{ "architectures": [ "MiniCPMO" ], "attention_bias": false, "attention_dropout": 0.0, "audio_chunk_length": 1.0, "audio_config": { "_attn_implementation_autoset": true, "_name_or_path": "openai/whisper-medium", "activation_dropout": 0.0, "activation_function": "gelu", "apply_spec_augment": false, "architectures": [ "MiniCPMWhisperEncoder" ], "attention_dropout": 0.0, "begin_suppress_tokens": [ 220, 50257 ], "bos_token_id": 50257, "classifier_proj_size": 256, "d_model": 1024, "decoder_attention_heads": 16, "decoder_ffn_dim": 4096, "decoder_layerdrop": 0.0, "decoder_layers": 24, "decoder_start_token_id": 50258, "dropout": 0.0, "dtype": "float32", "encoder_attention_heads": 16, "encoder_ffn_dim": 4096, "encoder_layerdrop": 0.0, "encoder_layers": 24, "eos_token_id": 50257, "forced_decoder_ids": [ [ 1, 50259 ], [ 2, 50359 ], [ 3, 50363 ] ], "init_std": 0.02, "mask_feature_length": 10, "mask_feature_min_masks": 0, "mask_feature_prob": 0.0, "mask_time_length": 10, "mask_time_min_masks": 2, "mask_time_prob": 0.05, "max_source_positions": 1500, "max_target_positions": 448, "median_filter_width": 7, "model_type": "whisper", "num_mel_bins": 80, "pad_token_id": 50257, "scale_embedding": false, "suppress_tokens": [ 1, 2, 7, 8, 9, 10, 14, 25, 26, 27, 28, 29, 31, 58, 59, 60, 61, 62, 63, 90, 91, 92, 93, 359, 503, 522, 542, 873, 893, 902, 918, 922, 931, 1350, 1853, 1982, 2460, 2627, 3246, 3253, 3268, 3536, 3846, 3961, 4183, 4667, 6585, 6647, 7273, 9061, 9383, 10428, 10929, 11938, 12033, 12331, 12562, 13793, 14157, 14635, 15265, 15618, 16553, 16604, 18362, 18956, 20075, 21675, 22520, 26130, 26161, 26435, 28279, 29464, 31650, 32302, 32470, 36865, 42863, 47425, 49870, 50254, 50258, 50358, 50359, 50360, 50361, 50362 ], "tie_word_embeddings": true, "use_cache": true, "use_weighted_layer_sum": false, "vocab_size": 51865 }, "audio_pool_step": 5, "auto_map": { "AutoConfig": "configuration_minicpmo.MiniCPMOConfig", "AutoModel": "modeling_minicpmo.MiniCPMO", "AutoModelForCausalLM": "modeling_minicpmo.MiniCPMO" }, "batch_vision_input": true, "bos_token_id": 151643, "drop_vision_last_layer": false, "dtype": "bfloat16", "eos_token_id": 151645, "head_dim": 128, "hidden_act": "silu", "hidden_size": 4096, "image_size": 448, "init_audio": true, "init_tts": true, "init_vision": true, "initializer_range": 0.02, "intermediate_size": 12288, "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", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention" ], "listen_speak_type": "asr", "max_position_embeddings": 40960, "max_window_layers": 36, "model_type": "minicpmo", "num_attention_heads": 32, "num_hidden_layers": 36, "num_key_value_heads": 8, "pad_token_id": null, "patch_size": 14, "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": 8, "observer": "memoryless_minmax", "observer_kwargs": {}, "scale_dtype": null, "strategy": "group", "symmetric": true, "type": "int", "zp_dtype": null } } }, "format": "pack-quantized", "global_compression_ratio": null, "ignore": [ "vpm.encoder.layers.0.self_attn.k_proj", "vpm.encoder.layers.0.self_attn.v_proj", "vpm.encoder.layers.0.self_attn.q_proj", "vpm.encoder.layers.0.self_attn.out_proj", "vpm.encoder.layers.0.mlp.fc1", "vpm.encoder.layers.0.mlp.fc2", "vpm.encoder.layers.1.self_attn.k_proj", "vpm.encoder.layers.1.self_attn.v_proj", "vpm.encoder.layers.1.self_attn.q_proj", "vpm.encoder.layers.1.self_attn.out_proj", "vpm.encoder.layers.1.mlp.fc1", "vpm.encoder.layers.1.mlp.fc2", "vpm.encoder.layers.2.self_attn.k_proj", "vpm.encoder.layers.2.self_attn.v_proj", "vpm.encoder.layers.2.self_attn.q_proj", "vpm.encoder.layers.2.self_attn.out_proj", "vpm.encoder.layers.2.mlp.fc1", "vpm.encoder.layers.2.mlp.fc2", "vpm.encoder.layers.3.self_attn.k_proj", "vpm.encoder.layers.3.self_attn.v_proj", "vpm.encoder.layers.3.self_attn.q_proj", "vpm.encoder.layers.3.self_attn.out_proj", "vpm.encoder.layers.3.mlp.fc1", "vpm.encoder.layers.3.mlp.fc2", "vpm.encoder.layers.4.self_attn.k_proj", "vpm.encoder.layers.4.self_attn.v_proj", "vpm.encoder.layers.4.self_attn.q_proj", "vpm.encoder.layers.4.self_attn.out_proj", "vpm.encoder.layers.4.mlp.fc1", "vpm.encoder.layers.4.mlp.fc2", "vpm.encoder.layers.5.self_attn.k_proj", "vpm.encoder.layers.5.self_attn.v_proj", "vpm.encoder.layers.5.self_attn.q_proj", "vpm.encoder.layers.5.self_attn.out_proj", "vpm.encoder.layers.5.mlp.fc1", "vpm.encoder.layers.5.mlp.fc2", "vpm.encoder.layers.6.self_attn.k_proj", "vpm.encoder.layers.6.self_attn.v_proj", "vpm.encoder.layers.6.self_attn.q_proj", "vpm.encoder.layers.6.self_attn.out_proj", "vpm.encoder.layers.6.mlp.fc1", "vpm.encoder.layers.6.mlp.fc2", "vpm.encoder.layers.7.self_attn.k_proj", "vpm.encoder.layers.7.self_attn.v_proj", "vpm.encoder.layers.7.self_attn.q_proj", "vpm.encoder.layers.7.self_attn.out_proj", "vpm.encoder.layers.7.mlp.fc1", "vpm.encoder.layers.7.mlp.fc2", "vpm.encoder.layers.8.self_attn.k_proj", "vpm.encoder.layers.8.self_attn.v_proj", "vpm.encoder.layers.8.self_attn.q_proj", "vpm.encoder.layers.8.self_attn.out_proj", "vpm.encoder.layers.8.mlp.fc1", "vpm.encoder.layers.8.mlp.fc2", "vpm.encoder.layers.9.self_attn.k_proj", "vpm.encoder.layers.9.self_attn.v_proj", "vpm.encoder.layers.9.self_attn.q_proj", "vpm.encoder.layers.9.self_attn.out_proj", "vpm.encoder.layers.9.mlp.fc1", "vpm.encoder.layers.9.mlp.fc2", "vpm.encoder.layers.10.self_attn.k_proj", "vpm.encoder.layers.10.self_attn.v_proj", "vpm.encoder.layers.10.self_attn.q_proj", "vpm.encoder.layers.10.self_attn.out_proj", "vpm.encoder.layers.10.mlp.fc1", "vpm.encoder.layers.10.mlp.fc2", "vpm.encoder.layers.11.self_attn.k_proj", "vpm.encoder.layers.11.self_attn.v_proj", "vpm.encoder.layers.11.self_attn.q_proj", "vpm.encoder.layers.11.self_attn.out_proj", "vpm.encoder.layers.11.mlp.fc1", "vpm.encoder.layers.11.mlp.fc2", "vpm.encoder.layers.12.self_attn.k_proj", "vpm.encoder.layers.12.self_attn.v_proj", "vpm.encoder.layers.12.self_attn.q_proj", "vpm.encoder.layers.12.self_attn.out_proj", "vpm.encoder.layers.12.mlp.fc1", "vpm.encoder.layers.12.mlp.fc2", "vpm.encoder.layers.13.self_attn.k_proj", "vpm.encoder.layers.13.self_attn.v_proj", "vpm.encoder.layers.13.self_attn.q_proj", "vpm.encoder.layers.13.self_attn.out_proj", "vpm.encoder.layers.13.mlp.fc1", "vpm.encoder.layers.13.mlp.fc2", "vpm.encoder.layers.14.self_attn.k_proj", "vpm.encoder.layers.14.self_attn.v_proj", "vpm.encoder.layers.14.self_attn.q_proj", "vpm.encoder.layers.14.self_attn.out_proj", "vpm.encoder.layers.14.mlp.fc1", "vpm.encoder.layers.14.mlp.fc2", "vpm.encoder.layers.15.self_attn.k_proj", "vpm.encoder.layers.15.self_attn.v_proj", "vpm.encoder.layers.15.self_attn.q_proj", "vpm.encoder.layers.15.self_attn.out_proj", "vpm.encoder.layers.15.mlp.fc1", "vpm.encoder.layers.15.mlp.fc2", "vpm.encoder.layers.16.self_attn.k_proj", "vpm.encoder.layers.16.self_attn.v_proj", "vpm.encoder.layers.16.self_attn.q_proj", "vpm.encoder.layers.16.self_attn.out_proj", "vpm.encoder.layers.16.mlp.fc1", "vpm.encoder.layers.16.mlp.fc2", "vpm.encoder.layers.17.self_attn.k_proj", "vpm.encoder.layers.17.self_attn.v_proj", "vpm.encoder.layers.17.self_attn.q_proj", "vpm.encoder.layers.17.self_attn.out_proj", "vpm.encoder.layers.17.mlp.fc1", "vpm.encoder.layers.17.mlp.fc2", "vpm.encoder.layers.18.self_attn.k_proj", "vpm.encoder.layers.18.self_attn.v_proj", "vpm.encoder.layers.18.self_attn.q_proj", "vpm.encoder.layers.18.self_attn.out_proj", "vpm.encoder.layers.18.mlp.fc1", "vpm.encoder.layers.18.mlp.fc2", "vpm.encoder.layers.19.self_attn.k_proj", "vpm.encoder.layers.19.self_attn.v_proj", "vpm.encoder.layers.19.self_attn.q_proj", "vpm.encoder.layers.19.self_attn.out_proj", "vpm.encoder.layers.19.mlp.fc1", "vpm.encoder.layers.19.mlp.fc2", "vpm.encoder.layers.20.self_attn.k_proj", "vpm.encoder.layers.20.self_attn.v_proj", "vpm.encoder.layers.20.self_attn.q_proj", "vpm.encoder.layers.20.self_attn.out_proj", "vpm.encoder.layers.20.mlp.fc1", "vpm.encoder.layers.20.mlp.fc2", "vpm.encoder.layers.21.self_attn.k_proj", "vpm.encoder.layers.21.self_attn.v_proj", "vpm.encoder.layers.21.self_attn.q_proj", "vpm.encoder.layers.21.self_attn.out_proj", "vpm.encoder.layers.21.mlp.fc1", "vpm.encoder.layers.21.mlp.fc2", "vpm.encoder.layers.22.self_attn.k_proj", "vpm.encoder.layers.22.self_attn.v_proj", "vpm.encoder.layers.22.self_attn.q_proj", "vpm.encoder.layers.22.self_attn.out_proj", "vpm.encoder.layers.22.mlp.fc1", "vpm.encoder.layers.22.mlp.fc2", "vpm.encoder.layers.23.self_attn.k_proj", "vpm.encoder.layers.23.self_attn.v_proj", "vpm.encoder.layers.23.self_attn.q_proj", "vpm.encoder.layers.23.self_attn.out_proj", "vpm.encoder.layers.23.mlp.fc1", "vpm.encoder.layers.23.mlp.fc2", "vpm.encoder.layers.24.self_attn.k_proj", "vpm.encoder.layers.24.self_attn.v_proj", "vpm.encoder.layers.24.self_attn.q_proj", "vpm.encoder.layers.24.self_attn.out_proj", "vpm.encoder.layers.24.mlp.fc1", "vpm.encoder.layers.24.mlp.fc2", "vpm.encoder.layers.25.self_attn.k_proj", "vpm.encoder.layers.25.self_attn.v_proj", "vpm.encoder.layers.25.self_attn.q_proj", "vpm.encoder.layers.25.self_attn.out_proj", "vpm.encoder.layers.25.mlp.fc1", "vpm.encoder.layers.25.mlp.fc2", "vpm.encoder.layers.26.self_attn.k_proj", "vpm.encoder.layers.26.self_attn.v_proj", "vpm.encoder.layers.26.self_attn.q_proj", "vpm.encoder.layers.26.self_attn.out_proj", "vpm.encoder.layers.26.mlp.fc1", "vpm.encoder.layers.26.mlp.fc2", "resampler.kv_proj", "apm.layers.0.self_attn.k_proj", "apm.layers.0.self