Record production-default full TensorRT smoke validation

Record successful checkpoint-free full TensorRT E2E validation

Validate corrected SLat DAE custom TensorRT branches and heads

Validate sparse-to-FlexiCubes TensorRT plugin at production res256

Validate production sparse mesh topology TensorRT plugin at res256

Match upstream CUDA autocast in native SLat DAE validation

Add native-vs-TensorRT validation for full SLat DAE branches and heads

Validate all sparse custom TensorRT ops after final SparseConv3D fix

Add FlexiCubes TensorRT mesh topology validation

Add numerical validation for sparse TensorRT custom ops

Validate AniGen SLat TensorRT core against sparse native model

Add AniGen TensorRT numerical and latency validation

Add reproducible L4 end-to-end smoke test

Add TensorRT vs ONNX numerical validation

Add L4 ONNX TensorRT hybrid runtime scaffold