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
Router: the vega lane lives beside the checkpoint registry, not in it
2fda478 Nandakishor committed 14 days ago
README and docs: how to switch the model underneath, by Hub name, in comments on the Agent line
5a6dc4b Nandakishor committed 14 days ago
fit: a CSV in, a calibration measured on rows the heads never saw, a floor on rows
9d3fcc8 Nandakishor committed 14 days ago
README and docs: the input limits as comments on the lines that take them
5024e2f Nandakishor committed 14 days ago
README and docs: image, audio and video limits as one paragraph, not a table
cc472f4 Nandakishor committed 14 days ago
README and docs: what each input accepts and where its limit is
2422243 Nandakishor committed 14 days ago
load(): declare the VegaAgent forward reference for the type checker
aec33ac Nandakishor committed 14 days ago
fix(ts): cut stateText budgets at code points like Python
b87976e Rahul Ranjan committed 16 days ago
fix(confidence): normalize finite weights when their total overflows
b9bda0f Jin Long committed 17 days ago
fix(ts): forward batch budgets through the single-call fallback
75033a4 xinpingzh committed 17 days ago
chore(ci): correct the pypi-publish pin comment to v1.14.3
0565524 Nandakishor committed 18 days ago