config.json

{
  "model_type": "music2pose",
  "architectures": ["Music2PoseTransformer"],
  "task": "audio-to-pose-generation",
  "checkpoint_epoch": 180,
  "audio_dim": 82,
  "pose_dim": 297,
  "pose_landmarks": 33,
  "pose_channels": 9,
  "audio_sample_rate": 32000,
  "pose_fps": 25,
  "context_len": 60,
  "d_model": 256,
  "nhead": 8,
  "num_layers": 6,
  "dropout": 0.1,
  "max_seq_len": 260,
  "training_dataset": "AIST++ (camera c01)",
  "author": {
    "name": "Emma Genthon",
    "email": "genthonemma@gmail.com",
    "username": "Emma5099"
  }
}