headroom 0.1.0 copy "headroom: ^0.1.0" to clipboard
headroom: ^0.1.0 copied to clipboard

A two-second on-device probe that tells your app how fast an LLM of a given size will decode on this phone, whether it fits, and on what basis.

0.1.0 #

First release.

  • Headroom.probe() runs the Headroom Swift package's probe on iOS (Metal STREAM ceiling, C STREAM triad, memory, thermal, power, device) and the same C triad with Android's memory, thermal, battery and device facts on Android, returning a typed ProbeReport with a basis on every figure.
  • ProbeReport.estimate() is a Dart port of the package's estimator and memory-fit maths, driven by a copy of its calibration.json; predictions on Android carry basis unknown until an Android calibration exists.
  • ModelSpec.tinyLlama1_1BQ4_0 and ModelSpec.qwen2_5_0_5BInstructQ4_K_M, the two presets verified from GGUF headers.
  • tool/sync_native.sh keeps the native sources identical to the package.
0
likes
150
points
337
downloads

Documentation

API reference

Publisher

unverified uploader

Weekly Downloads

A two-second on-device probe that tells your app how fast an LLM of a given size will decode on this phone, whether it fits, and on what basis.

Repository (GitHub)
View/report issues

Topics

#llm #on-device-ai #benchmark #memory-bandwidth #inference

License

MIT (license)

Dependencies

flutter

More

Packages that depend on headroom

Packages that implement headroom