llava_benchmark module

llava_benchmark.llava_benchmark(yaml_file_path, benchmarks)

Runs the LLAVA benchmark for different models, prompts, and media files. It calculates the average evaluation rates in tokens/s for each model and prints license plate or call audio benchmark results.

Parameters:

• yaml_file_path (str) – Path to the YAML configuration file.

• benchmarks (list) – List of benchmark objects.