{ "_name_or_path": "facebook/wav2vec2-base-960h", "activation_dropout": 0.1, "adapter_attn_dim": null, "adapter_kernel_size": 3, "adapter_stride": 2, "add_adapter": false, "apply_spec_augment": true, "architectures": [ "Wav2Vec2ForSequenceClassification" ], "attention_dropout": 0.1, "bos_token_id": 1, "classifier_proj_size": 256, "codevector_dim": 256, "contrastive_logits_temperature": 0.1, "conv_bias": false, "conv_dim": [ 512, 512, 512, 512, 512, 512, 512 ], "conv_kernel": [ 10, 3, 3, 3, 3, 2, 2 ], "conv_stride": [ 5, 2, 2, 2, 2, 2, 2 ], "ctc_loss_reduction": "sum", "ctc_zero_infinity": false, "diversity_loss_weight": 0.1, "do_stable_layer_norm": false, "eos_token_id": 2, "feat_extract_activation": "gelu", "feat_extract_dropout": 0.0, "feat_extract_norm": "group", "feat_proj_dropout": 0.1, "feat_quantizer_dropout": 0.0, "final_dropout": 0.1, "gradient_checkpointing": false, "hidden_act": "gelu", "hidden_dropout": 0.1, "hidden_dropout_prob": 0.1, "hidden_size": 768, "id2label": { "0": "yoruba", "1": "igbo", "10": "isizulu", "100": "bassa", "101": "etche", "102": "kubi", "103": "jukun", "104": "igbo and yoruba", "105": "urobo", "106": "kalabari", "107": "ibani", "108": "obolo", "109": "idah", "11": "igala", "110": "bassa-nge/nupe", "111": "yala mbembe", "112": "eket", "113": "afo", "114": "ebiobo", "115": "nyandang", "116": "ishan", "117": "bagi", "118": "estako", "119": "gerawa", "12": "izon", "13": "kiswahili", "14": "ebira", "15": "luganda", "16": "urhobo", "17": "nembe", "18": "ibibio", "19": "pidgin", "2": "swahili", "20": "luhya", "21": "kinyarwanda", "22": "xhosa", "23": "tswana", "24": "esan", "25": "alago", "26": "tshivenda", "27": "fulani", "28": "isoko", "29": "akan (fante)", "3": "hausa", "30": "ikwere", "31": "sepedi", "32": "efik", "33": "edo", "34": "luo", "35": "kikuyu", "36": "bekwarra", "37": "isixhosa", "38": "hausa/fulani", "39": "epie", "4": "ijaw", "40": "isindebele", "41": "venda and xitsonga", "42": "sotho", "43": "akan", "44": "nupe", "45": "anaang", "46": "english", "47": "afemai", "48": "shona", "49": "eggon", "5": "afrikaans", "50": "luganda and kiswahili", "51": "ukwuani", "52": "sesotho", "53": "benin", "54": "kagoma", "55": "nasarawa eggon", "56": "tiv", "57": "south african english", "58": "borana", "59": "swahili ,luganda ,arabic", "6": "idoma", "60": "ogoni", "61": "mada", "62": "bette", "63": "berom", "64": "bini", "65": "ngas", "66": "etsako", "67": "okrika", "68": "venda", "69": "siswati", "7": "zulu", "70": "damara", "71": "yoruba, hausa", "72": "southern sotho", "73": "kanuri", "74": "itsekiri", "75": "ekpeye", "76": "mwaghavul", "77": "bajju", "78": "luo, swahili", "79": "dholuo", "8": "setswana", "80": "ekene", "81": "jaba", "82": "ika", "83": "angas", "84": "ateso", "85": "brass", "86": "ikulu", "87": "eleme", "88": "chichewa", "89": "oklo", "9": "twi", "90": "meru", "91": "agatu", "92": "okirika", "93": "igarra", "94": "ijaw(nembe)", "95": "khana", "96": "ogbia", "97": "gbagyi", "98": "portuguese", "99": "delta" }, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "afemai": 47, "afo": 113, "afrikaans": 5, "agatu": 91, "akan": 43, "akan (fante)": 29, "alago": 25, "anaang": 45, "angas": 83, "ateso": 84, "bagi": 117, "bajju": 77, "bassa": 100, "bassa-nge/nupe": 110, "bekwarra": 36, "benin": 53, "berom": 63, "bette": 62, "bini": 64, "borana": 58, "brass": 85, "chichewa": 88, "damara": 70, "delta": 99, "dholuo": 79, "ebiobo": 114, "ebira": 14, "edo": 33, "efik": 32, "eggon": 49, "ekene": 80, "eket": 112, "ekpeye": 75, "eleme": 87, "english": 46, "epie": 39, "esan": 24, "estako": 118, "etche": 101, "etsako": 66, "fulani": 27, "gbagyi": 97, "gerawa": 119, "hausa": 3, "hausa/fulani": 38, "ibani": 107, "ibibio": 18, "idah": 109, "idoma": 6, "igala": 11, "igarra": 93, "igbo": 1, "igbo and yoruba": 104, "ijaw": 4, "ijaw(nembe)": 94, "ika": 82, "ikulu": 86, "ikwere": 30, "ishan": 116, "isindebele": 40, "isixhosa": 37, "isizulu": 10, "isoko": 28, "itsekiri": 74, "izon": 12, "jaba": 81, "jukun": 103, "kagoma": 54, "kalabari": 106, "kanuri": 73, "khana": 95, "kikuyu": 35, "kinyarwanda": 21, "kiswahili": 13, "kubi": 102, "luganda": 15, "luganda and kiswahili": 50, "luhya": 20, "luo": 34, "luo, swahili": 78, "mada": 61, "meru": 90, "mwaghavul": 76, "nasarawa eggon": 55, "nembe": 17, "ngas": 65, "nupe": 44, "nyandang": 115, "obolo": 108, "ogbia": 96, "ogoni": 60, "okirika": 92, "oklo": 89, "okrika": 67, "pidgin": 19, "portuguese": 98, "sepedi": 31, "sesotho": 52, "setswana": 8, "shona": 48, "siswati": 69, "sotho": 42, "south african english": 57, "southern sotho": 72, "swahili": 2, "swahili ,luganda ,arabic": 59, "tiv": 56, "tshivenda": 26, "tswana": 23, "twi": 9, "ukwuani": 51, "urhobo": 16, "urobo": 105, "venda": 68, "venda and xitsonga": 41, "xhosa": 22, "yala mbembe": 111, "yoruba": 0, "yoruba, hausa": 71, "zulu": 7 }, "layer_norm_eps": 1e-05, "layerdrop": 0.1, "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, "model_type": "wav2vec2", "num_adapter_layers": 3, "num_attention_heads": 12, "num_codevector_groups": 2, "num_codevectors_per_group": 320, "num_conv_pos_embedding_groups": 16, "num_conv_pos_embeddings": 128, "num_feat_extract_layers": 7, "num_hidden_layers": 12, "num_negatives": 100, "output_hidden_size": 768, "pad_token_id": 0, "proj_codevector_dim": 256, "tdnn_dilation": [ 1, 2, 3, 1, 1 ], "tdnn_dim": [ 512, 512, 512, 512, 1500 ], "tdnn_kernel": [ 5, 3, 3, 1, 1 ], "torch_dtype": "float32", "transformers_version": "4.46.2", "use_weighted_layer_sum": false, "vocab_size": 32, "xvector_output_dim": 512 }