YAML Metadata
Error:
"datasets[0]" with value "https://github.com/egorsmkv/speech-recognition-uk" is not valid. If possible, use a dataset id from https://hf.co/datasets.
pruned_transducer_stateless5
with Conformer encoder for Ukrainian: https://github.com/proger/icefall/tree/uk
./pruned_transducer_stateless5/train.py \
--world-size 2 \
--num-epochs 30 \
--start-epoch 1 \
--full-libri 1 \
--exp-dir pruned_transducer_stateless5/exp-uk-shuf \
--max-duration 500 \
--use-fp16 1 \
--num-encoder-layers 18 \
--dim-feedforward 1024 \
--nhead 4 \
--encoder-dim 256 \
--decoder-dim 512 \
--joiner-dim 512 \
--bpe-model uk/data/lang_bpe_250/bpe.model
./pruned_transducer_stateless5/decode.py \
--epoch 27 \
--avg 15 \
--use-averaged-model True \
--exp-dir pruned_transducer_stateless5/exp-uk-shuf \
--decoding-method fast_beam_search \
--num-encoder-layers 18 \
--dim-feedforward 1024 \
--nhead 4 \
--encoder-dim 256 \
--decoder-dim 512 \
--joiner-dim 512 \
--bpe-model uk/data/lang_bpe_250/bpe.model \
--lang-dir uk/data/lang_bpe_250
Unable to determine this model's library. Check the
docs
.
Evaluation results
- Validation WER on Common Voice uktest set self-reported13.370