Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
Commits on main
hooks: calibrate_reliability accepts NumPy booleans and reals after #1314
e7249b7 Nandakishor committed 27 days ago
fix(ts): count the same English function words as Python
3a86508 Rishit Sharma committed 28 days ago
fix(lang): count question words and auxiliaries as English evidence
047f186 Rishit Sharma committed 28 days ago
ci: re-run (previous run hit a flaky .NET test and a Windows runner crash; no code change)
4fb0065 Adeshina Lasisi committed 28 days ago
laya-dotnet: stop ConcurrentLoadsWarnOnce counting other classes' warnings
c038faf Adeshina Lasisi committed 29 days ago
ci: re-run the Windows job (runner crashed with 0xC000001D in tests/test_runtime_fixes.py; this PR changes no Python)
46e0557 Adeshina Lasisi committed 29 days ago
laya-go: hookrun refuses a value that is not equal to itself; floor 3868 -> 3869
b5ae168 Adeshina Lasisi committed 29 days ago
laya-go: router hook test, blank func field so staticcheck stays clean
d261274 Adeshina Lasisi committed 29 days ago
laya-go: export the hook dispatcher from agent so the router runs on_route through it
b1afb54 Adeshina Lasisi committed 29 days ago