config.json

{
  "activation_type": "silu",
  "add_faster_video": false,
  "add_time_instruction": false,
  "alibi": false,
  "alibi_bias_max": 8.0,
  "architectures": [
    "LlavaLladaForMaskedDiffusion"
  ],
  "attention_dropout": 0.0,
  "attention_layer_norm": false,
  "attention_layer_norm_with_affine": true,
  "auto_map": {
    "AutoConfig": "configuration_llada.LLaDAConfig",
    "AutoModel": "modeling_llada.LLaDAModelLM",
    "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"
  },
  "bias_for_layer_norm": false,
  "block_causal": true,
  "block_causal_purge": 0.1,
  "block_group_size": 1,
  "block_type": "llama",
  "d_model": 4096,
  "d_model_gen": 2048,
  "downsample": true,
  "dual_tower": true,
  "dual_tower_layers": 32,
  "dual_tower_out": true,
  "effective_n_kv_heads": 32,
  "embedding_dropout": 0.0,
  "embedding_size": 127378,
  "enc_use_image_branch": true,
  "eos_token_id": 126081,
  "faster_token_stride": 10,
  "flash_attention": false,
  "force_sample": false,
  "gen_enc_add_pos_emb": true,
  "gen_mlp_type": "dit2",
  "image_aspect_ratio": "anyres",
  "image_crop_resolution": null,
  "image_enc_drop_rate": 0.5,
  "image_grid_pinpoints": [
    [
      384,
      768
    ],
    [
      768,
      384
    ],
    [
      768,
      768
    ],
    [
      1152,
      384
    ],
    [
      384,
      1152
    ]
  ],
  "image_split_resolution": null,
  "include_bias": false,
  "include_qkv_bias": false,
  "init_cutoff_factor": null,
  "init_device": "cpu",
  "init_fn": "mitchell",
  "init_std": 0.02,
  "input_emb_norm": false,
  "layer_norm_type": "rms",
  "layer_norm_with_affine": true,
  "mask_token_id": 126336,
  "max_sequence_length": 4096,
  "mlp_hidden_size": 12288,
  "mlp_hidden_size_gen": 8192,
  "mlp_ratio": 4,
  "mm_hidden_size": 1152,
  "mm_newline_position": "grid",
  "mm_patch_merge_type": "spatial_unpad",
  "mm_pooler_ratio": 2,
  "mm_projector_lr": null,
  "mm_projector_type": "mlp2x_gelu",
  "mm_resampler_type": "spatial_pool",
  "mm_spatial_pool_mode": "conv",
  "mm_spatial_pool_out_channels": 1152,
  "mm_spatial_pool_stride": 2,
  "mm_tunable_parts": "mm_vision_tower,mm_mlp_adapter,mm_vision_resampler,mm_language_model,mm_language_model_vision_parms",
  "mm_use_im_patch_token": false,
  "mm_use_im_start_end": false,
  "mm_vision_select_feature": "patch",
  "mm_vision_select_layer": -2,
  "mm_vision_tower": "/mnt/localssd/siglip-so400m-patch14-384",
  "mm_vision_tower_lr": 2e-06,
  "mm_vqvae": "/mnt/localssd/Meissonic/vqvae",
  "model_type": "llada",
  "multi_query_attention": null,
  "n_heads": 32,
  "n_kv_heads": 32,
  "n_layers": 32,
  "num_register_groups": "25,25",
  "num_register_tokens": "64,64",
  "pad_token_id": 126081,
  "pos_skipping_range": 4096,
  "precision": "amp_bf16",
  "prompt_drop_rate": 0.1,
  "resampler_type": null,
  "residual_dropout": 0.0,
  "rms_norm_eps": 1e-05,
  "rope": true,
  "rope_full_precision": true,
  "rope_theta": 500000.0,
  "scale_logits": false,
  "tie_word_embeddings": false,
  "tokenizer_model_max_length": 8192,
  "tokenizer_padding_side": "right",
  "torch_dtype": "bfloat16",
  "transformers_version": "4.50.3",
  "use_cache": false,
  "use_mm_proj": true,
  "use_pos_skipping": false,
  "vision_tower_pretrained": null,
  "vocab_size": 127378,
  "weight_tying": false
}