parakeet-tdt-0.6b-v2 / config.json
senstella's picture
Upload folder using huggingface_hub
b8e276d verified
{
"sample_rate": 16000,
"compute_eval_loss": false,
"log_prediction": true,
"rnnt_reduction": "mean_volume",
"skip_nan_grad": false,
"model_defaults": {
"enc_hidden": 1024,
"pred_hidden": 640,
"joint_hidden": 640,
"tdt_durations": [
0,
1,
2,
3,
4
],
"num_tdt_durations": 5
},
"train_ds": {
"use_lhotse": true,
"skip_missing_manifest_entries": true,
"input_cfg": null,
"tarred_audio_filepaths": null,
"manifest_filepath": null,
"sample_rate": 16000,
"shuffle": true,
"num_workers": 2,
"pin_memory": true,
"max_duration": 40.0,
"min_duration": 0.1,
"text_field": "answer",
"batch_duration": null,
"use_bucketing": true,
"bucket_duration_bins": null,
"bucket_batch_size": null,
"num_buckets": 30,
"bucket_buffer_size": 20000,
"shuffle_buffer_size": 10000
},
"validation_ds": {
"use_lhotse": true,
"manifest_filepath": null,
"sample_rate": 16000,
"batch_size": 16,
"shuffle": false,
"max_duration": 40.0,
"min_duration": 0.1,
"num_workers": 2,
"pin_memory": true,
"text_field": "answer"
},
"tokenizer": {
"dir": "/lustre/fsw/portfolios/llmservice/users/nkoluguri/datasets/tokenizers/en_mm_ytc/tokenizer_spe_bpe_v1024/",
"type": "bpe",
"model_path": "nemo:705f11d22dc04b169effc35ce5cd1361_tokenizer.model",
"vocab_path": "nemo:4cf78c8ca4ca44fca36c3754478fb188_vocab.txt",
"spe_tokenizer_vocab": "nemo:a4715c7f6b2d4c2bb709306073d0c0a4_tokenizer.vocab"
},
"preprocessor": {
"_target_": "nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor",
"sample_rate": 16000,
"normalize": "per_feature",
"window_size": 0.025,
"window_stride": 0.01,
"window": "hann",
"features": 128,
"n_fft": 512,
"log": true,
"frame_splicing": 1,
"dither": 1e-05,
"pad_to": 0,
"pad_value": 0.0
},
"spec_augment": {
"_target_": "nemo.collections.asr.modules.SpectrogramAugmentation",
"freq_masks": 2,
"time_masks": 10,
"freq_width": 27,
"time_width": 0.05
},
"encoder": {
"_target_": "nemo.collections.asr.modules.ConformerEncoder",
"feat_in": 128,
"feat_out": -1,
"n_layers": 24,
"d_model": 1024,
"use_bias": false,
"subsampling": "dw_striding",
"subsampling_factor": 8,
"subsampling_conv_channels": 256,
"causal_downsampling": false,
"reduction": null,
"reduction_position": null,
"reduction_factor": 1,
"ff_expansion_factor": 4,
"self_attention_model": "rel_pos",
"n_heads": 8,
"att_context_size": [
-1,
-1
],
"att_context_style": "regular",
"xscaling": false,
"untie_biases": true,
"pos_emb_max_len": 5000,
"conv_kernel_size": 9,
"conv_norm_type": "batch_norm",
"conv_context_size": null,
"dropout": 0.1,
"dropout_pre_encoder": 0.1,
"dropout_emb": 0.0,
"dropout_att": 0.1,
"stochastic_depth_drop_prob": 0.0,
"stochastic_depth_mode": "linear",
"stochastic_depth_start_layer": 1
},
"decoder": {
"_target_": "nemo.collections.asr.modules.RNNTDecoder",
"normalization_mode": null,
"random_state_sampling": false,
"blank_as_pad": true,
"prednet": {
"pred_hidden": 640,
"pred_rnn_layers": 2,
"t_max": null,
"dropout": 0.2
},
"vocab_size": 1024
},
"joint": {
"_target_": "nemo.collections.asr.modules.RNNTJoint",
"log_softmax": null,
"preserve_memory": false,
