transformer/config.json

{
  "_class_name": "StableAudioDiTModel",
  "_diffusers_version": "0.30.0.dev0",
  "attention_head_dim": 64,
  "cross_attention_dim": 768,
  "cross_attention_input_dim": 768,
  "global_states_input_dim": 1536,
  "in_channels": 64,
  "num_attention_heads": 24,
  "num_key_value_attention_heads": 12,
  "num_layers": 24,
  "out_channels": 64,
  "sample_size": 1024.0,
  "time_proj_dim": 256
}