ZiweiLi commited on
Commit ·
b6e70d8
1
Parent(s): 743b95d
add config
Browse files- config.json +41 -0
- hyper-p.json +68 -0
config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"specaug": {
|
| 3 |
+
"freq_mask_width_range": 0.35,
|
| 4 |
+
"num_freq_mask": 2,
|
| 5 |
+
"num_time_mask": 10,
|
| 6 |
+
"time_mask_width_range": 0.05
|
| 7 |
+
},
|
| 8 |
+
"encoder": {
|
| 9 |
+
"type": "ConformerNet",
|
| 10 |
+
"kwargs": {
|
| 11 |
+
"num_cells": 14,
|
| 12 |
+
"idim": 80,
|
| 13 |
+
"hdim": 512,
|
| 14 |
+
"conv": "vgg2l",
|
| 15 |
+
"num_heads": 4,
|
| 16 |
+
"kernel_size": 15,
|
| 17 |
+
"num_classes": 75
|
| 18 |
+
}
|
| 19 |
+
},
|
| 20 |
+
"scheduler": {
|
| 21 |
+
"type": "SchedulerNoamEarlyStop",
|
| 22 |
+
"kwargs": {
|
| 23 |
+
"warmup_step": 20000,
|
| 24 |
+
"min_step": 30000,
|
| 25 |
+
"dim_model": 512,
|
| 26 |
+
"peak_factor": 1.0,
|
| 27 |
+
"n_tol": 50
|
| 28 |
+
},
|
| 29 |
+
"optimizer": {
|
| 30 |
+
"type": "Adam",
|
| 31 |
+
"kwargs": {
|
| 32 |
+
"lr": 0.0003,
|
| 33 |
+
"betas": [
|
| 34 |
+
0.9,
|
| 35 |
+
0.98
|
| 36 |
+
],
|
| 37 |
+
"weight_decay": 1e-06
|
| 38 |
+
}
|
| 39 |
+
}
|
| 40 |
+
}
|
| 41 |
+
}
|
hyper-p.json
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data": {
|
| 3 |
+
"dev": [
|
| 4 |
+
"dev_en_mul",
|
| 5 |
+
"dev_es_mul",
|
| 6 |
+
"dev_fr_mul",
|
| 7 |
+
"dev_it_mul",
|
| 8 |
+
"dev_ky_mul",
|
| 9 |
+
"dev_ru_mul",
|
| 10 |
+
"dev_nl_mul",
|
| 11 |
+
"dev_tt_mul",
|
| 12 |
+
"dev_tr_mul",
|
| 13 |
+
"dev_sv-SE_mul"
|
| 14 |
+
],
|
| 15 |
+
"test": [
|
| 16 |
+
"test_en_mul",
|
| 17 |
+
"test_es_mul",
|
| 18 |
+
"test_fr_mul",
|
| 19 |
+
"test_it_mul",
|
| 20 |
+
"test_ky_mul",
|
| 21 |
+
"test_ru_mul",
|
| 22 |
+
"test_nl_mul",
|
| 23 |
+
"test_tt_mul",
|
| 24 |
+
"test_tr_mul",
|
| 25 |
+
"test_sv-SE_mul"
|
| 26 |
+
],
|
| 27 |
+
"filter": "10:1200"
|
| 28 |
+
},
|
| 29 |
+
"tokenizer": {
|
| 30 |
+
"type": "LexiconTokenizer",
|
| 31 |
+
"option-init": {
|
| 32 |
+
"lexicon": "dict/ten/lexicon_mul10.txt"
|
| 33 |
+
},
|
| 34 |
+
"file": "dict/ten/tokenizer_mul10.tknz",
|
| 35 |
+
"|V|": 75
|
| 36 |
+
},
|
| 37 |
+
"env": {
|
| 38 |
+
"CUDA_VISIBLE_DEVICES": "0,1,2,3,4,5,6,7,8,9"
|
| 39 |
+
},
|
| 40 |
+
"train": {
|
| 41 |
+
"bin": "cat.ctc.train",
|
| 42 |
+
"option": {
|
| 43 |
+
"amp": true,
|
| 44 |
+
"batch_size": 640,
|
| 45 |
+
"check_freq": 500,
|
| 46 |
+
"large-dataset": true,
|
| 47 |
+
"ld": "data/tar_mul/*/{11_1000,1001_1200}/*.tar"
|
| 48 |
+
}
|
| 49 |
+
},
|
| 50 |
+
"inference": {
|
| 51 |
+
"avgmodel": {
|
| 52 |
+
"mode": "best",
|
| 53 |
+
"num": 3
|
| 54 |
+
},
|
| 55 |
+
"infer": {
|
| 56 |
+
"bin": "cat.ctc.decode",
|
| 57 |
+
"option": {
|
| 58 |
+
"beam_size": 16,
|
| 59 |
+
"nj": 16,
|
| 60 |
+
"store_ark": true
|
| 61 |
+
}
|
| 62 |
+
},
|
| 63 |
+
"er": {
|
| 64 |
+
"per": true
|
| 65 |
+
}
|
| 66 |
+
},
|
| 67 |
+
"commit": "1b77bc21f7a5faac5e47f6bc4445f23ebf0e0633"
|
| 68 |
+
}
|