Upload model files
Browse files- config.json +92 -0
- preprocessor_config.json +175 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
config.json
ADDED
|
@@ -0,0 +1,92 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_dropout": 0.1,
|
| 3 |
+
"apply_spec_augment": true,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"EmformerForRNNT"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"codevector_dim": 256,
|
| 10 |
+
"contrastive_logits_temperature": 0.1,
|
| 11 |
+
"conv_bias": false,
|
| 12 |
+
"conv_dim": [
|
| 13 |
+
512,
|
| 14 |
+
512,
|
| 15 |
+
512,
|
| 16 |
+
512,
|
| 17 |
+
512,
|
| 18 |
+
512,
|
| 19 |
+
512
|
| 20 |
+
],
|
| 21 |
+
"conv_kernel": [
|
| 22 |
+
10,
|
| 23 |
+
3,
|
| 24 |
+
3,
|
| 25 |
+
3,
|
| 26 |
+
3,
|
| 27 |
+
2,
|
| 28 |
+
2
|
| 29 |
+
],
|
| 30 |
+
"conv_stride": [
|
| 31 |
+
5,
|
| 32 |
+
2,
|
| 33 |
+
2,
|
| 34 |
+
2,
|
| 35 |
+
2,
|
| 36 |
+
2,
|
| 37 |
+
2
|
| 38 |
+
],
|
| 39 |
+
"ctc_loss_reduction": "sum",
|
| 40 |
+
"ctc_zero_infinity": false,
|
| 41 |
+
"diversity_loss_weight": 0.1,
|
| 42 |
+
"do_stable_layer_norm": false,
|
| 43 |
+
"eos_token_id": 2,
|
| 44 |
+
"feat_extract_activation": "gelu",
|
| 45 |
+
"feat_extract_norm": "group",
|
| 46 |
+
"feat_proj_dropout": 0.0,
|
| 47 |
+
"feat_quantizer_dropout": 0.0,
|
| 48 |
+
"ffn_dim": 2048,
|
| 49 |
+
"final_dropout": 0.1,
|
| 50 |
+
"hidden_act": "gelu",
|
| 51 |
+
"hidden_dropout": 0.1,
|
| 52 |
+
"hidden_size": 768,
|
| 53 |
+
"initializer_range": 0.02,
|
| 54 |
+
"input_dim": 80,
|
| 55 |
+
"intermediate_size": 3072,
|
| 56 |
+
"joiner_activation": "relu",
|
| 57 |
+
"layer_norm_eps": 1e-05,
|
| 58 |
+
"layerdrop": 0.1,
|
| 59 |
+
"left_context_length": 30,
|
| 60 |
+
"lstm_dropout": 0.3,
|
| 61 |
+
"lstm_hidden_dim": 512,
|
| 62 |
+
"lstm_layer_norm": 0.001,
|
| 63 |
+
"lstm_layer_norm_epsilon": 0.001,
|
| 64 |
+
"mask_feature_length": 10,
|
| 65 |
+
"mask_feature_min_masks": 0,
|
| 66 |
+
"mask_feature_prob": 0.0,
|
| 67 |
+
"mask_time_length": 10,
|
| 68 |
+
"mask_time_min_masks": 2,
|
| 69 |
+
"mask_time_prob": 0.05,
|
| 70 |
+
"model_type": "emformer",
|
| 71 |
+
"num_attention_heads": 8,
|
| 72 |
+
"num_codevector_groups": 2,
|
| 73 |
+
"num_codevectors_per_group": 320,
|
| 74 |
+
"num_conv_pos_embedding_groups": 16,
|
| 75 |
+
"num_conv_pos_embeddings": 128,
|
| 76 |
+
"num_feat_extract_layers": 7,
|
| 77 |
+
"num_hidden_layers": 20,
|
| 78 |
+
"num_lstm_layers": 3,
|
| 79 |
+
"num_negatives": 100,
|
| 80 |
+
"output_dim": 1024,
|
| 81 |
+
"pad_token_id": 0,
|
| 82 |
+
"proj_codevector_dim": 256,
|
| 83 |
+
"right_context_length": 4,
|
| 84 |
+
"segment_length": 16,
|
| 85 |
+
"symbol_embedding_dim": 512,
|
| 86 |
+
"time_reduction_input_dim": 128,
|
| 87 |
+
"time_reduction_stride": 4,
|
| 88 |
+
"torch_dtype": "float32",
|
| 89 |
+
"transformer_ffn_dim": 2048,
|
| 90 |
+
"transformers_version": "4.20.0.dev0",
|
| 91 |
+
"vocab_size": 4097
|
| 92 |