_attn.v_proj", "apm.layers.0.self_attn.q_proj", "apm.layers.0.self_attn.out_proj", "apm.layers.0.fc1", "apm.layers.0.fc2", "apm.layers.1.self_attn.k_proj", "apm.layers.1.self_attn.v_proj", "apm.layers.1.self_attn.q_proj", "apm.layers.1.self_attn.out_proj", "apm.layers.1.fc1", "apm.layers.1.fc2", "apm.layers.2.self_attn.k_proj", "apm.layers.2.self_attn.v_proj", "apm.layers.2.self_attn.q_proj", "apm.layers.2.self_attn.out_proj", "apm.layers.2.fc1", "apm.layers.2.fc2", "apm.layers.3.self_attn.k_proj", "apm.layers.3.self_attn.v_proj", "apm.layers.3.self_attn.q_proj", "apm.layers.3.self_attn.out_proj", "apm.layers.3.fc1", "apm.layers.3.fc2", "apm.layers.4.self_attn.k_proj", "apm.layers.4.self_attn.v_proj", "apm.layers.4.self_attn.q_proj", "apm.layers.4.self_attn.out_proj", "apm.layers.4.fc1", "apm.layers.4.fc2", "apm.layers.5.self_attn.k_proj", "apm.layers.5.self_attn.v_proj", "apm.layers.5.self_attn.q_proj", "apm.layers.5.self_attn.out_proj", "apm.layers.5.fc1", "apm.layers.5.fc2", "apm.layers.6.self_attn.k_proj", "apm.layers.6.self_attn.v_proj", "apm.layers.6.self_attn.q_proj", "apm.layers.6.self_attn.out_proj", "apm.layers.6.fc1", "apm.layers.6.fc2", "apm.layers.7.self_attn.k_proj", "apm.layers.7.self_attn.v_proj", "apm.layers.7.self_attn.q_proj", "apm.layers.7.self_attn.out_proj", "apm.layers.7.fc1", "apm.layers.7.fc2", "apm.layers.8.self_attn.k_proj", "apm.layers.8.self_attn.v_proj", "apm.layers.8.self_attn.q_proj", "apm.layers.8.self_attn.out_proj", "apm.layers.8.fc1", "apm.layers.8.fc2", "apm.layers.9.self_attn.k_proj", "apm.layers.9.self_attn.v_proj", "apm.layers.9.self_attn.q_proj", "apm.layers.9.self_attn.out_proj", "apm.layers.9.fc1", "apm.layers.9.fc2", "apm.layers.10.self_attn.k_proj", "apm.layers.10.self_attn.v_proj", "apm.layers.10.self_attn.q_proj", "apm.layers.10.self_attn.out_proj", "apm.layers.10.fc1", "apm.layers.10.fc2", "apm.layers.11.self_attn.k_proj", "apm.layers.11.self_attn.v_proj", "apm.layers.11.self_attn.q_proj", "apm.layers.11.self_attn.out_proj", "apm.layers.11.fc1", "apm.layers.11.fc2", "apm.layers.12.self_attn.k_proj", "apm.layers.12.self_attn.v_proj", "apm.layers.12.self_attn.q_proj", "apm.layers.12.self_attn.out_proj", "apm.layers.12.fc1", "apm.layers.12.fc2", "apm.layers.13.self_attn.k_proj", "apm.layers.13.self_attn.v_proj", "apm.layers.13.self_attn.q_proj", "apm.layers.13.self_attn.out_proj", "apm.layers.13.fc1", "apm.layers.13.fc2", "apm.layers.14.self_attn.k_proj", "apm.layers.14.self_attn.v_proj", "apm.layers.14.self_attn.q_proj", "apm.layers.14.self_attn.out_proj", "apm.layers.14.fc1", "apm.layers.14.fc2", "apm.layers.15.self_attn.k_proj", "apm.layers.15.self_attn.v_proj", "apm.layers.15.self_attn.q_proj", "apm.layers.15.self_attn.out_proj", "apm.layers.15.fc1", "apm.layers.15.fc2", "apm.layers.16.self_attn.k_proj", "apm.layers.16.self_attn.v_proj", "apm.layers.16.self_attn.q_proj", "apm.layers.16.self_attn.out_proj", "apm.layers.16.fc1", "apm.layers.16.fc2", "apm.layers.17.self_attn.k_proj", "apm.layers.17.self_attn.v_proj", "apm.layers.17.self_attn.q_proj", "apm.layers.17.self_attn.out_proj", "apm.layers.17.fc1", "apm.layers.