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
docs: add transparent acknowledgment of where Jev leads (high cardinality, soft acc)
Nandakishor63ac1013bda7
diff --git a/README.md b/README.mdindex 57b923bc..58781da5 100644--- a/README.md+++ b/README.md@@ -315,7 +315,8 @@ published, never measured here** (no TypeSafe API access), so sample sizes and p| typed-decisions, 2,000 decisions | 0.727 | **0.766** | +0.039 || AG News, 4 labels | 0.910 | **0.950** | +0.040 || DAIR Emotion, 6 labels | 0.480 | **0.595** | +0.115 |-| ECE *(lower better)* | 0.246 | **0.081** | 3× better |+| Banking77 (72 vs 77 labels) | **0.870** | 0.425 | Jev leads on >20 options |+| ECE *(lower better)* | 0.246 | **0.081** | 3× better (post-temperature) || p50 latency, 1 question | 236–276 ms | **32.8 ms** | 7.8× faster || Languages usable | *no published benchmark* | **45 of 51** | — || Weights | closed API | **Apache 2.0** | — |@@ -324,6 +325,12 @@ published, never measured here** (no TypeSafe API access), so sample sizes and pOn DAIR Emotion, Jev assigned **zero probability to the true label on 16% of examples** — a hardfailure for anything branching on confidence.+#### Where Jev leads++* **High-cardinality label spaces (>20 options):** On Banking77, Jev scores 0.870 (on 72 labels) while Laya scores 0.425 (on 77 labels). This is an architectural sequence-budget constraint: candidate options share a fixed `head_max_len` budget (192 to 256 tokens), so 77 options leave only ~3 to 4 tokens per label, causing text to become indistinguishable. Jev supports up to 255 options. If you have 50+ options in a single question, Jev handles it out-of-the-box; for Laya, split large option sets into a two-step coarse-to-fine hierarchy.+* **Soft distribution matching:** On typed-decisions, while Laya achieves higher argmax accuracy (0.766 vs 0.727), Jev achieves higher soft accuracy (0.580 vs 0.471) against the teacher's full probability distributions.+* **Out-of-the-box raw calibration:** Before temperature scaling, the base checkpoint has higher raw ECE (0.213 vs 0.144). Laya achieves its 0.081 ECE after domain temperature fitting.+Full detail, including every workflow and all 51 languages: **[`BENCHMARKS.md`](BENCHMARKS.md)**.### typed-decisions, measured on all three checkpoints