wav2vec2-ls100-sid / config.json
techsword's picture
Upload Wav2Vec2ForSequenceClassification
605fc43 verified
{
"_name_or_path": "/home/gshen/work_dir/unprobe/finetuned_models/wav2vec2-sid-finetuned/checkpoint-1300",
"activation_dropout": 0.0,
"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_norm": "group",
"feat_proj_dropout": 0.1,
"feat_quantizer_dropout": 0.0,
"final_dropout": 0.0,
"freeze_feat_extract_train": true,
"hidden_act": "gelu",
"hidden_dropout": 0.1,
"hidden_size": 768,
"id2label": {
"0": "103",
"1": "1034",
"2": "1040",
"3": "1069",
"4": "1081",
"5": "1088",
"6": "1098",
"7": "1116",
"8": "118",
"9": "1183",
"10": "1235",
"11": "1246",
"12": "125",
"13": "1263",
"14": "1334",
"15": "1355",
"16": "1363",
"17": "1447",
"18": "1455",
"19": "150",
"20": "1502",
"21": "1553",
"22": "1578",
"23": "1594",
"24": "1624",
"25": "163",
"26": "1723",
"27": "1737",
"28": "1743",
"29": "1841",
"30": "1867",
"31": "1898",
"32": "19",
"33": "1926",
"34": "196",
"35": "1963",
"36": "1970",
"37": "198",
"38": "1992",
"39": "200",
"40": "2002",
"41": "2007",
"42": "201",
"43": "2092",
"44": "211",
"45": "2136",
"46": "2159",
"47": "2182",
"48": "2196",
"49": "226",
"50": "2289",
"51": "229",
"52": "233",
"53": "2384",
"54": "2391",
"55": "2416",
"56": "2436",
"57": "248",
"58": "250",
"59": "2514",
"60": "2518",
"61": "254",
"62": "26",
"63": "2691",
"64": "27",
"65": "2764",
"66": "2817",
"67": "2836",
"68": "2843",
"69": "289",
"70": "2893",
"71": "2910",
"72": "2911",
"73": "2952",
"74": "298",
"75": "2989",
"76": "302",
"77": "307",
"78": "311",
"79": "3112",
"80": "3168",
"81": "32",
"82": "3214",
"83": "322",
"84": "3235",
"85": "3240",
"86": "3242",
"87": "3259",
"88": "328",
"89": "332",
"90": "3374",
"91": "3436",
"92": "3440",
"93": "3486",
"94": "3526",
"95": "3607",
"96": "3664",
"97": "3699",
"98": "3723",
"99": "374",
"100": "3807",
"101": "3830",
"102": "3857",
"103": "3879",
"104": "39",
"105": "3947",
"106": "3982",
"107": "3983",
"108": "40",
"109": "4014",
"110": "4018",
"111": "403",
"112": "405",
"113": "4051",
"114": "4088",
"115": "412",
"116": "4137",
"117": "4160",
"118": "4195",
"119": "4214",
"120": "426",
"121": "4267",
"122": "4297",
"123": "4340",
"124": "4362",
"125": "4397",
"126": "4406",
"127": "441",
"128": "4441",
"129": "445",
"130": "446",
"131": "4481",
"132": "458",
"133": "460",
"134": "4640",
"135": "4680",
"136": "4788",
"137": "481",
"138": "4813",
"139": "4830",
"140": "4853",
"141": "4859",
"142": "4898",
"143": "5022",
"144": "5049",
"145": "5104",
"146": "5163",
"147": "5192",
"148": "5322",
"149": "5339",
"150": "5390",
"151": "5393",
"152": "5456",
"153": "5463",
"154": "5514",
"155": "5561",
"156": "5652",
"157": "5678",
"158": "5688",
"159": "5703",
"160": "5750",
"161": "5778",
"162": "5789",
"163": "5808",
"164": "5867",
"165": "587",
"166": "60",
"167": "6000",
"168": "6019",
"169": "6064",
"170": "6078",
"171": "6081",
"172": "6147",
"173": "6181",
"174": "6209",
"175": "625",
"176": "6272",
"177": "6367",
"178": "6385",
"179": "6415",
"180": "6437",
"181": "6454",
"182": "6476",
"183": "6529",
"184": "6531",
"185": "6563",
"186": "669",
"187": "6818",
"188": "6836",
"189": "6848",
"190": "6880",
"191": "6925",
"192": "696",
"193": "7059",
"194": "7067",
"195": "7078",
"196": "7113",
"197": "7148",
"198": "7178",
"199": "7190",
"200": "7226",
"201": "7264",
"202": "7278",
"203": "730",
"204": "7302",
"205": "7312",
"206": "7367",
"207": "7402",
"208": "7447",
"209": "7505",
"210": "7511",
"211": "7517",
"212": "7635",
"213": "7780",
"214": "7794",
"215": "78",
"216": "7800",
"217": "7859",
"218": "8014",
"219": "8051",
"220": "8063",
"221": "8088",
"222": "8095",
"223": "8098",
"224": "8108",
"225": "8123",
"226": "8226",
"227": "8238",
"228": "83",
"229": "831",
"230": "8312",
"231": "8324",
"232": "839",
"233": "8419",
"234": "8425",
"235": "8465",
"236": "8468",
"237": "8580",