"fuse_loss_wer": true,
"fused_batch_size": 4,
"jointnet": {
"joint_hidden": 640,
"activation": "relu",
"dropout": 0.2,
"encoder_hidden": 1024,
"pred_hidden": 640
},
"num_extra_outputs": 5,
"num_classes": 1024,
"vocabulary": [
"<unk>",
"\u2581t",
"\u2581th",
"\u2581a",
"in",
"\u2581the",
"re",
"\u2581w",
"\u2581o",
"\u2581s",
"at",
"ou",
"er",
"nd",
"\u2581i",
"\u2581b",
"\u2581c",
"on",
"\u2581h",
"ing",
"\u2581to",
"\u2581m",
"en",
"\u2581f",
"\u2581p",
"an",
"\u2581d",
"es",
"or",
"ll",
"\u2581of",
"\u2581and",
"\u2581y",
"\u2581l",
"\u2581I",
"it",
"\u2581in",
"is",
"ed",
"\u2581g",
"\u2581you",
"ar",
"\u2581that",
"om",
"as",
"\u2581n",
"ve",
"us",
"ic",
"ow",
"al",
"\u2581it",
"\u2581be",
"\u2581wh",
"le",
"ion",
"ut",
"ot",
"\u2581we",
"\u2581is",
"\u2581e",
"et",
"ay",
"\u2581re",
"\u2581on",
"\u2581T",
"\u2581A",
"\u2581ha",
"ent",
"ke",
"ct",
"\u2581S",
"ig",
"ver",
"\u2581Th",
"all",
"id",
"\u2581for",
"ro",
"\u2581he",
"se",
"\u2581this",
"ld",
"ly",
"\u2581go",
"\u2581k",
"\u2581st",
"st",
"ch",
"\u2581li",
"\u2581u",
"am",
"ur",
"ce",
"ith",
"im",
"\u2581so",
"\u2581have",
"\u2581do",
"ht",
"th",
"\u2581an",
"\u2581with",
"ad",
"\u2581r",
"ir",
"\u2581was",
"\u2581as",
"\u2581W",
"\u2581are",
"ust",
"ally",
"\u2581j",
"\u2581se",
"ation",
"od",
"ere",
"\u2581like",
"\u2581not",
"\u2581kn",
"ight",
"\u2581B",
"\u2581they",
"\u2581And",
"\u2581know",
"ome",
"op",
"\u2581can",
"\u2581or",
"\u2581sh",
"\u2581me",
"ill",
"ant",
"ck",
"\u2581what",
"\u2581at",
"\u2581ab",
"ould",
"ol",
"\u2581So",
"\u2581C",
"use",
"ter",
"il",
"\u2581but",
"\u2581just",
"\u2581ne",
"\u2581de",
"ra",
"ore",
"\u2581there",
"ul",
"out",
"\u2581con",
"\u2581all",
"\u2581The",
"ers",
"\u2581H",
"\u2581fr",
"\u2581pro",
"ge",
"ea",
"\u2581Y",
"\u2581O",
"\u2581M",
"pp",
"\u2581com",
"ess",
"\u2581ch",
"\u2581al",
"est",
"ate",
"qu",
"\u2581lo",
"\u2581ex",
"very",
"\u2581su",
"ain",
"\u2581one",
"ca",
"art",
"ist",
"if",
"ive",
"\u2581if",
"ink",
"nt",
"ab",
"\u2581about",
"\u2581going",
"\u2581v",
"\u2581wor",
"um",
"ok",
"\u2581your",
"\u2581my",
"ind",
"\u2581get",
"cause",
"\u2581from",
"\u2581don",
"ri",
"pe",
"un",
"ity",
"\u2581up",
"\u2581P",
"\u2581out",
"ort",
"\u2581L",
"ment",
"el",
"\u2581N",
"\u2581some",
"ich",
"and",
"\u2581think",
"em",
"oug",
"\u2581G",
"os",
"\u2581D",
"res",
"\u2581because",
"\u2581by",
"ake",
"\u2581int",
"ie",
"\u2581us",
"\u2581tr",
"\u2581then",
"ack",
"\u2581pl",
"\u2581here",
"\u2581pe",
"her",
"\u2581will",
"\u2581F",
"\u2581which",
"ard",
"\u2581right",
"\u2581thing",
"\u2581want",
"ies",
"ople",
"\u2581It",
"\u2581them",
"ame",
"\u2581We",
"our",
"\u2581say",
"\u2581R",
"\u2581people",
"\u2581see",
"\u2581who",
"ast",
"ure",
"ect",
"ear",
"\u2581tim",
"\u2581E",
"\u2581You",
"\u2581would",
"\u2581when",
"ven",
"\u2581our",
"ci",
"\u2581really",
"\u2581more",
"ound",
"ose",
"ak",
"\u2581co",
"ide",
"ough",
"\u2581had",
"so",
"\u2581qu",
"eah",
"\u2581were",
"ine",
"\u2581act",
"ther",
"\u2581these",
"\u2581how",
"\u2581now",
"\u2581sa",
"ud",
"\u2581Wh",
"\u2581man",
"ous",
"one",
"pt",
"ff",
"ong",
"\u2581has",
"\u2581any",
"\u2581very",
"\u2581But",
"\u2581look",
