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
ci(windows): run the shared CI and release test suites
Rishit Sharma3fa307cee3aa
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.ymlindex dbdc1bc8..ed0b734c 100644--- a/.github/workflows/ci.yml+++ b/.github/workflows/ci.yml@@ -232,67 +232,22 @@ jobs:run: python -c "import laya; print(laya.__version__); print(sorted(laya.__all__))"shell: bash- # Same suites as the Linux job. Guarded with -f so a suite added by another open branch- # cannot make this job red on a tree that does not have it yet.- - name: Routing and language-detection tests- run: |- for t in test_router test_criteria test_batch test_predict_long \- test_structured test_structured_api test_onnx_lang_parity \- test_download test_verify_checkpoints test_shortlist test_decision_model \- test_head_checkpointing test_packaging test_doc_tables \- test_tokenizer_cache test_tokenizer_concurrency test_lazy_import \- test_runtime_fixes test_lang_guess test_identifier_complexity \- test_lang_stats test_calibration_persistence test_calibrate test_context_manager \- test_criteria_normalization test_docker_entrypoint \- test_empty_questions test_router_memory test_shortlist_cosine \- test_temperature_loading test_confidence test_cli test_mcp \- test_mcp_device \- test_langchain test_portability test_training test_train \- test_example_server_limits test_blank_lang_routing \- test_export_onnx_safety test_load_errors test_revision_pinning; do- if [ -f "tests/$t.py" ]; then python "tests/$t.py"; fi- done- shell: bash-- - name: Prediction hook tests- run: |- python tests/test_hooks.py- python tests/test_hooks_api.py- shell: bash-- - name: Question token reuse tests- run: python tests/test_question_token_reuse.py- shell: bash-- # Same pytest suites as the Linux job. They are named explicitly rather than added to- # the loop above, because the loop runs each file as a script and that is exactly what does- # not execute a pytest suite (#374).- - name: Pytest suites (serve, batch, system-one, audit regressions, truncation, compile)- run: |- python -m pytest tests/test_serve.py \- tests/test_router_batch.py \- tests/test_predict_batch.py \- tests/test_system_one_lang.py \- tests/test_audit_regressions.py \- tests/test_truncation_direction.py \- tests/test_compile.py \- -q- shell: bash-# The Linux job's guard, repeated here: the install is what makes test_mcp.py and# test_langchain.py exercise their real base classes, and nothing else proves it took. mcp- # and httpx are asserted the same way because their suites skip instead of failing.+ # and httpx are asserted the same way because their suites skip instead of failing. Run+ # this before the shared list so a missing extra cannot make the suites silently skip.- name: Integrations run on their real base classesrun: |python -c "import mcp, httpx, langchain_core"python -c "from laya.integrations.langchain import RunnableSerializable; \assert RunnableSerializable is not object, \'langchain-core is missing: laya.integrations.langchain fell back to a plain object'"- python tests/test_langchain_real_mode.pyshell: bash- - name: Email cleaning tests- run: python tests/test_email.py+ # The same canonical list as Linux and the release gate: an inline Windows copy had+ # omitted 32 suites, and the -f guard silently accepted missing suite files.+ - name: Test suites+ run: python scripts/test_suites.pyshell: bashlint:diff --git a/tests/test_packaging.py b/tests/test_packaging.pyindex fd1a328e..a341cc98 100644--- a/tests/test_packaging.py+++ b/tests/test_packaging.py@@ -329,10 +329,16 @@ untested_suites = [if f not in EXEMPT_TEST_SUITES and f not in registered_test_files]check("ci/wires every non-exempt test suite", untested_suites, [])-# The point of the shared list is that CI and the publish gate run the same suites. Both invoking-# the script is the entire mechanism, so assert neither workflow dropped back to an inline list.-check_true("ci and the release gate share one suite list",- "scripts/test_suites.py" in workflow and "scripts/test_suites.py" in release_workflow, "")+# Linux, Windows and the publish gate must invoke the shared list in their own jobs. A mention+# elsewhere in the workflow (including a comment) cannot prove that a lane runs those suites.+for label, source, job in (("ci/linux", workflow, "test"),+ ("ci/windows", workflow, "test-windows"),+ ("release/build", release_workflow, "build")):+ job_block = re.search(r"^ %s:\s*\n(.*?)(?=^ [\w-]+:|\Z)" % re.escape(job), source, re.M | re.S)+ check_true("%s runs the shared suite list" % label,+ job_block is not None and re.search(r"^\s+run:\s+python scripts/test_suites\.py\s*$",+ job_block.group(1), re.M) is not None,+ "expected run: python scripts/test_suites.py in %s" % job)# -------------------------------------------------- CONTRIBUTING.md's "Running the tests" section