transformer/config.json

{
  "_class_name": "Gen3CTransformer3DModel",
  "in_channels": 16,
  "out_channels": 16,
  "num_attention_heads": 32,
  "attention_head_dim": 128,
  "num_layers": 28,
  "mlp_ratio": 4.0,
  "text_embed_dim": 1024,
  "adaln_lora_dim": 256,
  "use_adaln_lora": true,
  "add_augment_sigma_embedding": false,
  "frame_buffer_max": 2,
  "max_size": [
    128,
    240,
    240
  ],
  "patch_size": [
    1,
    2,
    2
  ],
  "rope_scale": [
    2.0,
    1.0,
    1.0
  ],
  "extra_pos_embed_type": "learnable",
  "concat_padding_mask": true,
  "affine_emb_norm": true,
  "qk_norm": "rms_norm",
  "eps": 1e-06
}