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
test(evals): the foreign per-example model case is a usage error too
Bruce-Yii95b24f905182
diff --git a/tests/test_evals_onnx.py b/tests/test_evals_onnx.pyindex d3b25684..3bd17e0d 100644--- a/tests/test_evals_onnx.py+++ b/tests/test_evals_onnx.py@@ -147,7 +147,9 @@ try:row["model"] = "some-other-checkpoint"f.write(json.dumps(row) + "\n")rc = evals_cli.main(["run", mismatch, "--onnx", "laya.onnx"])- check("cli/foreign per-example model exits 1", rc, 1)+ # Asking for a different checkpoint than the ONNX export serves is a caller mistake, so+ # exit 2 (docs/evals.md:28), not the 1 a quality failure uses.+ check("cli/foreign per-example model exits 2", rc, 2)finally:onnx_agent_module.ONNXAgent = _real_onnx_agent