{ "architectures": [ "VoxtralRealtimeModel" ], "audio_config": { "activation_function": "silu", "attention_dropout": 0.0, "head_dim": 8, "hidden_act": "silu", "hidden_size": 32, "initializer_range": 0.02, "intermediate_size": 32, "max_position_embeddings": 256, "model_type": "voxtral_realtime_encoder", "num_attention_heads": 2, "num_hidden_layers": 2, "num_mel_bins": 80, "rms_norm_eps": 1e-06, "rope_parameters": { "rope_theta": 10000.0, "rope_type": "default" }, "sliding_window": 750, "vocab_size": 99 }, "audio_length_per_tok": 8, "default_num_delay_tokens": 6, "downsample_factor": 4, "dtype": "float32", "hidden_size": 32, "model_type": "voxtral_realtime", "pad_token_id": 1, "projector_hidden_act": "gelu", "text_config": { "attention_dropout": 0.0, "bos_token_id": 1, "eos_token_id": 2, "head_dim": 8, "hidden_act": "silu", "hidden_size": 32, "initializer_range": 0.02, "intermediate_size": 32, "max_position_embeddings": 256, "model_type": "voxtral_realtime_text", "num_attention_heads": 2, "num_hidden_layers": 2, "num_key_value_heads": 2, "pad_token_id": 1, "rms_norm_eps": 1e-06, "rope_parameters": { "rope_theta": 10000.0, "rope_type": "default" }, "sliding_window": 4096, "use_cache": true, "vocab_size": 99 }, "tie_word_embeddings": true, "transformers_version": "5.16.0.dev0" }