"iv",
"itt",
"\u2581time",
"\u2581mo",
"\u2581ar",
"hing",
"\u2581le",
"\u2581work",
"\u2581their",
"are",
"\u2581his",
"per",
"ions",
"\u2581im",
"\u2581ag",
"\u2581J",
"\u2581no",
"\u2581en",
"\u2581got",
"ag",
"\u2581sp",
"ans",
"act",
"\u2581te",
"\u2581also",
"iz",
"ice",
"\u2581That",
"\u2581cl",
"\u2581been",
"\u2581way",
"\u2581fe",
"\u2581did",
"ple",
"ually",
"\u2581other",
"\u2581U",
"ite",
"age",
"omet",
"ber",
"reat",
"ree",
"\u2581into",
"own",
"\u2581tw",
"\u2581part",
"alk",
"\u2581where",
"\u2581need",
"\u2581every",
"pl",
"\u2581ad",
"ry",
"\u2581over",
"ble",
"ap",
"ue",
"\u2581kind",
"\u2581po",
"\u2581back",
"\u2581cont",
"iff",
"\u2581somet",
"\u2581pr",
"nder",
"ire",
"\u2581good",
"\u2581than",
"ace",
"\u2581gu",
"ep",
"og",
"ick",
"way",
"\u2581lot",
"\u2581un",
"\u2581things",
"\u2581In",
"ish",
"kay",
"\u2581well",
"\u2581could",
"\u2581pre",
"\u2581two",
"irst",
"\u2581diff",
"ach",
"cc",
"ittle",
"int",
"\u2581He",
"\u2581those",
"ence",
"ip",
"ase",
"\u2581him",
"\u2581make",
"\u2581little",
"ical",
"\u2581gr",
"\u2581year",
"ass",
"\u2581thr",
"uch",
"ated",
"\u2581This",
"\u2581off",
"\u2581res",
"ac",
"ance",
"\u2581actually",
"\u2581talk",
"ult",
"able",
"orm",
"\u2581dis",
"\u2581first",
"ations",
"\u2581something",
"\u2581she",
"sel",
"\u2581let",
"ord",
"\u2581may",
"ia",
"\u2581am",
"\u2581her",
"\u2581said",
"\u2581bo",
"be",
"ount",
"\u2581much",
"\u2581per",
"\u2581even",
"\u2581differe",
"vel",
"ary",
"\u2581app",
"ving",
"\u2581comm",
"\u2581imp",
"ys",
"\u2581again",
"ress",
"\u2581yeah",
"\u2581down",
"ang",
"\u2581mean",
"na",
"ens",
"\u2581does",
"\u2581fo",
"\u2581comp",
"\u2581ro",
"\u2581bl",
"ody",
"\u2581K",
"\u2581through",
"\u2581start",
"uct",
"\u2581only",
"\u2581bet",
"\u2581under",
"\u2581br",
"\u2581take",
"ning",
"\u2581bu",
"\u2581use",
"\u2581Ch",
"xt",
"co",
"ory",
"ild",
"\u2581put",
"\u2581call",
"\u2581new",
"other",
"ting",
"\u2581happ",
"ater",
"\u2581inc",
"ition",
"\u2581different",
"\u2581should",
"ade",
"ign",
"thing",
"\u2581day",
"fore",
"\u2581Yeah",
"ark",
"ile",
"ial",
"\u2581come",
"\u2581They",
"\u2581being",
"\u2581try",
"ious",
"\u2581sc",
"\u2581bit",
"\u2581spe",
"ub",
"fe",
"\u2581doing",
"\u2581St",
"vers",
"av",
"ty",
"ian",
"onna",
"red",
"wn",
"\u2581ke",
"form",
"ors",
"\u2581fl",
"fter",
"ail",
"ents",
"\u2581gonna",
"\u2581point",
"ces",
"\u2581There",
"self",
"\u2581many",
"\u2581If",
"\u2581same",
"\u2581sy",
"\u2581quest",
"\u2581most",
"\u2581great",
"\u2581What",
"\u2581fu",
"ug",
"\u2581show",
"we",
"ual",
"ons",
"\u2581Be",
"ically",
"\u2581ser",
"\u2581rem",
"\u2581ind",
"\u2581pers",
"\u2581V",
"he",
"\u2581str",
"ved",
"\u2581still",
"ank",
"\u2581rec",
"\u2581wr",
"ought",
"day",
"ath",
"\u2581end",
"\u2581bas",
"ft",
"erm",
"body",
"ph",
"ject",
"ict",
"\u2581play",
"\u2581Is",
"ates",
"\u2581ph",
"oth",
"\u2581acc",
"get",
"\u2581years",
"\u2581em",
"\u2581id",
"\u2581Oh",
"ves",
"ever",
"\u2581inter",
"\u2581rel",
"\u2581before",
"\u2581feel",
"igh",
"\u2581three",
"iss",
"\u2581des",
"ne",
"\u2581why",
"\u2581uh",
"\u2581To",
"\u2581cons",
"\u2581hel",
"\u2581after",
"ower",
