This benchmark measures the execution time of Xenova/all-MiniLM-L6-v2 (bert-based embedding model) using the WASM and WebGPU execution providers across different batch sizes.