Observed arrival · 2026-09-28
Leafparity: checking whether ONNX conversions agree
A tool for checking whether XGBoost, LightGBM, or scikit-learn tree models give the same answers as their ONNX conversions.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams converting tree models to ONNX
- Worth noticing
- The displayed LightGBM case reports four problems across 699 tree locations, including a mismatch caused by zero_as_missing.
Field notes
The page distinguishes three outcomes: equivalent (exit code 0), a mismatch with witness inputs (1), and an unsupported analysis that stops rather than guesses (2). Its sample LightGBM/ONNX report attributes a zero-input discrepancy to zero_as_missing: the original treats 0 as missing, while the converted runtime evaluates it numerically. The page also says its equivalence result allows a stated floating-point rounding tolerance.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue