Skip to content

QuantForge v1.0.0 — Arm AI Auto-Tuning Benchmark

Choose a tag to compare

View all tags
@ash3353f ash3353f released this 14 Aug 12:16
· 4 commits to main since this release
v1.0.0
f46e20c

Description:
First public release of QuantForge.

Highlights:

  • Real FP32 vs INT8 benchmarking on Arm64 Android
  • Automatic 1 / 2 / 4 CPU-thread tuning
  • MobileNetV3-Small optimization
  • 71.6% model-size reduction measured
  • Best INT8 configuration achieved 8.55 ms median latency on OPPO CPH2591
  • 37.5% lower median latency than the best FP32 configuration
  • Dark and light themes
  • QuantForge Lab developer tools
Assets 3
Loading