Files
vllm-omni/tests/dfx/perf/tests/test_higgs_audio_v3.json
T

91 lines
2.1 KiB
JSON

[
{
"test_name": "test_higgs_audio_v3",
"mark": [
{
"hardware_marks": {
"res": {
"cuda": "H100"
},
"num_cards": 1
}
},
"full_model",
"tts"
],
"server_params": {
"model": "bosonai/higgs-audio-v3-tts-4b",
"stage_overrides": {
"0": {
"gpu_memory_utilization": 0.3
},
"1": {
"gpu_memory_utilization": 0.2
}
},
"extra_cli_args": [
"--deploy-config",
"vllm_omni/deploy/higgs_multimodal_qwen3.yaml",
"--trust-remote-code"
]
},
"benchmark_params": [
{
"task": "default_voice",
"eval_phase": "latency",
"dataset_name": "seed-tts-text",
"backend": "openai-audio-speech",
"endpoint": "/v1/audio/speech",
"dataset_path": "benchmarks/build_dataset/seed_tts_smoke",
"num_prompts": [
20
],
"max_concurrency": [
1
],
"seed_tts_locale": "en",
"percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"median_audio_ttfp_ms": [
223.3584
],
"median_audio_rtf": [
0.4393
]
}
}
},
{
"task": "default_voice",
"eval_phase": "throughput",
"dataset_name": "seed-tts-text",
"backend": "openai-audio-speech",
"endpoint": "/v1/audio/speech",
"dataset_path": "benchmarks/build_dataset/seed_tts_smoke",
"num_prompts": [
500
],
"max_concurrency": [
16
],
"seed_tts_locale": "en",
"percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"median_audio_ttfp_ms": [
276.575
],
"median_audio_rtf": [
0.5073
],
"audio_throughput": [
27.3728
]
}
}
}
]
}
]