{ "audio_processor": { "audio_pool_step": 5, "auto_map": { "AutoFeatureExtractor": "processing_minicpmo.MiniCPMAAudioProcessor", "AutoImageProcessor": "processing_minicpmo.MiniCPMVImageProcessor", "AutoProcessor": "processing_minicpmo.MiniCPMOProcessor" }, "chunk_length": 30, "dither": 0.0, "dynamic_log_norm": true, "dynamic_range_db": 8.0, "feature_extractor_type": "MiniCPMAAudioProcessor", "feature_size": 80, "hop_length": 160, "im_end": "", "im_id_end": "", "im_id_start": "", "im_start": "", "image_feature_size": 64, "image_processor_type": "MiniCPMVImageProcessor", "log_floor_db": -10.0, "max_slice_nums": 9, "n_fft": 400, "n_samples": 480000, "nb_max_frames": 3000, "norm_mean": [ 0.5, 0.5, 0.5 ], "norm_std": [ 0.5, 0.5, 0.5 ], "padding_side": "right", "padding_value": 0.0, "patch_size": 14, "return_attention_mask": false, "sampling_rate": 16000, "scale_resolution": 448, "slice_end": "", "slice_mode": true, "slice_start": "", "unk": "", "use_image_id": true, "version": 4.5 }, "auto_map": { "AutoProcessor": "processing_minicpmo.MiniCPMOProcessor" }, "image_processor": { "audio_pool_step": 5, "auto_map": { "AutoFeatureExtractor": "processing_minicpmo.MiniCPMAAudioProcessor", "AutoImageProcessor": "processing_minicpmo.MiniCPMVImageProcessor", "AutoProcessor": "processing_minicpmo.MiniCPMOProcessor" }, "im_end": "", "im_end_token": "", "im_id_end": "", "im_id_start": "", "im_start": "", "im_start_token": "", "image_feature_size": 64, "image_processor_type": "MiniCPMVImageProcessor", "max_slice_nums": 9, "mean": [ 0.5, 0.5, 0.5 ], "norm_mean": [ 0.5, 0.5, 0.5 ], "norm_std": [ 0.5, 0.5, 0.5 ], "patch_size": 14, "scale_resolution": 448, "slice_end": "", "slice_end_token": "", "slice_mode": true, "slice_start": "", "slice_start_token": "", "std": [ 0.5, 0.5, 0.5 ], "unk": "", "unk_token": "", "use_image_id": true, "version": 4.5 }, "processor_class": "MiniCPMOProcessor" }