tiny_model_marker.json

{
  "marker_version": 1,
  "source_model_id": "diffusers/LTX-2.3-Diffusers",
  "seed": 42,
  "text": {
    "hidden_size": 32,
    "num_hidden_layers": 2,
    "vocab_size": 1024
  },
  "transformer": {
    "num_layers": 1,
    "num_attention_heads": 4,
    "attention_head_dim": 8,
    "cross_attention_dim": 32,
    "caption_channels": 32,
    "in_channels": 8,
    "out_channels": 8,
    "audio_num_attention_heads": 4,
    "audio_attention_head_dim": 8,
    "audio_cross_attention_dim": 32
  },
  "vae": {
    "latent_channels": 8,
    "block_out_channels": [
      8,
      8,
      8,
      8
    ],
    "decoder_block_out_channels": [
      8,
      16,
      16,
      32
    ],
    "layers_per_block": [
      1,
      1,
      1,
      1,
      1
    ],
    "decoder_layers_per_block": [
      1,
      1,
      1,
      1,
      1
    ]
  },
  "audio_vae": {
    "ch_mult": [
      1
    ],
    "num_res_blocks": 1
  },
  "connectors": {
    "caption_channels": 32,
    "text_proj_in_factor": 3,
    "video_connector_num_layers": 1,
    "video_connector_num_attention_heads": 4,
    "video_connector_attention_head_dim": 8,
    "video_connector_num_learnable_registers": 8,
    "video_hidden_dim": 32,
    "audio_connector_num_layers": 1,
    "audio_connector_num_attention_heads": 4,
    "audio_connector_attention_head_dim": 8,
    "audio_connector_num_learnable_registers": 8,
    "audio_hidden_dim": 32
  },
  "vocoder": {
    "hidden_channels": 64,
    "resnet_kernel_sizes": [
      3
    ],
    "resnet_dilations": [
      [
        1,
        3,
        5
      ]
    ],
    "bwe_hidden_channels": 32,
    "bwe_resnet_kernel_sizes": [
      3
    ],
    "bwe_resnet_dilations": [
      [
        1,
        3,
        5
      ]
    ]
  }
}