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(compile-and-fast-path): point the CPU-path citations at the lines they describe
Aashish567892a090fb
diff --git a/docs/compile-and-fast-path.md b/docs/compile-and-fast-path.mdindex 2be421f5..188171a2 100644--- a/docs/compile-and-fast-path.md+++ b/docs/compile-and-fast-path.md@@ -188,14 +188,15 @@ The C backend does accept:- `T.gemm` with a `T.alloc_local` accumulator, which lowers to a scalar triple loop.That list is now the specification of a CPU lowering, not an argument against one. Each of the five-factories takes a keyword-only `cpu=False`, and `compile_cpu` (`laya/tl_kernels.py:258-268`) sets+factories takes a keyword-only `cpu=False`, and `compile_cpu` (`laya/tl_kernels.py:261-271`) sets`cpu=True` with `dtype="float32"` and lowers the same `prim_func` on `target="c"` with`TIR_DISABLE_VECTORIZE`. The three rejected constructs are the three the selector swaps:`T.alloc_fragment` and `T.alloc_shared` become `T.alloc_local`, `T.Parallel` becomes `T.grid`, and16-bit activations become fp32. Everything else is shared with the GPU form — tile blocking,`T.Pipelined`, `T.gemm` with `GemmWarpPolicy`, `T.reduce_sum` / `T.reduce_max` — and the only-construct the CPU branch drops is the swizzled shared-memory layout annotation-(`laya/tl_kernels.py:215-216`), which is a GPU concern anyway.+construct the CPU branch drops is the swizzled shared-memory layout annotation, the `if not cpu:`+guard on `T.annotate_layout` inside `attn_kernel` (`laya/tl_kernels.py:218-219`), which is a GPU+concern anyway.What this is *not* is still what the paragraph before it argued: not a faster way to run on CPU.`compile_cpu`'s own docstring calls it "a scalar correctness path, not an Agent.accelerate backend",