transformer/config.json

{
  "_class_name": "CausalHunyuanVideo15Transformer3DModel",
  "_diffusers_version": "0.36.0.dev0",
  "attention_head_dim": 128,
  "image_embed_dim": 1152,
  "in_channels": 65,
  "mlp_ratio": 4.0,
  "num_attention_heads": 16,
  "num_layers": 54,
  "num_refiner_layers": 2,
  "out_channels": 32,
  "patch_size": 1,
  "patch_size_t": 1,
  "qk_norm": "rms_norm",
  "rope_axes_dim": [
    16,
    56,
    56
  ],
  "rope_theta": 256.0,
  "target_size": 640,
  "task_type": "t2v",
  "text_embed_2_dim": 1472,
  "text_embed_dim": 3584,
  "use_meanflow": false
}