"238": "8609",
"239": "8629",
"240": "8630",
"241": "87",
"242": "8747",
"243": "8770",
"244": "8797",
"245": "8838",
"246": "887",
"247": "89",
"248": "8975",
"249": "909",
"250": "911"
},
"initializer_range": 0.02,
"intermediate_size": 3072,
"label2id": {
"103": 0,
"1034": 1,
"1040": 2,
"1069": 3,
"1081": 4,
"1088": 5,
"1098": 6,
"1116": 7,
"118": 8,
"1183": 9,
"1235": 10,
"1246": 11,
"125": 12,
"1263": 13,
"1334": 14,
"1355": 15,
"1363": 16,
"1447": 17,
"1455": 18,
"150": 19,
"1502": 20,
"1553": 21,
"1578": 22,
"1594": 23,
"1624": 24,
"163": 25,
"1723": 26,
"1737": 27,
"1743": 28,
"1841": 29,
"1867": 30,
"1898": 31,
"19": 32,
"1926": 33,
"196": 34,
"1963": 35,
"1970": 36,
"198": 37,
"1992": 38,
"200": 39,
"2002": 40,
"2007": 41,
"201": 42,
"2092": 43,
"211": 44,
"2136": 45,
"2159": 46,
"2182": 47,
"2196": 48,
"226": 49,
"2289": 50,
"229": 51,
"233": 52,
"2384": 53,
"2391": 54,
"2416": 55,
"2436": 56,
"248": 57,
"250": 58,
"2514": 59,
"2518": 60,
"254": 61,
"26": 62,
"2691": 63,
"27": 64,
"2764": 65,
"2817": 66,
"2836": 67,
"2843": 68,
"289": 69,
"2893": 70,
"2910": 71,
"2911": 72,
"2952": 73,
"298": 74,
"2989": 75,
"302": 76,
"307": 77,
"311": 78,
"3112": 79,
"3168": 80,
"32": 81,
"3214": 82,
"322": 83,
"3235": 84,
"3240": 85,
"3242": 86,
"3259": 87,
"328": 88,
"332": 89,
"3374": 90,
"3436": 91,
"3440": 92,
"3486": 93,
"3526": 94,
"3607": 95,
"3664": 96,
"3699": 97,
"3723": 98,
"374": 99,
"3807": 100,
"3830": 101,
"3857": 102,
"3879": 103,
"39": 104,
"3947": 105,
"3982": 106,
"3983": 107,
"40": 108,
"4014": 109,
"4018": 110,
"403": 111,
"405": 112,
"4051": 113,
"4088": 114,
"412": 115,
"4137": 116,
"4160": 117,
"4195": 118,
"4214": 119,
"426": 120,
"4267": 121,
"4297": 122,
"4340": 123,
"4362": 124,
"4397": 125,
"4406": 126,
"441": 127,
"4441": 128,
"445": 129,
"446": 130,
"4481": 131,
"458": 132,
"460": 133,
"4640": 134,
"4680": 135,
"4788": 136,
"481": 137,
"4813": 138,
"4830": 139,
"4853": 140,
"4859": 141,
"4898": 142,
"5022": 143,
"5049": 144,
"5104": 145,
"5163": 146,
"5192": 147,
"5322": 148,
"5339": 149,
"5390": 150,
"5393": 151,
"5456": 152,
"5463": 153,
"5514": 154,
"5561": 155,
"5652": 156,
"5678": 157,
"5688": 158,
"5703": 159,
"5750": 160,
"5778": 161,
"5789": 162,
"5808": 163,
"5867": 164,
"587": 165,
"60": 166,
"6000": 167,
"6019": 168,
"6064": 169,
"6078": 170,
"6081": 171,
"6147": 172,
"6181": 173,
"6209": 174,
"625": 175,
"6272": 176,
"6367": 177,
"6385": 178,
"6415": 179,
"6437": 180,
"6454": 181,
"6476": 182,
"6529": 183,
"6531": 184,
"6563": 185,
"669": 186,
"6818": 187,
"6836": 188,
"6848": 189,
"6880": 190,
"6925": 191,
"696": 192,
"7059": 193,
"7067": 194,
"7078": 195,
"7113": 196,
"7148": 197,
"7178": 198,
"7190": 199,
"7226": 200,
"7264": 201,
"7278": 202,
"730": 203,
"7302": 204,
"7312": 205,
"7367": 206,
"7402": 207,
"7447": 208,
"7505": 209,
"7511": 210,
"7517": 211,
"7635": 212,
"7780": 213,
"7794": 214,
"78": 215,
"7800": 216,
"7859": 217,
"8014": 218,
"8051": 219,
"8063": 220,
"8088": 221,
"8095": 222,
"8098": 223,
"8108": 224,
"8123": 225,
"8226": 226,
"8238": 227,
"83": 228,
"831": 229,
"8312": 230,
"8324": 231,
"839": 232,
"8419": 233,
"8425": 234,
"8465": 235,
"8468": 236,
"8580": 237,
"8609": 238,
"8629": 239,
"8630": 240,
"87": 241,
"8747": 242,
"8770": 243,
"8797": 244,
"8838": 245,
"887": 246,
"89": 247,
"8975": 248,
"909": 249,
"911": 250
},
"layer_norm_eps": 1e-05,
"layerdrop": 0.0,
"mask_channel_length": 10,
"mask_channel_min_space": 1,
"mask_channel_other": 0.0,
"mask_channel_prob": 0.0,
"mask_channel_selection": "static",
"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_min_space": 1,
"mask_time_other": 0.0,
"mask_time_prob": 0.05,
"mask_time_selection": "static",
"model_type": "wav2vec2",
"no_mask_channel_overlap": false,
"no_mask_time_overlap": false,
"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": "4.49.0",
"use_weighted_layer_sum": false,
"vocab_size": 32,
"xvector_output_dim": 512
}