dit/config.json

{
  "_class_name": "MGVideoDiffusionTransformerI2V",
  "_diffusers_version": "0.32.2",
  "action_config": {
    "enable_keyboard": true,
    "enable_mouse": true,
    "heads_num": 24,
    "hidden_size": 128,
    "img_hidden_size": 3072,
    "keyboard_dim_in": 6,
    "mlp_act_type": "gelu_tanh",
    "mouse_dim_in": 2,
    "mouse_qk_dim_list": [
      16,
      56,
      56
    ],
    "patch_size": [
      1,
      2,
      2
    ],
    "qk_norm": "True",
    "qk_norm_type": "rms",
    "qkv_bias": "False",
    "rope_dim_list": [
      16,
      56,
      56
    ],
    "rope_theta": 256,
    "vae_time_compression_ratio": 4,
    "windows_size": 3
  },
  "device": null,
  "dtype": null,
  "guidance_embed": false,
  "heads_num": 24,
  "hidden_size": 3072,
  "i2v_condition_type": "token_replace",
  "in_channels": 33,
  "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
  ],
  "rope_theta": 256,
  "text_projection": "single_refiner",
  "text_states_dim": 4096,
  "text_states_dim_2": 768,
  "use_attention_mask": true
}