transformer/config.json

{
  "_class_name": "LTXModel",
  "_diffusers_version": "0.36.0",
  "attention_head_dim": 128,
  "attention_type": "default",
  "audio_attention_head_dim": 64,
  "audio_cross_attention_dim": 2048,
  "audio_in_channels": 128,
  "audio_num_attention_heads": 32,
  "audio_out_channels": 128,
  "audio_positional_embedding_max_pos": [
    20
  ],
  "av_ca_timestep_scale_multiplier": 1000.0,
  "caption_channels": 3840,
  "cross_attention_dim": 4096,
  "double_precision_rope": true,
  "in_channels": 128,
  "model_type": "audio-video",
  "norm_eps": 1e-06,
  "num_attention_heads": 32,
  "num_layers": 48,
  "out_channels": 128,
  "positional_embedding_max_pos": [
    20,
    2048,
    2048
  ],
  "positional_embedding_theta": 10000.0,
  "rope_type": "split",
  "timestep_scale_multiplier": 1000,
  "use_middle_indices_grid": true
}