17.fc2", "apm.layers.18.self_attn.k_proj", "apm.layers.18.self_attn.v_proj", "apm.layers.18.self_attn.q_proj", "apm.layers.18.self_attn.out_proj", "apm.layers.18.fc1", "apm.layers.18.fc2", "apm.layers.19.self_attn.k_proj", "apm.layers.19.self_attn.v_proj", "apm.layers.19.self_attn.q_proj", "apm.layers.19.self_attn.out_proj", "apm.layers.19.fc1", "apm.layers.19.fc2", "apm.layers.20.self_attn.k_proj", "apm.layers.20.self_attn.v_proj", "apm.layers.20.self_attn.q_proj", "apm.layers.20.self_attn.out_proj", "apm.layers.20.fc1", "apm.layers.20.fc2", "apm.layers.21.self_attn.k_proj", "apm.layers.21.self_attn.v_proj", "apm.layers.21.self_attn.q_proj", "apm.layers.21.self_attn.out_proj", "apm.layers.21.fc1", "apm.layers.21.fc2", "apm.layers.22.self_attn.k_proj", "apm.layers.22.self_attn.v_proj", "apm.layers.22.self_attn.q_proj", "apm.layers.22.self_attn.out_proj", "apm.layers.22.fc1", "apm.layers.22.fc2", "apm.layers.23.self_attn.k_proj", "apm.layers.23.self_attn.v_proj", "apm.layers.23.self_attn.q_proj", "apm.layers.23.self_attn.out_proj", "apm.layers.23.fc1", "apm.layers.23.fc2", "audio_projection_layer.linear1", "audio_projection_layer.linear2", "tts.model.layers.0.self_attn.q_proj", "tts.model.layers.0.self_attn.k_proj", "tts.model.layers.0.self_attn.v_proj", "tts.model.layers.0.self_attn.o_proj", "tts.model.layers.0.mlp.gate_proj", "tts.model.layers.0.mlp.up_proj", "tts.model.layers.0.mlp.down_proj", "tts.model.layers.1.self_attn.q_proj", "tts.model.layers.1.self_attn.k_proj", "tts.model.layers.1.self_attn.v_proj", "tts.model.layers.1.self_attn.o_proj", "tts.model.layers.1.mlp.gate_proj", "tts.model.layers.1.mlp.up_proj", "tts.model.layers.1.mlp.down_proj", "tts.model.layers.2.self_attn.q_proj", "tts.model.layers.2.self_attn.k_proj", "tts.model.layers.2.self_attn.v_proj", "tts.model.layers.2.self_attn.o_proj", "tts.model.layers.2.mlp.gate_proj", "tts.model.layers.2.mlp.up_proj", "tts.model.layers.2.mlp.down_proj", "tts.model.layers.3.self_attn.q_proj", "tts.model.layers.3.self_attn.k_proj", "tts.model.layers.3.self_attn.v_proj", "tts.model.layers.3.self_attn.o_proj", "tts.model.layers.3.mlp.gate_proj", "tts.model.layers.3.mlp.up_proj", "tts.model.layers.3.mlp.down_proj", "tts.model.layers.4.self_attn.q_proj", "tts.model.layers.4.self_attn.k_proj", "tts.model.layers.4.self_attn.v_proj", "tts.model.layers.4.self_attn.o_proj", "tts.model.layers.4.mlp.gate_proj", "tts.model.layers.4.mlp.up_proj", "tts.model.layers.4.mlp.down_proj", "tts.model.layers.5.self_attn.q_proj", "tts.model.layers.5.self_attn.k_proj", "tts.model.layers.5.self_attn.v_proj", "tts.model.layers.5.self_attn.o_proj", "tts.model.layers.5.mlp.gate_proj", "tts.model.layers.5.mlp.up_proj", "tts.model.layers.5.mlp.down_proj", "tts.model.layers.6.self_attn.q_proj", "tts.model.layers.6.self_attn.k_proj", "tts.model.layers.6.self_attn.v_proj", "tts.model.layers.6.self_attn.o_proj", "tts.model.layers.6.mlp.gate_proj", "tts.model.layers.6.mlp.up_proj", "tts.model.layers.6.mlp.down_proj", "tts.model.layers.7.self_attn.q_proj", "tts.model.layers.7.self_attn.k_proj", "tts.model.layers.7.self_attn.v_proj", "tts.model.layers.7.self_attn.o_proj", "tts.model.layers.7.mlp.gate_proj", "tts.model.layers.7.mlp.up_proj", "tts.model.layers.7.mlp.down_proj", "tts.model.layers.8.self_attn.q_proj", "tts.model.layers.8.self_attn.k_proj", "tts.model.layers.8.self_attn.v_proj", "tts.model.layers.8.self_attn.o_proj", "tts.model.layers.8.mlp.gate_proj", "tts.model.layers.8.mlp.up_proj", "tts.model.layers.8.mlp.down_proj", "tts.model.layers.9.self_attn.q_proj", "tts.model.layers.9.self_attn.k_proj", "tts.model.layers.9.self_attn.v_proj", "tts.model.layers.9.self_attn.o_proj", "tts.model.layers.9.mlp.gate_proj", "tts.model.layers.9.mlp.up_proj", "tts.model.layers.9.mlp.down_proj", "tts.model.layers.10.self_attn.q_proj", "tts.model.layers.10.self_attn.k_proj", "tts.model.layers.10.self_attn.v_proj", "tts.model.layers.10.self_attn.o_proj", "tts.model.layers.10.mlp.gate_proj", "tts.model.layers.10.mlp.up_proj", "tts.model.layers.10.mlp.down_proj", "tts.model.layers.11.self_attn.q_proj", "tts.model.layers.11.self_attn.k_proj", "tts.model.layers.11.self_attn.v_proj", "tts.model.layers.11.self_attn.o_proj", "tts.model.layers.11.mlp.gate_proj", "tts.model.layers.11.mlp.up_proj", "tts.model.layers.11.mlp.down_proj", "tts.model.layers.12.self_attn.q_proj", "tts.model.layers.12.self_attn.k_proj", "tts.model.layers.12.self_attn.v_proj", "tts.model.layers.12.self_attn.o_proj", "tts.model.layers.12.mlp.gate_proj", "tts.model.layers.12.mlp.up_proj", "tts.model.layers.12.mlp.down_proj", "tts.model.layers.13.self_attn.q_proj", "tts.model.layers.13.self_attn.k_proj", "tts.model.layers.13.self_attn.v_proj", "tts.model.layers.13.self_attn.o_proj", "tts.model.layers.13.mlp.gate_proj", "tts.model.layers.13.mlp.up_proj", "tts.model.layers.13.mlp.down_proj", "tts.model.layers.14.self_attn.q_proj", "tts.model.layers.14.self_attn.k_proj", "tts.model.layers.14.self_attn.v_proj", "tts.model.layers.14.self_attn.o_proj", "tts.model.layers.14.mlp.gate_proj", "tts.model.layers.14.mlp.up_proj", "tts.model.layers.14.mlp.down_proj", "tts.model.layers.15.self_attn.q_proj", "tts.model.layers.15.self_attn.k_proj", "tts.model.layers.15.self_attn.v_proj", "tts.model.layers.15.self_attn.o_proj", "tts.model.layers.15.mlp.gate_proj", "tts.model.layers.15.mlp.up_proj", "tts.model.layers.15.mlp.down_proj", "tts.model.layers.16.self_attn.q_proj", "tts.model.layers.16.self_attn.k_proj", "tts.model.layers.16.self_attn.v_proj", "tts.model.layers.16.self_attn.o_proj", "tts.model.layers.16.mlp.gate_proj", "tts.model.layers.16.mlp.up_proj", "tts.model.layers.16.mlp.down_proj", "tts.model.layers.17.self_attn.q_proj", "tts.model.layers.17.self_attn.k_proj", "tts.model.layers.17.self_attn.v_proj", "tts.model.layers.17.self_attn.o_proj", "tts.model.layers.17.mlp.gate_proj", "tts.model.layers.17.mlp.up_proj", "tts.model.layers.17.mlp.down_proj", "tts.model.layers.18.self_attn.q_proj", "tts.model.layers.18.self_attn.k_proj", "tts.model.layers.18.self_attn.v_proj", "tts.model.layers.18.self_attn.o_proj", "tts.model.layers.18.mlp.gate_proj", "tts.model.layers.18.mlp.up_proj", "tts.model.layers.18.mlp.down_proj", "tts.model.layers.19.self_attn.q_proj", "tts.model