"urn",
"\u2581okay",
"\u2581long",
"\u2581bel",
"\u2581around",
"ful",
"te",
"ise",
"\u2581ob",
"\u2581supp",
"ady",
"ange",
"aking",
"\u2581pos",
"atch",
"\u2581tra",
"gr",
"\u2581might",
"ert",
"\u2581help",
"ost",
"\u2581too",
"cial",
"\u2581world",
"\u2581give",
"ike",
"\u2581Okay",
"ways",
"\u2581min",
"ward",
"ily",
"\u2581gen",
"\u2581find",
"\u2581dec",
"ular",
"ob",
"\u2581tell",
"\u2581Now",
"\u2581sm",
"\u2581cour",
"\u2581real",
"cess",
"nds",
"\u2581big",
"\u2581num",
"ction",
"\u2581add",
"\u2581set",
"\u2581um",
"ood",
"ible",
"\u2581own",
"\u2581life",
"ities",
"\u2581its",
"\u2581God",
"pect",
"\u2581didn",
"stem",
"les",
"uc",
"ib",
"ating",
"olog",
"\u2581person",
"\u2581inv",
"ably",
"\u2581sure",
"\u2581reg",
"lic",
"\u2581stu",
"\u2581cr",
"\u2581ev",
"ments",
"\u2581another",
"\u2581la",
"\u2581last",
"\u2581sub",
"\u2581att",
"\u2581op",
"\u2581inst",
"\u2581sl",
"\u2581happen",
"\u2581rep",
"\u2581import",
"ific",
"ix",
"\u2581made",
"\u2581ear",
"\u2581ac",
"\u2581def",
"ute",
"\u2581next",
"ative",
"\u2581form",
"\u2581guys",
"\u2581system",
"ew",
"\u2581able",
"ied",
"\u2581always",
"ren",
"erest",
"\u2581As",
"\u2581mod",
"\u2581done",
"ings",
"\u2581love",
"ism",
"\u2581ask",
"old",
"ered",
"\u2581trans",
"\u2581count",
"ility",
"\u2581high",
"\u2581fin",
"\u2581prob",
"\u2581pol",
"\u2581exam",
"\u2581pres",
"\u2581maybe",
"ell",
"\u2581stud",
"\u2581prod",
"\u2581car",
"ock",
"\u2581used",
"oy",
"stand",
"\u2581No",
"\u2581mon",
"ks",
"\u2581interest",
"\u2581ent",
"ited",
"\u2581sort",
"\u2581For",
"\u2581today",
"ics",
"\u2581vide",
"\u2581bec",
"\u2581Well",
"\u2581Al",
"\u2581important",
"\u2581such",
"\u2581run",
"\u2581keep",
"\u2581fact",
"ata",
"ss",
"\u2581never",
"ween",
"\u2581stuff",
"ract",
"\u2581question",
"als",
"\u2581sim",
"vern",
"ather",
"\u2581course",
"\u2581Of",
"oc",
"ness",
"arch",
"ize",
"\u2581All",
"ense",
"blem",
"\u2581probably",
"hip",
"\u2581number",
"ention",
"\u2581saying",
"\u2581commun",
"\u2581An",
"akes",
"\u2581belie",
"\u2581between",
"\u2581better",
"cus",
"\u2581place",
"\u2581gener",
"\u2581ca",
"\u2581ins",
"\u2581ass",
"cond",
"cept",
"ull",
"\u2581understand",
"\u2581fun",
"\u2581thought",
"gan",
"iew",
"cy",
"ution",
"ope",
"ason",
"\u2581problem",
"\u2581doesn",
"ational",
"\u2581read",
"\u2581trying",
"\u2581sch",
"\u2581el",
"ah",
"atter",
"\u2581exper",
"\u2581four",
"\u2581ele",
"\u2581cou",
"ont",
"\u2581called",
"\u2581partic",
"\u2581open",
"\u2581gl",
"\u2581everything",
"\u2581eff",
"\u2581getting",
"\u2581ty",
"\u2581Am",
"\u2581Because",
"ave",
"\u2581met",
"\u2581Like",
"oney",
"\u2581",
"e",
"t",
"o",
"a",
"n",
"i",
"s",
"h",
"r",
"l",
"d",
"u",
"c",
"y",
"m",
"g",
"w",
"f",
"p",
",",
"b",
".",
"k",
"v",
"'",
"I",
"T",
"A",
"S",
"j",
"x",
"W",
"B",
"C",
"?",
"0",
"O",
"-",
"M",
"H",
"Y",
"q",
"1",
"P",
"z",
"L",
"D",
"N",
"G",
"F",
"R",
"E",
"2",
"J",
"U",
":",
"5",
"9",
"3",
"K",
"4",
"V",
"8",
"6",
"7",
"!",
"%",
"Q",
"$",
"Z",
"X",
"\u00e9",
"/",
"\u00ed",
"\u00e1",
"\u00a3",
"\u00f3",
"\u0101",
"\u00fc",
"\u00f1",
"\u00f6",
"\u00e8",
"\u00e7",
"\u00e0",
"\u00bf",
"\u03bc",
"\u03c0",
