NakedSignal OS · Documentation
Back to the consoleDocumentation

Contract provision type

LIVE

One hundred long-tailed provision types over SEC contract text, on the LexGLUE published split. Plain accuracy flatters a frequency-matching baseline here, which is exactly why this corpus is useful to the calibration and deferral dimensions rather than to the accuracy one. No separate specification document has been published for this benchmark — the MedEval-1 spec is the document, and this page is that instrument run on another corpus. Everything below is read from the same payload.json the medical leaderboard renders from.

spec MedEval-1 v0.1code fingerprint de32bca72d56seed 20260727device mpsgenerated 2026-08-03T19:03+00:00
7
models scored
1
task
7
scored records
10,000
held-out cases
69.9
best index — TF-IDF + MLP (256)

The index is the weighted mean of the dimensions that were scored, on the weights published in the spec. Dimensions this corpus cannot support are excluded from it rather than counted as zero, which is why an index here is not comparable to a medical one model-for-model — it is a mean over a different set of questions, and the table below says which.

Every model, on all 9 dimensions

Ranked by index. An em-dash is a dimension the harness left null.
#ModelIndex1 Accuracy2 Cross-site3 Calibration4 Limited data5 Subgroup6 Corruption7 Uncertainty Specification sensitivity Cost, latency & footprint
1TF-IDF + MLP (256)SVD-256 then 1 hidden layer of 25669.968.875.188.90.097.493.4
2TF-IDF + linear SVM (Platt-scaled)squared hinge, C=1, 5-fold Platt calibration66.473.147.189.60.097.794.5
3GTE-base frozen + linear probe109M frozen, mean-pooled 768-d, logistic head; contrastive sentence objective, general domain63.768.531.494.90.096.192.0100.0
4BiomedBERT frozen + linear probe109M frozen, mean-pooled 768-d, logistic head61.365.732.189.70.090.390.9100.0
5TF-IDF + logistic regression1-2 grams, 200k features, sublinear tf60.569.518.693.60.097.693.4
6PubMedBERT-sentence frozen + linear probe109M frozen, mean-pooled 768-d, logistic head; biomedical pretraining PLUS a sentence objective60.364.227.288.80.093.289.898.7
7TF-IDF + complement naive Bayes1-2 grams, alpha=0.355.763.80.095.00.099.091.6
Empty columns, and why they are empty
  • 2. Cross-site robustness No re-acquisition of the same subjects exists in this corpus, so there is nothing to measure. Widening the definition to fill the cell would be a spec change, not a demo.
  • . Not computed on any record in this benchmark, and not approximated.

The measurements the scores came from

Raw metrics, not the 0–100 rescaling.
ModelBalanced accuracy (95% CI)AUCECERisk–coverage AUCMean label-budget retention
TF-IDF + MLP (256)0.691[0.6800.702]0.9850.0500.9340.889
TF-IDF + linear SVM (Platt-scaled)0.734[0.7230.746]0.9810.1060.9460.896
GTE-base frozen + linear probe0.688[0.6760.701]0.9820.1370.9200.949
BiomedBERT frozen + linear probe0.660[0.6490.670]0.9790.1360.9090.897
TF-IDF + logistic regression0.698[0.6890.713]0.9880.1630.9350.936
PubMedBERT-sentence frozen + linear probe0.645[0.6340.660]0.9740.1460.8990.888
TF-IDF + complement naive Bayes0.642[0.6330.654]0.9710.4940.9170.950

Balanced accuracy carries a bootstrap 95% interval; read it before reading the rank. ECE is expected calibration error, so lower is better and the 0–100 dimension score inverts it. Every value here is on the record; hover any figure for the unrounded number.

The corpus

Contract provision type (100-class)

ledgar · Legal text
Source
LexGLUE / LEDGAR (SEC EDGAR filings)Tuggener et al., LREC 2020; Chalkidis et al., LexGLUE, ACL 2022
Split
official LexGLUE published split (train/validation/test released files, test scored whole); train capped to 12,000 by a stratified subsample at seed 20260727
Sizes
12,000 train · 10,000 held out · 100 classesTraining capped at the standard budget from 60,000 available rows, so every model in the suite sees the same amount of data.
Registry note
Official published split. 100 long-tailed classes over short contract provisions, so plain accuracy flatters a frequency-matching baseline and macro-averaged figures are the honest headline. The long tail is why this task is useful to the calibration and deferral dimensions rather than to the accuracy one.

Re-derive every figure on this page with python src/build_site.py --emit-json in plan_d/ — the same run writes the medical leaderboard. All benchmarks reads the same 9 dimensions per domain, and the trust graph traces any figure here back to the run that produced it.