+
}
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,175 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"feature_extractor_type": "EmformerFeatureExtractor",
|
| 3 |
+
"feature_size": 80,
|
| 4 |
+
"global_invstddev": [
|
| 5 |
+
0.25952333211898804,
|
| 6 |
+
0.2590482831001282,
|
| 7 |
+
0.24866817891597748,
|
| 8 |
+
0.24776232242584229,
|
| 9 |
+
0.22200720012187958,
|
| 10 |
+
0.21363843977451324,
|
| 11 |
+
0.20652402937412262,
|
| 12 |
+
0.19909949600696564,
|
| 13 |
+
0.2021811604499817,
|
| 14 |
+
0.20355898141860962,
|
| 15 |
+
0.20546883344650269,
|
| 16 |
+
0.2061648815870285,
|
| 17 |
+
0.20569036900997162,
|
| 18 |
+
0.20412985980510712,
|
| 19 |
+
0.20357738435268402,
|
| 20 |
+
0.2041499763727188,
|
| 21 |
+
0.2055872678756714,
|
| 22 |
+
0.20807604491710663,
|
| 23 |
+
0.21054454147815704,
|
| 24 |
+
0.21341396868228912,
|
| 25 |
+
0.21418628096580505,
|
| 26 |
+
0.22065168619155884,
|
| 27 |
+
0.2248840034008026,
|
| 28 |
+
0.22723940014839172,
|
| 29 |
+
0.230172261595726,
|
| 30 |
+
0.23371541500091553,
|
| 31 |
+
0.23734734952449799,
|
| 32 |
+
0.23960146307945251,
|
| 33 |
+
0.24088498950004578,
|
| 34 |
+
0.241532102227211,
|
| 35 |
+
0.24218633770942688,
|
| 36 |
+
0.24371792376041412,
|
| 37 |
+
0.2447739839553833,
|
| 38 |
+
0.25564682483673096,
|
| 39 |
+
0.2632736265659332,
|
| 40 |
+
0.2549223005771637,
|
| 41 |
+
0.24608071148395538,
|
| 42 |
+
0.2464841604232788,
|
| 43 |
+
0.2470586597919464,
|
| 44 |
+
0.24785254895687103,
|
| 45 |
+
0.24904784560203552,
|
| 46 |
+
0.2503036856651306,
|
| 47 |
+
0.25226327776908875,
|
| 48 |
+
0.2532329559326172,
|
| 49 |
+
0.2527913451194763,
|
| 50 |
+
0.2518651783466339,
|
| 51 |
+
0.2504975199699402,
|
| 52 |
+
0.24836081266403198,
|
| 53 |
+
0.24765831232070923,
|
| 54 |
+
0.24767662584781647,
|
| 55 |
+
0.24965286254882812,
|
| 56 |
+
0.2501370906829834,
|
| 57 |
+
0.2508895993232727,
|
| 58 |
+
0.2512582540512085,
|
| 59 |
+
0.25150999426841736,
|
| 60 |
+
0.2525503635406494,
|
| 61 |
+
0.25313329696655273,
|
| 62 |
+
0.2534785270690918,
|
| 63 |
+
0.25330957770347595,
|
| 64 |
+
0.25366073846817017,
|
| 65 |
+
0.25502219796180725,
|
| 66 |
+
0.2608155608177185,
|
| 67 |
+
0.25662899017333984,
|
| 68 |
+
0.2558451294898987,
|
| 69 |
+
0.25671014189720154,
|
| 70 |
+
0.2577403485774994,
|
| 71 |
+
0.25914356112480164,
|
| 72 |
+
0.2596718966960907,
|
| 73 |
+
0.25953933596611023,
|
| 74 |
+
0.2610883116722107,
|
| 75 |
+
0.26132410764694214,
|
| 76 |
+
0.26272818446159363,
|
| 77 |
+
0.26397505402565,
|
| 78 |
+
0.26440608501434326,
|
| 79 |
+
0.26543495059013367,
|
| 80 |
+
0.26753780245780945,
|
| 81 |
+
0.26935192942619324,
|
| 82 |
+
0.26732245087623596,
|
| 83 |
+
0.26666897535324097,
|
| 84 |
+
0.2663257420063019
|
| 85 |
+
],
|
| 86 |
+
"global_mean": [
|
| 87 |
+
15.058613777160645,
|
| 88 |
+
16.34557342529297,
|
| 89 |
+
16.34653663635254,
|
| 90 |
+
16.240671157836914,
