| { | |
| "architectures": [ | |
| "VoxtralAudioFeatureExtractor" | |
| ], | |
| "model_type": "voxtral_audio_feature_extractor", | |
| "auto_map": { | |
| "AutoConfig": "modeling_voxtral_audio.VoxtralAudioFeatureConfig", | |
| "AutoModel": "modeling_voxtral_audio.VoxtralAudioFeatureExtractor" | |
| }, | |
| "concat_factor": 4, | |
| "projector_hidden_act": "gelu", | |
| "torch_dtype": "bfloat16", | |
| "audio_config": { | |
| "model_type": "voxtral_encoder", | |
| "d_model": 1280, | |
| "encoder_attention_heads": 20, | |
| "encoder_ffn_dim": 5120, | |
| "encoder_layers": 32, | |
| "num_mel_bins": 128 | |
| }, | |
| "text_config": { | |
| "model_type": "llama", | |
| "hidden_size": 3072 | |
| } | |
| } |