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
feat(common): report options that collapse into the same token span
ac1bc78 AlKor13 committed 1 month ago
test(mcp): cover the per-call controls at the tool layer and over the wire
be4934d Aashish committed 1 month ago
feat(mcp): expose task/lang/max_len/head_max_len on the decision tools
d38aab2 Aashish committed 1 month ago
fix(mcp): refuse a non-string noul label before stringifying it
8d1db1f Bruce-Yii committed 1 month ago
fix(serve): forward and validate per-request token budget overrides (#549)
fba55b3 Swatantra Yadav committed 1 month ago
fix(mcp): reject a malformed question as a caller error, not a server fault
0b1ccf7 Bruce-Yii committed 1 month ago
fix(mcp): laya_preset answered about a field the state did not have
bbbefe0 Aashish committed 1 month ago
feat(presets): read the state field a question set uses out of its own instructions
6c8817a Aashish committed 1 month ago
ci(ts): add automated npm release workflow with provenance (#288)
92240a0 Swatantra Yadav committed 1 month ago
test(local-e2e): assert the noul label override where the checkpoint reads it
8e276af thread3d committed 1 month ago
fix(research): let the bench harnesses take their model root from LAYA_MODELS
c7958db thread3d committed 1 month ago
fix(common): keep the decision head's attention shapes when traced
a75eb0a thread3d committed 1 month ago
Fix eval-set: add per-item id tag for traceability; correct --slice flag in example
0cc34cf CodyQin committed 1 month ago
Add zh-CN/zh-TW evaluation: same-methodology rerun, business-scenario eval set, temperature refit
2cefb77 CodyQin committed 1 month ago
fix(agent): the over-budget message named the one knob that makes it worse
5c543b1 PerryLink committed 1 month ago
fix(ts): a score legend echoed the caller's own types, unlike the Python backends
32a3d41 PerryLink committed 1 month ago
fix(lang): a repeated English word that collides with a foreign function word routed English to multilingual
12be454 AlKor13 committed 1 month ago
fix(examples): key the answer cache on the full question definitions
e754ac1 thread3d committed 1 month ago
fix(examples): show the temperature the agent actually applies
d8a083c thread3d committed 1 month ago
fix(examples): gate on answer_confidence, not entropy confidence
05b49df thread3d committed 1 month ago
refactor(verify): drop setup_laya.sh and move the smoke test under verify/
582d9fa thread3d committed 1 month ago
feat(verify): fetch checkpoints at the manifest's pinned revision
83a1f35 thread3d committed 1 month ago
chore(setup): add setup_laya.sh and ignore the artifacts it writes
eaee58a thread3d committed 1 month ago