{ "architectures": [ "MiDashengLMModel" ], "audio_encoder_config": { "attn_drop_rate": 0.0, "center": true, "depth": 32, "drop_rate": 0.0, "embed_dim": 1280, "f_max": 8000.0, "f_min": 0.0, "hop_length": 160, "init_values": null, "input_channels": 1, "mlp_ratio": 4.0, "model_type": "midashenglm_dasheng_encoder", "n_fft": 512, "n_mels": 64, "num_heads": 16, "outputdim": 527, "patch_size": [ 64, 4 ], "patch_stride": [ 64, 4 ], "qkv_bias": true, "sample_rate": 16000, "target_length": 1008, "win_length": 512 }, "audio_token_id": 151646, "auto_map": { "AutoConfig": "configuration_midashenglm.MiDashengLMConfig", "AutoModelForCausalLM": "modeling_midashenglm.MiDashengLMModel" }, "model_type": "midashenglm", "quantization_config": { "config_groups": { "group_0": { "format": "pack-quantized", "input_activations": null, "output_activations": null, "targets": [ "Linear" ], "weights": { "actorder": null, "block_structure": null, "dynamic": false, "group_size": 128, "num_bits": 4, "observer": "minmax", "observer_kwargs": {}, "strategy": "group", "symmetric": true, "type": "int" } } }, "format": "pack-quantized", "global_compression_ratio": null, "ignore": [ "audio_encoder.blocks.0.attn.qkv", "audio_encoder.blocks.0.attn.proj", "audio_encoder.blocks.0.mlp.fc1", "audio_encoder.blocks.0.mlp.fc2", "audio_encoder.blocks.1.attn.qkv", "audio_encoder.blocks.1.attn.proj", "audio_encoder.blocks.1.mlp.fc1", "audio_encoder.blocks.1.mlp.fc2", "audio_encoder.blocks.2.attn.qkv", "audio_encoder.blocks.2.attn.proj", "audio_encoder.blocks.2.mlp.fc1", "audio_encoder.blocks.2.mlp.fc2", "audio_encoder.blocks.3.attn.qkv", "audio_encoder.blocks.3.attn.proj", "audio_encoder.blocks.3.mlp.fc1", "audio_encoder.blocks.3.mlp.fc2", "audio_encoder.blocks.4.attn.qkv", "audio_encoder.blocks.4.attn.proj", "audio_encoder.blocks.4.mlp.fc1", "audio_encoder.blocks.4.mlp.fc2", "audio_encoder.blocks.5.attn.qkv", "audio_encoder.blocks.5.attn.proj", "audio_encoder.blocks.5.mlp.fc1", "audio_encoder.blocks.5.mlp.fc2", "audio_encoder.blocks.6.attn.qkv", "audio_encoder.blocks.6.attn.proj", "audio_encoder.blocks.6.mlp.fc1", "audio_encoder.blocks.6.mlp.fc2", "audio_encoder.blocks.7.attn.qkv", "audio_encoder.blocks.7.attn.proj", "audio_encoder.blocks.7.mlp.fc1", "audio_encoder.blocks.7.mlp.fc2", "audio_encoder.blocks.8.attn.qkv", "audio_encoder.blocks.8.attn.proj", "audio_encoder.blocks.8.mlp.fc1", "audio_encoder.blocks.8.mlp.fc2", "audio_encoder.blocks.9.attn.qkv", "audio_encoder.blocks.9.attn.proj", "audio_encoder.blocks.9.mlp.fc1", "audio_encoder.blocks.9.mlp.fc2", "audio_encoder.blocks.10.attn.qkv", "audio_encoder.blocks.10.attn.proj", "audio_encoder.blocks.10.mlp.fc1", "audio_encoder.blocks.10.mlp.fc2", "audio_encoder.blocks.11.attn.qkv", "audio_encoder.blocks.11.attn.proj", "audio_encoder.blocks.11.mlp.fc1", "audio_encoder.blocks.11.mlp.fc2", "audio_encoder.blocks.12.attn.qkv", "audio_encoder.blocks.12.attn.proj", "audio_encoder.blocks.12.mlp.fc1", "audio_encoder.blocks.12.mlp.fc2", "audio_encoder.blocks.13.attn.qkv", "audio_encoder.blocks.13.attn.proj", "audio_encoder.blocks.13.mlp.fc1", "audio_encoder.blocks.13.mlp.fc2", "audio_encoder.blocks.14.attn.qkv", "audio_encoder.blocks.14.attn.proj", "audio_encoder