"\u00e4",
"\u00fa",
"\u03b8",
"\u00e3",
"\u03c6",
"\u012b",
"\u03c3",
"\u00ea",
"\u03c1",
"\u00e2",
"\u00f4",
"^",
"\u20ac",
"\u00c9",
"\u016b",
"\u0394",
"\u03bb",
"\u03b1",
"\u03c4",
"\u00e6",
"\u0430",
"\u043e",
"\u03bd",
"\u00ee",
"\u03b3",
"\u03c8",
"\u0113",
"\u0442",
"\u00df",
"\u03c9",
"\u00ef",
"\u0107",
"\u010d",
"\u03b5",
"\u0435",
"\u0438",
"\u00f2",
"\u0440",
"\u03b2",
"\u00f8",
"\u0142",
"\u03b4",
"\u03b7",
"\u043f",
"\u00eb",
"\u043d",
"\u0441",
"\u0161",
"\u00dc",
"\u00e5",
"\u0144",
"\u015b",
"\u044f",
"\u0111",
"\u043b",
"\u043c",
"\u00d6",
"\u00fb",
"\u0219",
"\u0432",
"\u00c1",
"\u00d8",
"\u00f9",
"\u03bf",
"\u0447",
"\u044c",
"\u017e",
"\u03a6",
"\u0443",
"\u0119",
"\u03b9",
"\u0431",
"\u0433",
"\u043a",
"\u0151",
"\u015a",
"\u03a9",
"\u03ba",
"\u03c5",
"\u00ec",
"\u010c",
"\u03ad",
"\u0445",
"\u044b",
"\u00c5",
"\u00c7",
"\u017c",
"\u03af",
"\u03b6",
"\u03c7",
"\u044d",
"\u00c6",
"\u00cd",
"\u00f5",
"\u011b",
"\u0127",
"\u0141",
"\u0153",
"\u017d",
"\u021b",
"\u0393",
"\u041f",
"\u0434",
"\u0437",
"\u0444",
"\u00a1",
"\u00c0",
"\u00ce",
"\u0100",
"\u0117",
"\u0160",
"\u017a",
"\u039a",
"\u03a8",
"\u03ac",
"\u03be",
"\u03cc"
]
},
"decoding": {
"strategy": "greedy_batch",
"model_type": "tdt",
"durations": [
0,
1,
2,
3,
4
],
"greedy": {
"max_symbols": 10
},
"beam": {
"beam_size": 2,
"return_best_hypothesis": false,
"score_norm": true,
"tsd_max_sym_exp": 50,
"alsd_max_target_len": 2.0
}
},
"aux_ctc": {
"ctc_loss_weight": 0.3,
"use_cer": false,
"ctc_reduction": "mean_batch",
"decoder": {
"_target_": "nemo.collections.asr.modules.ConvASRDecoder",
"feat_in": null,
"num_classes": -1,
"vocabulary": []
},
"decoding": {
"strategy": "greedy"
}
},
"interctc": {
"loss_weights": [],
"apply_at_layers": []
},
"loss": {
"loss_name": "tdt",
"tdt_kwargs": {
"fastemit_lambda": 0.0,
"clamp": -1.0,
"durations": [
0,
1,
2,
3,
4
],
"sigma": 0.02,
"omega": 0.1
}
},
"optim": {
"name": "adamw",
"lr": 0.0001,
"betas": [
0.9,
0.98
],
"weight_decay": 0.001,
"sched": {
"name": "CosineAnnealing",
"warmup_steps": 0,
"warmup_ratio": null,
"min_lr": 1e-06
}
},
"labels": [
"<unk>",
"\u2581t",
"\u2581th",
"\u2581a",
"in",
"\u2581the",
"re",
"\u2581w",
"\u2581o",
"\u2581s",
"at",
"ou",
"er",
"nd",
"\u2581i",
"\u2581b",
"\u2581c",
"on",
"\u2581h",
"ing",
"\u2581to",
"\u2581m",
"en",
"\u2581f",
"\u2581p",
"an",
"\u2581d",
"es",
"or",
"ll",
"\u2581of",
"\u2581and",
"\u2581y",
"\u2581l",
"\u2581I",
"it",
"\u2581in",
"is",
"ed",
"\u2581g",
"\u2581you",
"ar",
"\u2581that",
"om",
"as",
"\u2581n",
"ve",
"us",
"ic",
"ow",
"al",
"\u2581it",
"\u2581be",
"\u2581wh",
"le",
"ion",
"ut",
"ot",
"\u2581we",
"\u2581is",
"\u2581e",
"et",
"ay",
"\u2581re",
"\u2581on",
"\u2581T",
"\u2581A",
"\u2581ha",
"ent",
"ke",
"ct",
"\u2581S",
"ig",
"ver",
"\u2581Th",
"all",
"id",
"\u2581for",
"ro",
"\u2581he",
"se",
"\u2581this",
"ld",
"ly",
"\u2581go",
"\u2581k",
"\u2581st",
"st",
"ch",
"\u2581li",
"\u2581u",
"am",
"ur",
"ce",
"ith",
"im",
"\u2581so",
"\u2581have",
"\u2581do",
"ht",
"th",
"\u2581an",
"\u2581with",
"ad",
"\u2581r",
"ir",
"\u2581was",
"\u2581as",
"\u2581W",
"\u2581are",
"ust",
"ally",
"\u2581j",
"\u2581se",
"ation",
"od",
"ere",
"\u2581like",
"\u2581not",
"\u2581kn",
"ight",
"\u2581B",
"\u2581they",
"\u2581And",
"\u2581know",
"ome",
"op",
"\u2581can",
"\u2581or",
"\u2581sh",
"\u2581me",
"ill",
"ant",
"ck",
"\u2581what",
"\u2581at",
"\u2581ab",
"ould",
"ol",
"\u2581So",
"\u2581C",
"use",
"ter",
"il",
"\u2581but",
"\u2581just",
"\u2581ne",
"\u2581de",
"ra",
"ore",
"\u2581there",
"ul",
"out",
"\u2581con",
"\u2581all",
"\u2581The",
"ers",
"\u2581H",
"\u2581fr",
"\u2581pro",
"ge",
"ea",
"\u2581Y",
"\u2581O",
"\u2581M",
"pp",
"\u2581com",
"ess",
"\u2581ch",
"\u2581al",
"est",
"ate",
"qu",
"\u2581lo",
"\u2581ex",
"very",
"\u2581su",
"ain",
"\u2581one",
"ca",
"art",
"ist",
"if",
"ive",
"\u2581if",
"ink",
"nt",
"ab",
"\u2581about",
"\u2581going",
"\u2581v",
"\u2581wor",
"um",
"ok",
"\u2581your",
"\u2581my",
"ind",
"\u2581get",
"cause",
"\u2581from",
"\u2581don",
"ri",
"pe",
"un",
"ity",
"\u2581up",
"\u2581P",
"\u2581out",
"ort",
"\u2581L",
"ment",
"el",
"\u2581N",
"\u2581some",
"ich",
"and",
"\u2581think",
"em",
"oug",
"\u2581G",
"os",
"\u2581D",
"res",
"\u2581because",
"\u2581by",
"ake",
"\u2581int",
"ie",
"\u2581us",
"\u2581tr",
"\u2581then",
"ack",
"\u2581pl",
"\u2581here",
"\u2581pe",
"her",
"\u2581will",
"\u2581F",
"\u2581which",
"ard",
"\u2581right",
"\u2581thing",
"\u2581want",
"ies",
"ople",
"\u2581It",
"\u2581them",
"ame",
"\u2581We",
"our",
"\u2581say",
"\u2581R",
"\u2581people",
"\u2581see",
"\u2581who",
"ast",
"ure",
"ect",
"ear",
"\u2581tim",
"\u2581E",
"\u2581You",
"\u2581would",
"\u2581when",
"ven",
"\u2581our",
"ci",
"\u2581really",
"\u2581more",
"ound",
"ose",
"ak",
"\u2581co",
"ide",
"ough",
"\u2581had",
"so",
"\u2581qu",
"eah",
"\u2581were",
"ine",
"\u2581act",
"ther",
"\u2581these",
"\u2581how",
"\u2581now",
"\u2581sa",
"ud",
"\u2581Wh",
"\u2581man",
"ous",
"one",
"pt",
"ff",
"ong",
"\u2581has",
"\u2581any",
"\u2581very",
"\u2581But",
"\u2581look",
"iv",
"itt",
"\u2581time",
"\u2581mo",
"\u2581ar",
"hing",
"\u2581le",
"\u2581work",
"\u2581their",
"are",
"\u2581his",
"per",
"ions",
"\u2581im",
"\u2581ag",
"\u2581J",
"\u2581no",
"\u2581en",
"\u2581got",
"ag",
"\u2581sp",
"ans",
"act",
"\u2581te",
"\u2581also",
"iz",
"ice",
"\u2581That",
"\u2581cl",
"\u2581been",
"\u2581way",
"\u2581fe",
"\u2581did",
"ple",
"ually",
"\u2581other",
"\u2581U",
"ite",
"age",
"omet",
"ber",
"reat",
"ree",
"\u2581into",
"own",
"\u2581tw",
"\u2581part",
"alk",
"\u2581where",
"\u2581need",
"\u2581every",
"pl",
"\u2581ad",
"ry",
"\u2581over",
"ble",
"ap",
"ue",
"\u2581kind",
"\u2581po",
"\u2581back",
"\u2581cont",
"iff",
"\u2581somet",
"\u2581pr",
"nder",
"ire",
"\u2581good",
"\u2581than",
"ace",
"\u2581gu",
"ep",
"og",
"ick",
"way",
"\u2581lot",
"\u2581un",
"\u2581things",
"\u2581In",
"ish",
"kay",
"\u2581well",
"\u2581could",
"\u2581pre",
"\u2581two",
"irst",
"\u2581diff",
"ach",
"cc",
"ittle",
"int",
"\u2581He",
"\u2581those",
"ence",
"ip",
"ase",
"\u2581him",
"\u2581make",
"\u2581little",
"ical",
"\u2581gr",
"\u2581year",
"ass",
"\u2581thr",
"uch",
"ated",
"\u2581This",
"\u2581off",
"\u2581res",
"ac",
"ance",
"\u2581actually",
"\u2581talk",
"ult",
"able",
"orm",
"\u2581dis",
"\u2581first",
"ations",
"\u2581something",
