Speech Processing and Machine Intelligence (SPMI)
Add files using upload-large-folder tool
ce6ccc2 verified | { | |
| "data": { | |
| "train": "text_sample", | |
| "dev": [ | |
| "dev_en", | |
| "dev_es", | |
| "dev_fr", | |
| "dev_it", | |
| "dev_ky", | |
| "dev_nl", | |
| "dev_ru", | |
| "dev_sv-SE", | |
| "dev_tr", | |
| "dev_tt" | |
| ], | |
| "test": [ | |
| "test_en", | |
| "test_es", | |
| "test_fr", | |
| "test_it", | |
| "test_ky", | |
| "test_nl", | |
| "test_ru", | |
| "test_sv-SE", | |
| "test_tr", | |
| "test_tt" | |
| ], | |
| "filter": "10:1200" | |
| }, | |
| "tokenizer": { | |
| "type": "SentencePieceTokenizer", | |
| "option-train": { | |
| "model_type": "bpe", | |
| "vocab_size": 5000, | |
| "model_prefix": "dict/ten/ten_bpe/spm_bpe5000" | |
| }, | |
| "|V|": 5000, | |
| "file": "dict/ten/tokenizer_bpe5000.tknz" | |
| }, | |
| "env": { | |
| "CUDA_VISIBLE_DEVICES": "0,1,2,3,4,5,6,7,8,9" | |
| }, | |
| "train": { | |
| "bin": "cat.ctc.train", | |
| "option": { | |
| "amp": true, | |
| "batch_size": 345, | |
| "check_freq": 856, | |
| "large-dataset": true, | |
| "ld": "data/tar_wo_lid/*/{11_1000,1001_1200}/*.tar" | |
| } | |
| }, | |
| "inference": { | |
| "avgmodel": { | |
| "mode": "best", | |
| "num": 3 | |
| }, | |
| "infer": { | |
| "bin": "cat.ctc.decode", | |
| "option": { | |
| "beam_size": 16, | |
| "nj": 16, | |
| "store_ark": true | |
| } | |
| }, | |
| "er": {} | |
| }, | |
| "commit": "1b77bc21f7a5faac5e47f6bc4445f23ebf0e0633" | |
| } | |