SAM-3D-Body: disclose encoder-only scope in setup note

SAM3: correct AGX Orin rows to model-card table values

Parakeet: correct AGX numbers, add L4 device + peak-memory metric

Add deploy-benchmark model families (8 models, lat/ips/mem metrics)

Add FlashHead version to measurement setup

Add NVIDIA prefix to Device names

Qwen3.5-FlashHead (#6)

Add accuracy tables, RTX 3500 Ada latency, and fix 270M variant grouping

Group table rows by model sub-family (size) instead of individual model

Add Llama-3.2, Gemma-3, and Qwen3 benchmarks and improve chart rendering

Demo (#5)

Add accuracy data (#4)

Add subgroups to sidebar (#3)

Add Qwen3.5 Benchmarks (#2)

Update chart and table styling (#1)

Update page

Add data_file key-value to Cosmos-Reason2-2B

Upload 8 files

Initial website