Testing

The default suite uses small deterministic RT-J checkpoints and realistic customer contexts containing typed customer, order, and support-ticket cells. It does not need network access or a GPU.

pytest

The portable coverage gate excludes CUDA-only Triton kernel implementations, which cannot execute on CPU CI, and requires at least 90% coverage. The current portable suite reports over 97%:

make coverage

The suite covers:

• typed relational batches, wide production-style node IDs, foreign-key parents, and padding;

• classification, regression, batching, output views, and explicit support-ticket ablation;

• checkpoint save/reload plus int8 and packed-int4 dequantization;

• frozen-backbone binary and multiclass head tuning and full-model fine-tuning;

• dynamic-batch and dynamic-context ONNX export and inference;

• Triton sorting and relational attention work-list construction;

• malformed shapes, missing fields, invalid semantic types, and non-finite values;

• meta-device architecture inspection without parameter allocation.

Published-model tests are opt-in because they download large checkpoints:

RUN_HUB_TESTS=1 pytest -m hub

On a CUDA deployment host, compare the optimized Triton output directly with the PyTorch backend on the same relational context:

RUN_CUDA_TESTS=1 pytest -m cuda

The RelativeDB repository has an additional end-to-end integration test. It runs a real PREDICT query through retrieval, cell encoding, relational batch construction, and this package’s PyTorch runtime.