config.json

{
  "_name_or_path": "Phi-4-multimodal-instruct",
  "architectures": [
    "Phi4MMForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "audio_processor": {
    "config": {
      "activation": "swish",
      "activation_checkpointing": {
        "interval": 1,
        "module": "transformer",
        "offload": false
      },
      "attention_dim": 1024,
      "attention_heads": 16,
      "batch_norm": false,
      "bias_in_glu": true,
      "causal": true,
      "chunk_size": -1,
      "cnn_layer_norm": true,
      "conv_activation": "swish",
      "conv_glu_type": "swish",
      "depthwise_multiplier": 1,
      "depthwise_seperable_out_channel": 1024,
      "dropout_rate": 0.0,
      "encoder_embedding_config": {
        "input_size": 80
      },
      "ext_pw_kernel_size": 1,
      "ext_pw_out_channel": 1024,
      "input_layer": "nemo_conv",
      "input_size": 80,
      "kernel_size": 3,
      "left_chunk": 18,
      "linear_units": 1536,
      "nemo_conv_settings": {
        "conv_channels": 1024
      },
      "num_blocks": 24,
      "relative_attention_bias_args": {
        "t5_bias_max_distance": 500,
        "type": "t5"
      },
      "time_reduction": 8
    },
    "name": "cascades"
  },
  "auto_map": {
    "AutoConfig": "configuration_phi4mm.Phi4MMConfig",
    "AutoModelForCausalLM": "modeling_phi4mm.Phi4MMForCausalLM",
    "AutoTokenizer": "Xenova/gpt-4o"
  },
  "bos_token_id": 199999,
  "embd_layer": {
    "audio_embd_layer": {
      "compression_rate": 8,
      "downsample_rate": 1,
      "embedding_cls": "audio",
      "enable_gradient_checkpointing": true,
      "projection_cls": "mlp",
      "use_conv_downsample": false,
      "use_qformer": false
    },
    "embedding_cls": "image_audio",
    "image_embd_layer": {
      "crop_size": 448,
      "embedding_cls": "tune_image",
      "enable_gradient_checkpointing": true,
      "hd_transform_order": "sub_glb",
      "image_token_compression_cls": "avg_pool_2d",
      "projection_cls": "mlp",
      "use_hd_transform": true,
      "with_learnable_separator": true
    }
  },
  "embd_pdrop": 0.0,
  "eos_token_id": 199999,
  "full_attn_mod": 1,
  "hidden_act": "silu",
  "hidden_size": 3072,
  "initializer_range": 0.02,
  "intermediate_size": 8192,
  "interpolate_factor": 1,
  "lm_head_bias": false,
  "vision_lora": {
    "dp": 0.0,
    "layer": "layers.*((self_attn\\.(qkv_proj|o_proj))|(mlp\\.(gate_up|down)_proj))",
    "lora_alpha": 512,
    "r": 256
  },
  "speech_lora": {
    "dp": 0.01,
    "layer": "((layers.*self_attn\\.(qkv|o)_proj)|(layers.*mlp\\.(gate_up|down)_proj))",
    "lora_alpha": 640,
    "r": 320
  },
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "phi4mm",
  "num_attention_heads": 24,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "original_max_position_embeddings": 4096,
  "pad_token_id": 199999,
  "partial_rotary_factor": 0.75,
  "resid_pdrop": 0.0,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "long_factor": [
      1,
      1.118320672,
      1.250641126,
      1.398617824,
      1.564103225,
      1.74916897,
      1.956131817,
      2.187582649,
      2.446418898,
      2.735880826,
      3.059592084,
      3.421605075,
      3.826451687,
      4.279200023,
      4.785517845,
      5.351743533,
      5.984965424,
      6.693110555,
      7.485043894,
      8.370679318,
      9.36110372,
      10.4687158,
      11.70738129,
      13.09260651,
      14.64173252,
      16.37415215,
      18.31155283,
      20.47818807,
      22.90118105,
      25.61086418,
      28.64115884,
      32.03,
      32.1,
      32.13,
      32.23,
      32.6,
      32.61,
      32.64,
      32.66,
      32.7,
      32.71,
      32.93,
      32.97,
      33.28,
      33.49,
      33.5,
      44.16,
      47.77
    ],
    "short_factor": [
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      1.0
    ],
    "type": "longrope"
  },
  "rope_theta": 10000.0,
  "sliding_window": 262144,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.46.1",
  "use_cache": true,
  "vocab_size": 200064,
  "_attn_implementation": "flash_attention_2"
}