config.json

{
  "auto_map": {
    "AutoConfig": "configuration_meralion.MERaLiONConfig"
  },
  "head_dim": 256,
  "hidden_size": 3584,
  "intermediate_size": 14336,
  "model_type": "meralion",
  "num_attention_heads": 16,
  "num_hidden_layers": 42,
  "num_key_value_heads": 8,
  "sliding_window": 4096,
  "speech_config": {
    "_name_or_path": "openai/whisper-large-v3",
    "apply_spec_augment": true,
    "architectures": [
      "WhisperForConditionalGeneration"
    ],
    "begin_suppress_tokens": [
      220,
      50257
    ],
    "bos_token_id": 50257,
    "d_model": 1280,
    "decoder_attention_heads": 20,
    "decoder_ffn_dim": 5120,
    "decoder_layers": 32,
    "decoder_start_token_id": 50258,
    "encoder_attention_heads": 20,
    "encoder_ffn_dim": 5120,
    "encoder_layers": 32,
    "eos_token_id": 50257,
    "mask_time_length": 20,
    "max_length": 448,
    "model_type": "meralion_speech_encoder",
    "num_hidden_layers": 32,
    "num_mel_bins": 128,
    "torch_dtype": "bfloat16",
    "vocab_size": 51866
  },
  "speech_mlp_scale_factor": 15,
  "speech_token_index": 255999,
  "text_config": {
    "_name_or_path": "aisingapore/gemma2-9b-cpt-sea-lionv3-instruct",
    "architectures": [
      "Gemma2ForCausalLM"
    ],
    "eos_token_id": 107,
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 3584,
    "intermediate_size": 14336,
    "model_type": "meralion_text_decoder",
    "num_hidden_layers": 42,
    "num_key_value_heads": 8,
    "query_pre_attn_scalar": 256,
    "sliding_window_size": 4096,
    "torch_dtype": "bfloat16"
  },
  "transformers_version": "4.44.2"
}