48CVs with text
42Reviewed
42Corrected labels
48AI labels

Start Experiment

This trains a lightweight token classifier from matched label spans. It is the first baseline before LayoutLM-style layout training.

Latest Result

StatusCompleted
Samples48 total · 38 train · 10 test
Entities1569
Token F10.3705
Span F10.1728
Artifacts/app/ml_artifacts/0005-production-error-analysis-weak-ner

Experiment History

Experiment Status Data Metrics Started
production-error-analysis-weak-ner
token_logistic_regression · corrected_then_ai
Completed 48 samples · 1569 spans Token F1 0.3705 · Span F1 0.1728 2026-05-12 10:23
error-analysis-weak-ner-baseline
token_logistic_regression · corrected_then_ai
Completed 48 samples · 1569 spans Token F1 0.3705 · Span F1 0.1728 2026-05-12 10:22
weak-ner-baseline
token_logistic_regression · corrected_then_ai
Completed 48 samples · 1569 spans Token F1 0.3705 · Span F1 0.1728 2026-05-12 10:16
production-weak-ner-baseline
token_logistic_regression · corrected_then_ai
Completed 48 samples · 1569 spans Token F1 0.3705 · Span F1 0.1728 2026-05-12 10:13
first-weak-ner-baseline
token_logistic_regression · corrected_then_ai
Completed 48 samples · 1569 spans Token F1 0.3705 · Span F1 0.1728 2026-05-12 10:11