Commit History

Add native custom q4 WebGPU runtime mode
0d908d4

linoyts HF Staff commited on

Add full custom WebGPU q4 FFN benchmark
7b8a4fb

linoyts HF Staff commited on

Add q4 FFN benchmark assets
8a51694
verified

multimodalart HF Staff commited on

Add custom WebGPU q4 FFN benchmark
1bebf01
verified

multimodalart HF Staff commited on

Add q4 MatMul PoC packed weights
9c6183c
verified

multimodalart HF Staff commited on

Add q4 MatMul PoC scales
72ce47e
verified

multimodalart HF Staff commited on

Add q4 MatMul PoC ONNX reference
cb60000
verified

multimodalart HF Staff commited on

Add custom WGSL q4 MatMul benchmark
6c20192
verified

multimodalart HF Staff commited on

Add sink-dot PoC ONNX reference
64f1111
verified

multimodalart HF Staff commited on

Add sink-dot PoC weights
1e09158
verified

multimodalart HF Staff commited on

Add custom WGSL sink-dot benchmark
db22e73
verified

multimodalart HF Staff commited on

Use ONNX Runtime Web 1.26 for all engines
3f567eb
verified

multimodalart HF Staff commited on

Add q4-full ONNX Runtime 1.26 benchmark mode
9abbfd6
verified

multimodalart HF Staff commited on

Remove slower sliced q4-full mode
65df88c
verified

multimodalart HF Staff commited on

Add sliced q4-full benchmark mode
0792afd
verified

multimodalart HF Staff commited on

Remove experimental hybrid q4 mode
a4bc41b
verified

multimodalart HF Staff commited on

Fix hybrid GPU token readback
1a26cf0
verified

multimodalart HF Staff commited on

Add experimental hybrid q4 mode
38d88ed
verified

multimodalart HF Staff commited on

Optimize split q4 top-k sampling
af55ec9
verified

multimodalart HF Staff commited on

Show per-engine frame and decoder timing
c7df7cf
verified

multimodalart HF Staff commited on

Add engine selector and restore split q4 default
e8ff614
verified

multimodalart HF Staff commited on

Add opt-in fully quantized frame benchmark
1111b66
verified

multimodalart HF Staff commited on

Use fused WebGPU frame graph
af9f3b8
verified

multimodalart HF Staff commited on

Upload index.html with huggingface_hub
b09684b
verified

multimodalart HF Staff commited on

Upload README.md with huggingface_hub
5f464ed
verified

multimodalart HF Staff commited on

initial commit
dcbdefa
verified

multimodalart HF Staff commited on