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(agent): name the check that fails when the count is missing
Aashishc78bac4cabe4
diff --git a/tests/test_predict_long.py b/tests/test_predict_long.pyindex a174b689..1ba54c84 100644--- a/tests/test_predict_long.py+++ b/tests/test_predict_long.py@@ -75,8 +75,8 @@ a = make_agent(lambda s, q: [])short = a.predict_long({"body": "x" * 50}, Q) # 50 tokens <= budget 72check("short/delegates to system_one", short["answers"], {"_via": "system_one"})check("short/no predict_batch call", a._calls["batch_states"], None)-check("short/one window is reported", short["usage"]["windows"], 1)-check("short/system_one's own usage is kept", short["usage"]["input_tokens"], 1)+check("short/one window is reported", short["usage"].get("windows", "<absent>"), 1)+check("short/system_one's own usage is kept", short["usage"].get("input_tokens", "<absent>"), 1)# 1b. the count is added to a copy, because a start hook that skips hands back the caller's own# payload dict and that object may be cached and reused