sylphx-keel-infer
Runs models: the inference Port (InferPort) with on-device null, scripted, host-fed, cascade, dense MLP and tract ONNX backends and per-call limits, typed decision models on candle or ONNX Runtime, calibration, and a text generation Port (GeneratePort) with a cost ledger and spend policy, and an on-device speech-to-text Port (TranscribePort); the remote OpenAI-compatible client is the opt-in generate-http feature and SenseVoice speech-to-text the opt-in transcribe-sensevoice feature
Activity
- Latest release
- 11h ago
- Total releases
- 1
- Cadence
- —
- Last 12 months
- 1
Details
- License
- unknown
- First release
- Oct 29, 2026
Releases