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(fast): describe checkpoint test setup without an env var
cklxxd339aeb393af
diff --git a/docs/fast-backends.md b/docs/fast-backends.mdindex 3588cd2a..fe062bc4 100644--- a/docs/fast-backends.md+++ b/docs/fast-backends.md@@ -121,8 +121,9 @@ Before/after runs used the original module extracted withThe original was loaded as `laya.tl_kernels` via `importlib` for baseline runs;all other Laya code and the Python environment stayed the same.-- `python -m pytest tests/test_fast.py -q`, with `LAYA_TEST_MODEL` set:- **13 passed before; 13 passed after**. Initially without that variable it was+- `python -m pytest tests/test_fast.py -q`, with the full-forward tests configured+ to load the cached English checkpoint identified below:+ **13 passed before; 13 passed after**. Initially without a checkpoint it was11 passed / 2 skipped. Both full runs emitted the checkpoint's existingtemperature-clamping warning (`choice:11+=0.10058280825614929 -> 0.5`).- Seeded capture of the existing kernel tests: **24 output tensors bit-identical**,