forked from ailab/Qwen2-Audio-7B-Instruct
35 lines
853 B
JSON
35 lines
853 B
JSON
{
|
|
"architectures": [
|
|
"Qwen2AudioForConditionalGeneration"
|
|
],
|
|
"audio_config": {
|
|
"model_type": "qwen2_audio_encoder",
|
|
"num_mel_bins": 128,
|
|
"encoder_layers": 32,
|
|
"encoder_attention_heads": 20,
|
|
"encoder_ffn_dim": 5120,
|
|
"d_model": 1280,
|
|
"activation_function": "gelu",
|
|
"scale_embedding": false,
|
|
"max_source_positions": 1500
|
|
},
|
|
"audio_token_index": 151646,
|
|
"ignore_index": -100,
|
|
"model_type": "qwen2_audio",
|
|
"text_config": {
|
|
"bos_token_id": 151643,
|
|
"eos_token_id": 151645,
|
|
"intermediate_size": 11008,
|
|
"max_position_embeddings": 8192,
|
|
"model_type": "qwen2",
|
|
"rope_theta": 10000,
|
|
"rms_norm_eps": 1e-5,
|
|
"sliding_window": 32768,
|
|
"torch_dtype": "bfloat16",
|
|
"use_mrope": false,
|
|
"vocab_size": 156032
|
|
},
|
|
"transformers_version": "4.38.1",
|
|
"vocab_size": 156032
|
|
}
|