What's Changed
- Preserve sparseness when copying the storage template during pack creation by @BinSquare in #604
- cuda: llama.cpp + CUDA 13, correctness hardening, cuBLAS BLAS ops, and remote latency/coldstart cuts by @BinSquare in #607
- Bump the workspace to 1.5.2 for the storage-template sparse-copy fix by @BinSquare in #608
Full Changelog: v1.5.1...v1.5.2