config.json

{
  "architectures": [
    "MusicFlamingoForConditionalGeneration"
  ],
  "audio_bos_token_id": 151670,
  "audio_config": {
    "activation_dropout": 0.0,
    "activation_function": "gelu",
    "attention_dropout": 0.0,
    "dropout": 0.0,
    "hidden_size": 1280,
    "initializer_range": 0.02,
    "intermediate_size": 5120,
    "layerdrop": 0.0,
    "max_source_positions": 1500,
    "model_type": "audioflamingo3_encoder",
    "num_attention_heads": 20,
    "num_hidden_layers": 32,
    "num_mel_bins": 128,
    "scale_embedding": false
  },
  "audio_eos_token_id": 151671,
  "audio_frame_step": 0.01,
  "audio_token_id": 151667,
  "dtype": "bfloat16",
  "head_dim": 1280,
  "max_position_embeddings": 1200,
  "model_type": "musicflamingo",
  "projector_bias": true,
  "projector_hidden_act": "gelu",
  "rope_parameters": {
    "partial_rotary_factor": 0.2,
    "rope_theta": 1200,
    "rope_type": "default"
  },
  "text_config": {
    "attention_dropout": 0.0,
    "bos_token_id": 151668,
    "eos_token_id": 151645,
    "hidden_act": "silu",
    "hidden_size": 3584,
    "initializer_range": 0.02,
    "intermediate_size": 18944,
    "layer_types": [
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention",
      "full_attention"
    ],
    "max_position_embeddings": 32768,
    "max_window_layers": 28,
    "model_max_length": 32768,
    "model_type": "qwen2",
    "num_attention_heads": 28,
    "num_hidden_layers": 28,
    "num_key_value_heads": 4,
    "pad_token_id": 151669,
    "rms_norm_eps": 1e-06,
    "rope_parameters": {
      "rope_theta": 1000000.0,
      "rope_type": "default"
    },
    "sliding_window": null,
    "tie_word_embeddings": false,
    "use_cache": false,
    "use_sliding_window": false,
    "vocab_size": 151672
  },
  "transformers_version": "5.6.0.dev0",
  "vocab_size": 500
}