Name	Name	Last commit message	Last commit date
parent directory ..
README.md	README.md
host_ab_linkedin_summary.png	host_ab_linkedin_summary.png
netkit_linkedin_float32_suite.png	netkit_linkedin_float32_suite.png
netkit_linkedin_int8_suite.png	netkit_linkedin_int8_suite.png
netkit_linkedin_wins.png	netkit_linkedin_wins.png
render_wins_summary.py	render_wins_summary.py

Name

Last commit message

Last commit date

Peer-suite infographics

PNG summaries of completed inference A/B across MCU (vs TFLM), MPU (Arm Cortex-A53 vs TF Lite), and CPU (Apple M4 vs TF Lite / ORT).

File	Contents
`netkit_linkedin_wins.png`	LinkedIn one-panel — MCU · MPU · CPU wins vs TFLM / microTVM / TF Lite / ONNX Runtime. Shown on repo README and STATUS — Peer highlights
`netkit_linkedin_int8_suite.png`	Int8 latency (+ MCU flash/RAM) — shown on repo README
`netkit_linkedin_float32_suite.png`	Float32 latency — shown on repo README
`host_ab_linkedin_summary.png`	Older host-oriented LinkedIn summary card (not on README)

Numbers and methodology: docs/STATUS.md.

benchmark/tflite/.venv/bin/python benchmark/linkedin/render_wins_summary.py
# or any Python 3 with Pillow

Update the hardcoded latencies in render_wins_summary.py when STATUS peer tables change.

Navigation Menu