transformer/config.json

{
  "_class_name": "HYVideoDiffusionTransformer",
  "_diffusers_version": "0.31.0",
  "guidance_embed": true,
  "heads_num": 24,
  "hidden_size": 3072,
  "in_channels": 16,
  "mlp_act_type": "gelu_tanh",
  "mlp_width_ratio": 4,
  "mm_double_blocks_depth": 20,
  "mm_single_blocks_depth": 40,
  "out_channels": 16,
  "patch_size": [
    1,
    2,
    2
  ],
  "qk_norm": true,
  "qk_norm_type": "rms",
  "qkv_bias": true,
  "rope_dim_list": [
    16,
    56,
    56
  ],
  "text_states_dim": 4096,
  "text_states_dim_2": 768,
  "use_attention_mask": true
}