|
| 91 |
+
17.45355224609375,
|
| 92 |
+
17.445302963256836,
|
| 93 |
+
17.52323341369629,
|
| 94 |
+
18.076807022094727,
|
| 95 |
+
17.699262619018555,
|
| 96 |
+
17.706790924072266,
|
| 97 |
+
17.24724578857422,
|
| 98 |
+
17.153791427612305,
|
| 99 |
+
17.213361740112305,
|
| 100 |
+
17.347240447998047,
|
| 101 |
+
17.331117630004883,
|
| 102 |
+
17.21516227722168,
|
| 103 |
+
17.030071258544922,
|
| 104 |
+
16.818960189819336,
|
| 105 |
+
16.573062896728516,
|
| 106 |
+
16.29717254638672,
|
| 107 |
+
16.00996971130371,
|
| 108 |
+
15.794167518615723,
|
| 109 |
+
15.616395950317383,
|
| 110 |
+
15.459056854248047,
|
| 111 |
+
15.306838989257812,
|
| 112 |
+
15.199165344238281,
|
| 113 |
+
15.208144187927246,
|
| 114 |
+
14.883454322814941,
|
| 115 |
+
14.787869453430176,
|
| 116 |
+
14.947835922241211,
|
| 117 |
+
14.5912504196167,
|
| 118 |
+
14.76955509185791,
|
| 119 |
+
14.617781639099121,
|
| 120 |
+
14.840407371520996,
|
| 121 |
+
14.83073616027832,
|
| 122 |
+
14.909119606018066,
|
| 123 |
+
14.89070987701416,
|
| 124 |
+
14.918207168579102,
|
| 125 |
+
14.939517974853516,
|
| 126 |
+
14.913643836975098,
|
| 127 |
+
14.863334655761719,
|
| 128 |
+
14.803299903869629,
|
| 129 |
+
14.751264572143555,
|
| 130 |
+
14.688116073608398,
|
| 131 |
+
14.63498306274414,
|
| 132 |
+
14.615056037902832,
|
| 133 |
+
14.680213928222656,
|
| 134 |
+
14.616259574890137,
|
| 135 |
+
14.707776069641113,
|
| 136 |
+
14.630264282226562,
|
| 137 |
+
14.644737243652344,
|
| 138 |
+
14.547430038452148,
|
| 139 |
+
14.529033660888672,
|
| 140 |
+
14.49357795715332,
|
| 141 |
+
14.411538124084473,
|
| 142 |
+
14.33312702178955,
|
| 143 |
+
14.260393142700195,
|
| 144 |
+
14.204919815063477,
|
| 145 |
+
14.130182266235352,
|
| 146 |
+
14.06987476348877,
|
| 147 |
+
14.010197639465332,
|
| 148 |
+
13.938552856445312,
|
| 149 |
+
13.750232696533203,
|
| 150 |
+
13.607213973999023,
|
| 151 |
+
13.457777976989746,
|
| 152 |
+
13.31512451171875,
|
| 153 |
+
13.167718887329102,
|
| 154 |
+
13.019341468811035,
|
| 155 |
+
12.8869047164917,
|
| 156 |
+
12.795098304748535,
|
| 157 |
+
12.685126304626465,
|
| 158 |
+
12.620392799377441,
|
| 159 |
+
12.58949089050293,
|
| 160 |
+
12.537697792053223,
|
| 161 |
+
12.496938705444336,
|
| 162 |
+
12.410022735595703,
|
| 163 |
+
12.346826553344727,
|
| 164 |
+
12.221966743469238,
|
| 165 |
+
12.122841835021973,
|
| 166 |
+
12.005624771118164
|
| 167 |
+
],
|
| 168 |
+
"hop_length": 160,
|
| 169 |
+
"n_fft": 400,
|
| 170 |
+
"n_mels": 80,
|
| 171 |
+
"padding_side": "right",
|
| 172 |
+
"padding_value": 0.0,
|
| 173 |
+
"return_attention_mask": true,
|
| 174 |
+
"sampling_rate": 16000
|
| 175 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3c79e78f96c3bd9a1513079828569d9c6cfe622f8c03a0159eeaab391cff5e99
|
| 3 |
+
size 307068673
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "sp_model_kwargs": {}, "tokenizer_class": "EmformerTokenizer"}
|