.layers.19.self_attn.k_proj", "tts.model.layers.19.self_attn.v_proj", "tts.model.layers.19.self_attn.o_proj", "tts.model.layers.19.mlp.gate_proj", "tts.model.layers.19.mlp.up_proj", "tts.model.layers.19.mlp.down_proj", "tts.projector_spk.linear1", "tts.projector_spk.linear2", "tts.projector_semantic.linear1", "tts.projector_semantic.linear2" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed", "sparsity_config": {}, "transform_config": {}, "version": "0.15.1.a20260521" }, "query_num": 64, "rms_norm_eps": 1e-06, "rope_parameters": { "rope_theta": 1000000, "rope_type": "default" }, "slice_config": { "max_slice_nums": 1, "model_type": "minicpmv", "patch_size": 14, "scale_resolution": 448 }, "slice_mode": true, "sliding_window": null, "stream_input": true, "tie_word_embeddings": false, "transformers_version": "5.8.1", "tts_config": { "_attn_implementation_autoset": true, "attention_type": "full_attention", "attn_implementation": "eager", "audio_bos_token_id": 151687, "audio_tokenizer_sample_rate": 16000, "audio_tokenizer_type": "s3tokenizer", "aug_layer_loss_weight": false, "aug_loss_weight": false, "backbone_model": "llama", "condition_type": "hidden_text_merge", "cosyvoice_config_path": null, "cosyvoice_model_dir": null, "filter_tts_loss": false, "hidden_act": "silu", "hidden_size": 768, "interleaved": false, "intermediate_size": 3072, "llm_dim": 4096, "llm_dim_model_base": 256, "llm_down_scale": false, "llm_hidden_size": 4096, "llm_intermediate_size": 768, "long_weight": 0.1, "max_position_embeddings": 4096, "model_type": "minicpmtts", "normalize_projected_hidden": true, "num_attention_heads": 12, "num_audio_tokens": 6562, "num_hidden_layers": 20, "num_key_value_heads": 12, "num_mel_bins": 100, "num_text_tokens": 152064, "num_vq": 1, "projector_type": "mlp", "recomputed_chunks": 1, "s3_stream_chunk_size": 25, "s3_stream_generate": false, "s3_stream_n_timesteps": 10, "s3_stream_prelook_size": 3, "short_weight": 0.1, "streaming": false, "streaming_audio_chunk_size": 50, "streaming_sliding_window": false, "streaming_sliding_window_audio_frame_rate": 50, "streaming_sliding_window_audio_init_text_length": 10, "streaming_sliding_window_audio_window_size": 300, "streaming_sliding_window_average_speed": 5, "streaming_sliding_window_fast_speed": 7, "streaming_sliding_window_max_text_len": 500, "streaming_sliding_window_slow_speed": 3, "streaming_sliding_window_text_window_size": 50, "streaming_text_chunk_max": 7, "streaming_text_chunk_min": 3, "streaming_text_reserved_len": 300, "text_eos_token_id": 151692, "tts_filter_loss_fix": false, "use_llm_hidden_state": false, "use_text": true, "window_size": 2 }, "use_cache": true, "use_image_id": true, "use_sliding_window": false, "version": "4.5", "vision_batch_size": 16, "vision_config": { "_attn_implementation_autoset": true, "attention_dropout": 0.0, "hidden_act": "gelu_pytorch_tanh", "hidden_size": 1152, "image_size": 980, "intermediate_size": 4304, "layer_norm_eps": 1e-06, "model_type": "siglip_vision_model", "num_attention_heads": 16, "num_channels": 3, "num_hidden_layers": 27, "patch_size": 14 }, "vocab_size": 151748 }