"\u2581she",
"sel",
"\u2581let",
"ord",
"\u2581may",
"ia",
"\u2581am",
"\u2581her",
"\u2581said",
"\u2581bo",
"be",
"ount",
"\u2581much",
"\u2581per",
"\u2581even",
"\u2581differe",
"vel",
"ary",
"\u2581app",
"ving",
"\u2581comm",
"\u2581imp",
"ys",
"\u2581again",
"ress",
"\u2581yeah",
"\u2581down",
"ang",
"\u2581mean",
"na",
"ens",
"\u2581does",
"\u2581fo",
"\u2581comp",
"\u2581ro",
"\u2581bl",
"ody",
"\u2581K",
"\u2581through",
"\u2581start",
"uct",
"\u2581only",
"\u2581bet",
"\u2581under",
"\u2581br",
"\u2581take",
"ning",
"\u2581bu",
"\u2581use",
"\u2581Ch",
"xt",
"co",
"ory",
"ild",
"\u2581put",
"\u2581call",
"\u2581new",
"other",
"ting",
"\u2581happ",
"ater",
"\u2581inc",
"ition",
"\u2581different",
"\u2581should",
"ade",
"ign",
"thing",
"\u2581day",
"fore",
"\u2581Yeah",
"ark",
"ile",
"ial",
"\u2581come",
"\u2581They",
"\u2581being",
"\u2581try",
"ious",
"\u2581sc",
"\u2581bit",
"\u2581spe",
"ub",
"fe",
"\u2581doing",
"\u2581St",
"vers",
"av",
"ty",
"ian",
"onna",
"red",
"wn",
"\u2581ke",
"form",
"ors",
"\u2581fl",
"fter",
"ail",
"ents",
"\u2581gonna",
"\u2581point",
"ces",
"\u2581There",
"self",
"\u2581many",
"\u2581If",
"\u2581same",
"\u2581sy",
"\u2581quest",
"\u2581most",
"\u2581great",
"\u2581What",
"\u2581fu",
"ug",
"\u2581show",
"we",
"ual",
"ons",
"\u2581Be",
"ically",
"\u2581ser",
"\u2581rem",
"\u2581ind",
"\u2581pers",
"\u2581V",
"he",
"\u2581str",
"ved",
"\u2581still",
"ank",
"\u2581rec",
"\u2581wr",
"ought",
"day",
"ath",
"\u2581end",
"\u2581bas",
"ft",
"erm",
"body",
"ph",
"ject",
"ict",
"\u2581play",
"\u2581Is",
"ates",
"\u2581ph",
"oth",
"\u2581acc",
"get",
"\u2581years",
"\u2581em",
"\u2581id",
"\u2581Oh",
"ves",
"ever",
"\u2581inter",
"\u2581rel",
"\u2581before",
"\u2581feel",
"igh",
"\u2581three",
"iss",
"\u2581des",
"ne",
"\u2581why",
"\u2581uh",
"\u2581To",
"\u2581cons",
"\u2581hel",
"\u2581after",
"ower",
"urn",
"\u2581okay",
"\u2581long",
"\u2581bel",
"\u2581around",
"ful",
"te",
"ise",
"\u2581ob",
"\u2581supp",
"ady",
"ange",
"aking",
"\u2581pos",
"atch",
"\u2581tra",
"gr",
"\u2581might",
"ert",
"\u2581help",
"ost",
"\u2581too",
"cial",
"\u2581world",
"\u2581give",
"ike",
"\u2581Okay",
"ways",
"\u2581min",
"ward",
"ily",
"\u2581gen",
"\u2581find",
"\u2581dec",
"ular",
"ob",
"\u2581tell",
"\u2581Now",
"\u2581sm",
"\u2581cour",
"\u2581real",
"cess",
"nds",
"\u2581big",
"\u2581num",
"ction",
"\u2581add",
"\u2581set",
"\u2581um",
"ood",
"ible",
"\u2581own",
"\u2581life",
"ities",
"\u2581its",
"\u2581God",
"pect",
"\u2581didn",
"stem",
"les",
"uc",
"ib",
"ating",
"olog",
"\u2581person",
"\u2581inv",
"ably",
"\u2581sure",
"\u2581reg",
"lic",
"\u2581stu",
"\u2581cr",
"\u2581ev",
"ments",
"\u2581another",
"\u2581la",
"\u2581last",
"\u2581sub",
"\u2581att",
"\u2581op",
"\u2581inst",
"\u2581sl",
"\u2581happen",
"\u2581rep",
"\u2581import",
"ific",
"ix",
"\u2581made",
"\u2581ear",
"\u2581ac",
"\u2581def",
"ute",
"\u2581next",
"ative",
"\u2581form",
"\u2581guys",
"\u2581system",
"ew",
"\u2581able",
"ied",
"\u2581always",
"ren",
"erest",
"\u2581As",
"\u2581mod",
"\u2581done",
"ings",
"\u2581love",
"ism",
"\u2581ask",
"old",
"ered",
"\u2581trans",
"\u2581count",
