{ "architectures": [ "ASTForAudioClassification" ], "attention_probs_dropout_prob": 0.0, "dtype": "float32", "frequency_stride": 10, "hidden_act": "gelu", "hidden_dropout_prob": 0.0, "hidden_size": 768, "id2label": { "0": "0", "1": "1", "10": "brushing_teeth", "11": "can_opening", "12": "car_horn", "13": "cat", "14": "chainsaw", "15": "chirping_birds", "16": "church_bells", "17": "clapping", "18": "clock_alarm", "19": "clock_tick", "2": "2", "20": "coughing", "21": "cow", "22": "crackling_fire", "23": "crickets", "24": "crow", "25": "crying_baby", "26": "dog", "27": "door_wood_creaks", "28": "door_wood_knock", "29": "drinking_sipping", "3": "3", "30": "engine", "31": "fireworks", "32": "footsteps", "33": "frog", "34": "glass_breaking", "35": "hand_saw", "36": "helicopter", "37": "hen", "38": "insects", "39": "keyboard_typing", "4": "4", "40": "laughing", "41": "membo_1", "42": "mouse_click", "43": "pig", "44": "pouring_water", "45": "rain", "46": "rooster", "47": "sea_waves", "48": "sheep", "49": "siren", "5": "5", "50": "sneezing", "51": "snoring", "52": "thunderstorm", "53": "toilet_flush", "54": "train", "55": "vacuum_cleaner", "56": "washing_machine", "57": "water_drops", "58": "wind", "6": "6", "7": "airplane", "8": "bebop_1", "9": "breathing" }, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "0": "0", "1": "1", "2": "2", "3": "3", "4": "4", "5": "5", "6": "6", "airplane": "7", "bebop_1": "8", "breathing": "9", "brushing_teeth": "10", "can_opening": "11", "car_horn": "12", "cat": "13", "chainsaw": "14", "chirping_birds": "15", "church_bells": "16", "clapping": "17", "clock_alarm": "18", "clock_tick": "19", "coughing": "20", "cow": "21", "crackling_fire": "22", "crickets": "23", "crow": "24", "crying_baby": "25", "dog": "26", "door_wood_creaks": "27", "door_wood_knock": "28", "drinking_sipping": "29", "engine": "30", "fireworks": "31", "footsteps": "32", "frog": "33", "glass_breaking": "34", "hand_saw": "35", "helicopter": "36", "hen": "37", "insects": "38", "keyboard_typing": "39", "laughing": "40", "membo_1": "41", "mouse_click": "42", "pig": "43", "pouring_water": "44", "rain": "45", "rooster": "46", "sea_waves": "47", "sheep": "48", "siren": "49", "sneezing": "50", "snoring": "51", "thunderstorm": "52", "toilet_flush": "53", "train": "54", "vacuum_cleaner": "55", "washing_machine": "56", "water_drops": "57", "wind": "58" }, "layer_norm_eps": 1e-12, "max_length": 1024, "model_type": "audio-spectrogram-transformer", "num_attention_heads": 12, "num_hidden_layers": 12, "num_mel_bins": 128, "patch_size": 16, "problem_type": "single_label_classification", "qkv_bias": true, "time_stride": 10, "transformers_version": "5.10.2", "use_cache": false }