picolm_flutter 0.0.1
picolm_flutter: ^0.0.1 copied to clipboard
On-device LLM inference engine for Flutter. Run LLaMA-architecture models via Dart FFI — no cloud, no internet, no API keys.
0.0.1 #
- Initial release of
picolm_flutter. - Added
PicoLMclass withload(),generate(), andgenerateFull(). - Background isolate support for
load()andgenerate(). - Support for GGUF model files.
- JSON grammar mode (
setJsonMode). - Streaming token output.
- iOS, Android, and macOS platform support.