced_accuracy ↑0.963 ± 0.002 (n=5)0.998 ± 0.000 (n=5)55k1.3mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvcnnbalanced_accuracy ↑0.961 ± 0.003 (n=5)0.997 ± 0.001 (n=5)52k1.4mtrace: output · eval-code · log · train-code · config · W&B
fault_category_linetestgrid_110kvresnetbalanced_accuracy ↑0.960 ± 0.003 (n=5)0.997 ± 0.001 (n=5)509k4.0mtrace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39majoritybalanced_accuracy ↑0.500 ± 0.000 (n=5)0.500 ± 0.000 (n=5)—0strace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39random_forestbalanced_accuracy ↑0.951 ± 0.001 (n=5)0.999 ± 0.000 (n=5)—17strace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39mlpbalanced_accuracy ↑0.963 ± 0.003 (n=5)0.998 ± 0.002 (n=5)3.1M32strace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39grubalanced_accuracy ↑0.964 ± 0.001 (n=5)0.998 ± 0.001 (n=5)55k1.1mtrace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39cnnbalanced_accuracy ↑0.964 ± 0.003 (n=5)0.997 ± 0.002 (n=5)52k48strace: output · eval-code · log · train-code · config · W&B
fault_category_lineieee39resnetbalanced_accuracy ↑0.962 ± 0.003 (n=5)0.997 ± 0.002 (n=5)509k2.4mtrace: output · eval-code · log · train-code · config · W&B

Metric definitions

headline & reported metrics for this view — the metric set is a frozen contract; each links to the exact function that computes it

metricdefinitionentry
balanced_accuracymean of per-class recall — the majority class cannot buy a good scoreevaluate()
macro_f1unweighted mean of per-class F1 (zero_division=0)evaluate()
accuracyfraction of exact-match predictionsevaluate()

Reproduce one cell

PYTHONPATH=src python -m evemtbench.baselines.runner --task fault_category_line --protocol held_out \
    --grid <grid> --baseline <baseline> --seeds 0 1 2 3 4