audio/config.json

{
  "_name_or_path": "/mnt/petrelfs/caoyuhang/InternLM-XComposer/finetune_audio/models/sft_base_3x_with_pt_extra",
  "architectures": [
    "Qwen2AudioForConditionalGeneration"
  ],
  "audio_config": {
    "model_type": "qwen2_audio_encoder"
  },
  "audio_token_index": 151646,
  "hidden_size": 1536,
  "ignore_index": -100,
  "model_type": "qwen2_audio",
  "text_config": {
    "bos_token_id": 151643,
    "eos_token_id": 151645,
    "hidden_size": 1536,
    "intermediate_size": 8960,
    "max_position_embeddings": 8192,
    "model_type": "qwen2",
    "num_attention_heads": 12,
    "num_hidden_layers": 28,
    "num_key_value_heads": 2,
    "rms_norm_eps": 1e-05,
    "tie_word_embeddings": true,
    "torch_dtype": "bfloat16",
    "use_mrope": false,
    "vocab_size": 156032
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.45.0.dev0",
  "use_cache": false,
  "vocab_size": 156032
}