.blocks.14.mlp.fc1", "audio_encoder.blocks.14.mlp.fc2", "audio_encoder.blocks.15.attn.qkv", "audio_encoder.blocks.15.attn.proj", "audio_encoder.blocks.15.mlp.fc1", "audio_encoder.blocks.15.mlp.fc2", "audio_encoder.blocks.16.attn.qkv", "audio_encoder.blocks.16.attn.proj", "audio_encoder.blocks.16.mlp.fc1", "audio_encoder.blocks.16.mlp.fc2", "audio_encoder.blocks.17.attn.qkv", "audio_encoder.blocks.17.attn.proj", "audio_encoder.blocks.17.mlp.fc1", "audio_encoder.blocks.17.mlp.fc2", "audio_encoder.blocks.18.attn.qkv", "audio_encoder.blocks.18.attn.proj", "audio_encoder.blocks.18.mlp.fc1", "audio_encoder.blocks.18.mlp.fc2", "audio_encoder.blocks.19.attn.qkv", "audio_encoder.blocks.19.attn.proj", "audio_encoder.blocks.19.mlp.fc1", "audio_encoder.blocks.19.mlp.fc2", "audio_encoder.blocks.20.attn.qkv", "audio_encoder.blocks.20.attn.proj", "audio_encoder.blocks.20.mlp.fc1", "audio_encoder.blocks.20.mlp.fc2", "audio_encoder.blocks.21.attn.qkv", "audio_encoder.blocks.21.attn.proj", "audio_encoder.blocks.21.mlp.fc1", "audio_encoder.blocks.21.mlp.fc2", "audio_encoder.blocks.22.attn.qkv", "audio_encoder.blocks.22.attn.proj", "audio_encoder.blocks.22.mlp.fc1", "audio_encoder.blocks.22.mlp.fc2", "audio_encoder.blocks.23.attn.qkv", "audio_encoder.blocks.23.attn.proj", "audio_encoder.blocks.23.mlp.fc1", "audio_encoder.blocks.23.mlp.fc2", "audio_encoder.blocks.24.attn.qkv", "audio_encoder.blocks.24.attn.proj", "audio_encoder.blocks.24.mlp.fc1", "audio_encoder.blocks.24.mlp.fc2", "audio_encoder.blocks.25.attn.qkv", "audio_encoder.blocks.25.attn.proj", "audio_encoder.blocks.25.mlp.fc1", "audio_encoder.blocks.25.mlp.fc2", "audio_encoder.blocks.26.attn.qkv", "audio_encoder.blocks.26.attn.proj", "audio_encoder.blocks.26.mlp.fc1", "audio_encoder.blocks.26.mlp.fc2", "audio_encoder.blocks.27.attn.qkv", "audio_encoder.blocks.27.attn.proj", "audio_encoder.blocks.27.mlp.fc1", "audio_encoder.blocks.27.mlp.fc2", "audio_encoder.blocks.28.attn.qkv", "audio_encoder.blocks.28.attn.proj", "audio_encoder.blocks.28.mlp.fc1", "audio_encoder.blocks.28.mlp.fc2", "audio_encoder.blocks.29.attn.qkv", "audio_encoder.blocks.29.attn.proj", "audio_encoder.blocks.29.mlp.fc1", "audio_encoder.blocks.29.mlp.fc2", "audio_encoder.blocks.30.attn.qkv", "audio_encoder.blocks.30.attn.proj", "audio_encoder.blocks.30.mlp.fc1", "audio_encoder.blocks.30.mlp.fc2", "audio_encoder.blocks.31.attn.qkv", "audio_encoder.blocks.31.attn.proj", "audio_encoder.blocks.31.mlp.fc1", "audio_encoder.blocks.31.mlp.fc2", "audio_projector.net.0", "audio_projector.net.2", "decoder.lm_head" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed", "sparsity_config": {}, "transform_config": {}, "version": "0.11.0" }, "subsample_factor": 5, "text_config": { "attention_dropout": 0.0, "hidden_act": "silu", "hidden_size": 3584, "init_std": 0.02, "initializer_range": 0.02, "intermediate_size": 18944, "layer_types": [ "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention" ], "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_5_omni_text", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": null, "use_cache": true, "use_sliding_window": false, "vocab_size": 152064 }, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.55.2" }