{ "architectures": [ "LingV3TTSForConditionalGeneration" ], "model_type": "lingv3_tts", "auto_map": { "AutoConfig": "configuration_lingv3_tts.LingV3TTSConfig", "AutoModel": "modeling_lingv3_tts.LingV3TTSForConditionalGeneration" }, "tie_word_embeddings": false, "llm_config": { "architectures": [ "BailingMoeV3ForCausalLM" ], "model_type": "bailing_moe_v3", "vocab_size": 157184, "hidden_size": 1536, "intermediate_size": 4608, "num_hidden_layers": 24, "num_attention_heads": 16, "num_key_value_heads": 16, "hidden_act": "silu", "use_qkv_bias": false, "use_bias": false, "rms_norm_eps": 1e-06, "tie_word_embeddings": false, "embedding_dropout": 0.0, "attention_dropout": 0.0, "output_dropout": 0.0, "initializer_range": 0.006, "max_position_embeddings": 4096, "rope_theta": 10000, "use_cache": true, "max_window_layers": 24, "rope_scaling": null, "bos_token_id": 156891, "pad_token_id": 156892, "eos_token_id": [ 156892, 156895 ], "num_experts": 128, "num_shared_experts": 1, "num_experts_per_tok": 8, "n_group": 8, "topk_group": 4, "moe_intermediate_size": 512, "moe_shared_expert_intermediate_size": 512, "first_k_dense_replace": 1, "head_dim": 128, "output_router_logits": false, "use_qk_norm": true, "num_nextn_predict_layers": 0, "mtp_loss_scaling_factor": 0.1, "moe_router_enable_expert_bias": true, "routed_scaling_factor": 2.5, "moe_router_topk_scaling_factor": 2.5, "layer_group_size": 4, "kv_lora_rank": 512, "q_lora_rank": 256, "qk_rope_head_dim": 64, "v_head_dim": 128, "qk_nope_head_dim": 128, "rope_interleave": true, "score_function": "sigmoid", "scoring_func": "sigmoid", "seq_aux": true, "topk_method": "noaux_tc", "router_dtype": "fp32", "gated_attention_proj_granularity_type": "head_wise", "no_kda_lora": true, "kda_safe_gate": true, "kda_lower_bound": -5.0, "short_conv_kernel_size": 4, "moe_layer_freq": [ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ], "torch_dtype": "bfloat16", "norm_topk_prob": true, "use_expert_bias": true, "moe_z_loss_coeff": 2.9e-06, "rotary_percent": 0.5, "position_embedding_type": "rope" }, "audio_config": { "architectures": [ "AudioVAE" ], "sample_rate": 44100, "enc_kwargs": { "backbone": { "_attn_implementation": "flash_attention_2", "attention_dropout": 0.0, "hidden_act": "silu", "hidden_size": 896, "initializer_range": 0.02, "intermediate_size": 4864, "is_causal": true, "max_position_embeddings": 32768, "max_window_layers": 0, "model_type": "qwen2", "num_attention_heads": 14, "num_hidden_layers": 24, "num_key_value_heads": 2, "rms_norm_eps": 1e-06, "rope_theta": 1000000.0, "sliding_window": 64, "tie_word_embeddings": true, "torch_dtype": "bfloat16", "use_cache": false, "use_sliding_window": true, "vocab_size": 1 }, "hop_size": 882, "input_dim": 882, "latent_dim": 64 }, "dec_kwargs": { "backbone": { "_attn_implementation": "flash_attention_2", "attention_dropout": 0.0, "hidden_act": "silu", "hidden_size": 896, "initializer_range": 0.02, "intermediate_size": 4864, "is_causal": true, "max_position_embeddings": 32768, "max_window_layers": 0, "model_type": "qwen2", "num_attention_heads": 14, "num_hidden_layers": 24, "num_key_value_heads": 2, "rms_norm_eps": 1e-06, "rope_theta": 1000000.0, "sliding_window": 64, "tie_word_embeddings": true, "torch_dtype": "bfloat16", "use_cache": false, "use_sliding_window": true, "vocab_size": 1 }, "latent_dim": 64, "output_dim": 882 }, "semantic_module_kwargs": null, "init_method": "kaiming", "patch_size": 4, "torch_dtype": "bfloat16" }, "tts_config": { "protocol_version": "thinking_v1", "prompt_format": "legacy_antmmf", "tokenizer_contract": "legacy_antmmf", "sampling_contract": "native_fp32", "enable_tts_target_language": true, "infer_tts_language_from_text": true, "conditioning_mode": "speaker_only", "sampling_temperature": 1.5, "prompt_eos_token_id": 156895, "sample_rate": 44100, "speaker_embedding_dim": 192, "latent_dim": 64, "patch_size": 1, "history_size": 1, "audio_patch_token_id": 157164, "audio_start_token_id": 157165, "audio_end_token_id": 157166, "speaker_start_token_id": 157175, "speaker_end_token_id": 157176, "stop_threshold": 0.5, "minimum_decode_steps": 13, "maximum_decode_steps": 250, "decode_alignment": "old_shift", "task_contract": "lingv3_sft_multitask_v1", "task_decode_alignment": { "podcast": "old_shift" }, "with_rml": true }, "flow_config": { "z_channels": 64, "hidden_size": 1024, "depth": 16, "num_heads": 16, "mlp_ratio": 4, "dropout": 0.1, "qk_norm": null, "pe_attn_head": null, "attn_backend": "torch", "attn_mask_enabled": false }, "aggregator_config": { "in_channels": 64, "hidden_size": 1024, "depth": 8, "num_heads": 16, "mlp_ratio": 4, "dropout": 0.1, "qk_norm": null, "pe_attn_head": null, "attn_backend": "torch", "attn_mask_enabled": false }, "torch_dtype": "bfloat16" }