"ility",
"\u2581high",
"\u2581fin",
"\u2581prob",
"\u2581pol",
"\u2581exam",
"\u2581pres",
"\u2581maybe",
"ell",
"\u2581stud",
"\u2581prod",
"\u2581car",
"ock",
"\u2581used",
"oy",
"stand",
"\u2581No",
"\u2581mon",
"ks",
"\u2581interest",
"\u2581ent",
"ited",
"\u2581sort",
"\u2581For",
"\u2581today",
"ics",
"\u2581vide",
"\u2581bec",
"\u2581Well",
"\u2581Al",
"\u2581important",
"\u2581such",
"\u2581run",
"\u2581keep",
"\u2581fact",
"ata",
"ss",
"\u2581never",
"ween",
"\u2581stuff",
"ract",
"\u2581question",
"als",
"\u2581sim",
"vern",
"ather",
"\u2581course",
"\u2581Of",
"oc",
"ness",
"arch",
"ize",
"\u2581All",
"ense",
"blem",
"\u2581probably",
"hip",
"\u2581number",
"ention",
"\u2581saying",
"\u2581commun",
"\u2581An",
"akes",
"\u2581belie",
"\u2581between",
"\u2581better",
"cus",
"\u2581place",
"\u2581gener",
"\u2581ca",
"\u2581ins",
"\u2581ass",
"cond",
"cept",
"ull",
"\u2581understand",
"\u2581fun",
"\u2581thought",
"gan",
"iew",
"cy",
"ution",
"ope",
"ason",
"\u2581problem",
"\u2581doesn",
"ational",
"\u2581read",
"\u2581trying",
"\u2581sch",
"\u2581el",
"ah",
"atter",
"\u2581exper",
"\u2581four",
"\u2581ele",
"\u2581cou",
"ont",
"\u2581called",
"\u2581partic",
"\u2581open",
"\u2581gl",
"\u2581everything",
"\u2581eff",
"\u2581getting",
"\u2581ty",
"\u2581Am",
"\u2581Because",
"ave",
"\u2581met",
"\u2581Like",
"oney",
"\u2581",
"e",
"t",
"o",
"a",
"n",
"i",
"s",
"h",
"r",
"l",
"d",
"u",
"c",
"y",
"m",
"g",
"w",
"f",
"p",
",",
"b",
".",
"k",
"v",
"'",
"I",
"T",
"A",
"S",
"j",
"x",
"W",
"B",
"C",
"?",
"0",
"O",
"-",
"M",
"H",
"Y",
"q",
"1",
"P",
"z",
"L",
"D",
"N",
"G",
"F",
"R",
"E",
"2",
"J",
"U",
":",
"5",
"9",
"3",
"K",
"4",
"V",
"8",
"6",
"7",
"!",
"%",
"Q",
"$",
"Z",
"X",
"\u00e9",
"/",
"\u00ed",
"\u00e1",
"\u00a3",
"\u00f3",
"\u0101",
"\u00fc",
"\u00f1",
"\u00f6",
"\u00e8",
"\u00e7",
"\u00e0",
"\u00bf",
"\u03bc",
"\u03c0",
"\u00e4",
"\u00fa",
"\u03b8",
"\u00e3",
"\u03c6",
"\u012b",
"\u03c3",
"\u00ea",
"\u03c1",
"\u00e2",
"\u00f4",
"^",
"\u20ac",
"\u00c9",
"\u016b",
"\u0394",
"\u03bb",
"\u03b1",
"\u03c4",
"\u00e6",
"\u0430",
"\u043e",
"\u03bd",
"\u00ee",
"\u03b3",
"\u03c8",
"\u0113",
"\u0442",
"\u00df",
"\u03c9",
"\u00ef",
"\u0107",
"\u010d",
"\u03b5",
"\u0435",
"\u0438",
"\u00f2",
"\u0440",
"\u03b2",
"\u00f8",
"\u0142",
"\u03b4",
"\u03b7",
"\u043f",
"\u00eb",
"\u043d",
"\u0441",
"\u0161",
"\u00dc",
"\u00e5",
"\u0144",
"\u015b",
"\u044f",
"\u0111",
"\u043b",
"\u043c",
"\u00d6",
"\u00fb",
"\u0219",
"\u0432",
"\u00c1",
"\u00d8",
"\u00f9",
"\u03bf",
"\u0447",
"\u044c",
"\u017e",
"\u03a6",
"\u0443",
"\u0119",
"\u03b9",
"\u0431",
"\u0433",
"\u043a",
"\u0151",
"\u015a",
"\u03a9",
"\u03ba",
"\u03c5",
"\u00ec",
"\u010c",
"\u03ad",
"\u0445",
"\u044b",
"\u00c5",
"\u00c7",
"\u017c",
"\u03af",
"\u03b6",
"\u03c7",
"\u044d",
"\u00c6",
"\u00cd",
"\u00f5",
"\u011b",
"\u0127",
"\u0141",
"\u0153",
"\u017d",
"\u021b",
"\u0393",
"\u041f",
"\u0434",
"\u0437",
"\u0444",
"\u00a1",
"\u00c0",
"\u00ce",
"\u0100",
"\u0117",
"\u0160",
"\u017a",
"\u039a",
"\u03a8",
"\u03ac",
"\u03be",
"\u03cc"
],
"target": "nemo.collections.asr.models.rnnt_bpe_models.EncDecRNNTBPEModel",
"nemo_version": "2.4.0rc0"
}