A small, serialized wrapper over LiteRT-LM: one shared engine, GPU→CPU fallback ladder, retry-aware sampling, streamed and one-shot generation.

Latest Versions

2 versions →
Version ▼VulnerabilitiesUsagesDate
0.1.x
0.1.2
1
Nov 21, 2026
0.1.0
1
Sep 13, 2026