# FastH3 on two DGX Sparks (one GPU each) over QSFP RoCE.
# Bring up the Ray cluster first: docs/getting_started/installation/spark_pair.md
#
# request.sampling below is an example, not a required recipe. Change height,
# width, num_frames, num_inference_steps, seed, and prompt. Legal H3 frame
# counts are 17n+5, max 345 (15 s).
#
#   source examples/inference/optimizations/spark_pair_env.sh
#   export RAY_ADDRESS=<qsfp-head>:6379
#   export FASTVIDEO_HOST_IP=<qsfp-head>
#   FASTVIDEO_VSA_SM100A=0 FASTVIDEO_FA4=0 FASTVIDEO_ATTENTION_BACKEND=VIDEO_SPARSE_ATTN_H3 \
#     FASTVIDEO_VAE_PARALLEL_DECODE=1 FASTVIDEO_STAGE_LOGGING=1 \
#     fastvideo generate --config examples/inference/basic/basic_fasth3_spark_pair.yaml
generator:
  model_path: FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree
  engine:
    num_gpus: 2
    execution_backend: ray
    use_fsdp_inference: false
    parallelism:
      tp_size: 1
      sp_size: 2
    offload:
      dit: false
      dit_layerwise: false
      text_encoder: true
      vae: true
      pin_cpu_memory: true
      lazy_module_load: true
    compile:
      enabled: false
      vae_enabled: true
  pipeline:
    experimental:
      attention_backend: VIDEO_SPARSE_ATTN_H3
      VSA_sparsity: 0.9
      VSA_tile_size: 64
      inference_torch_compile: true
      vae_parallel_decode: true
      vae_parallel_decode_strategy: gather
request:
  prompt: >-
    A wide cinematic shot of an alpine meadow at sunrise, pale pink mountain
    peaks above a blue valley filled with thin morning mist.
  negative_prompt: ""
  sampling:
    seed: 2026
    height: 768
    width: 1344
    num_frames: 124
    fps: 24
    num_inference_steps: 5
    guidance_scale: 1.0
    batch_cfg: false
  output:
    output_path: outputs/fasth3_spark_pair/
    save_video: true
    return_frames: false