38 lines
892 B
YAML
38 lines
892 B
YAML
# OpenAI-compatible media server for MMAudio large-44k-v2 video-to-audio.
|
|
# fastvideo serve --config examples/serving/openai_mmaudio.yaml
|
|
# Send video_reference (local path or URL). Audio uses the /v1/videos job
|
|
# endpoints and returns media_type audio/wav.
|
|
generator:
|
|
model_path: FastVideo/MMAudio-large-44k-v2-Diffusers
|
|
engine:
|
|
num_gpus: 1
|
|
use_fsdp_inference: false
|
|
parallelism:
|
|
tp_size: 1
|
|
sp_size: 1
|
|
offload:
|
|
dit: false
|
|
vae: false
|
|
text_encoder: false
|
|
pin_cpu_memory: false
|
|
|
|
pipeline:
|
|
workload_type: v2a
|
|
|
|
server:
|
|
host: 0.0.0.0
|
|
port: 8000
|
|
output_dir: outputs/openai_mmaudio
|
|
served_model_name: mmaudio-large-44k-v2
|
|
|
|
default_request:
|
|
negative_prompt: music
|
|
sampling:
|
|
audio_start_in_s: 0.0
|
|
audio_end_in_s: 8.0
|
|
num_inference_steps: 25
|
|
guidance_scale: 4.5
|
|
seed: 42
|
|
output:
|
|
return_frames: false
|