Embed hfviewer card in model card

Replace benchmark tables with link to Edge Inference Benchmarks space

Update benchmark image to match FlashHead

Make quantization more robust

Add new_version

Update README.md

Model card: add top accuracy plot image for Edge2

Update README.md

Update README.md

Model card: align Edge2 optimization wording with review comment

Model card: rotate accuracy tables to model-comparison layout

Remove Edge2 text in Additional Benchmarks

Update README.md

Model card: split all performance tables by device

Model card: split final batch_size=8 additional benchmarks by device

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Update README.md

Move vLLM Python test hardware label above Python code block

Add test hardware annotations to usage examples

Reformat accuracy tables to one table per model with model headings

Update README.md

Update README.md

Update license section

Simplify measurement setup text per request

Fill missing unquantized accuracy table cells using 50.6 rerun (r39)

Remove divider before output example

Simplify output example source note

Move output example before model details and simplify quote labels

Embed fridge video in output example with fallback link

Update output example to fridge video and trim final sentence

Add public video URL to output example prompt

Replace output example with real logged verbatim output

Restructure model card to NVFP-style accuracy/performance layout

Update model card references to renamed W4A16-Edge2 repo

Fix Edge2 card: valid repo IDs, rename mixed labels, fill missing video f2 rows

Rename model mentions from W4A16-Mixed to W4A16-Edge2 in model card

Update model card with final mixed accuracy and latency tables

Remove dtype

Update model card with completed AGX mixed latency benchmarks

Update model card tables with latest Nano/AGX benchmark results

Update model card

Upload W4A16 mixed model (exclude recipe yaml)

initial commit