espnet/brianyan918_mustc-v2_en-de_st_ctc_rnnt_asrinit_raw_en_de_bpe_tc4000_sp - AI Model Zoo

Download model and run inference:

./run.sh --skip_data_prep false --skip_train true --download_model espnet/brianyan918_mustc-v2_en-de_st_ctc_rnnt_asrinit_raw_en_de_bpe_tc4000_sp --inference_config conf/tuning/decode_rnnt_tsd_mse4_scorenormduring_beam10.yaml

dataset	score	verbose_score
decode_rnnt_tsd_mse4_scorenormduring_beam10_st_model_valid.loss.ave_10best/tst-COMMON.en-de	27.6	60.2/33.6/21.0/13.7 (BP = 0.998 ratio = 0.998 hyp_len = 51602 ref_len = 51699)