CASIA-LM/DialectS2S
07
1{2 "adapter_inner_dim": 512,3 "architectures": [4 "OmniSpeechModel"5 ],6 "audio_encoder_config": {7 "_attn_implementation_autoset": true,8 "activation_dropout": 0.0,9 "activation_function": "gelu",10 "attention_dropout": 0.0,11 "d_model": 1280,12 "dropout": 0.0,13 "encoder_attention_heads": 20,14 "encoder_ffn_dim": 5120,15 "encoder_layerdrop": 0.0,16 "encoder_layers": 32,17 "init_std": 0.02,18 "max_source_positions": 1500,19 "model_type": "qwen2_audio_encoder",20 "num_hidden_layers": 32,21 "num_mel_bins": 128,22 "scale_embedding": false23 },24 "conv_kernel_sizes": "5,5",25 "interleave_strategy": "4:8",26 "llm_config": {27 "_name_or_path": "/public/zhangjiajun/PretrainModels/Qwen3-8B/",28 "architectures": [29 "Qwen3ForCausalLM"30 ],31 "attention_bias": false,32 "attention_dropout": 0.0,33 "bos_token_id": 151643,34 "eos_token_id": 151645,35 "head_dim": 128,36 "hidden_act": "silu",37 "hidden_size": 4096,38 "initializer_range": 0.02,39 "intermediate_size": 12288,40 "max_position_embeddings": 40960,41 "max_window_layers": 36,42 "model_type": "qwen3",43 "num_attention_heads": 32,44 "num_hidden_layers": 36,45 "num_key_value_heads": 8,46 "rms_norm_eps": 1e-06,47 "rope_scaling": null,48 "rope_theta": 1000000,49 "sliding_window": null,50 "torch_dtype": "bfloat16",51 "use_cache": true,52 "use_sliding_window": false,53 "vocab_size": 15167354 },55 "lora_config": null,56 "torch_dtype": "bfloat16",57 "transformers_version": "4.51.0",58 "tts_lm_config": {59 "_attn_implementation_autoset": true,60 "_name_or_path": "/public/zhangjiajun/cwang/omni_speech_v1/tts_interleave_checkpoints/",61 "architectures": [62 "Qwen3ForCausalLM"63 ],64 "attention_bias": false,65 "attention_dropout": 0.0,66 "bos_token_id": 151643,67 "eos_token_id": 151643,68 "head_dim": 128,69 "hidden_act": "silu",70 "hidden_size": 2048,71 "initializer_range": 0.02,72 "intermediate_size": 6144,73 "max_position_embeddings": 32768,74 "max_window_layers": 28,75 "model_type": "qwen3",76 "num_attention_heads": 16,77 "num_hidden_layers": 28,78 "num_key_value_heads": 8,79 "rms_norm_eps": 1e-06,80 "rope_scaling": null,81 "rope_theta": 1000000,82 "sliding_window": null,83 "tie_word_embeddings": true,84 "torch_dtype": "bfloat16",85 "use_cache": true,86 "use_sliding_window": false,87 "vocab_size": 16805588 }89}90 