mirror of
https://github.com/k2-fsa/icefall.git
synced 2025-12-11 06:55:27 +00:00
49 lines
1.4 KiB
Bash
Executable File
49 lines
1.4 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
# fix segmentation fault reported in https://github.com/k2-fsa/icefall/issues/674
|
|
export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
|
|
. ../../../tools/activate_python.sh
|
|
|
|
set -eou pipefail
|
|
|
|
stage=0
|
|
stop_stage=100
|
|
|
|
model=pruned_transducer_stateless_w2v
|
|
world_size=4
|
|
|
|
. shared/parse_options.sh || exit 1
|
|
|
|
log() {
|
|
# This function is from espnet
|
|
local fname=${BASH_SOURCE[1]##*/}
|
|
echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
|
|
}
|
|
|
|
if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then
|
|
log "Stage 0: Train model"
|
|
./transducer_unsupervised_finetuning_d2v_v2/train.py \
|
|
--wandb False \
|
|
--use-pseudo-labels True \
|
|
--load-prefinetuned-model ./model_locked/libri_finetuned.pt \
|
|
--input-strategy AudioSamples \
|
|
--enable-spec-aug False \
|
|
--multi-optim True \
|
|
--start-epoch 1 \
|
|
--world-size 4 \
|
|
--num-epochs 60 \
|
|
--exp-dir ./transducer_unsupervised_finetuning_d2v_v2/d2v-T-LJft \
|
|
--max-duration 150 \
|
|
--freeze-finetune-updates 1000 \
|
|
--encoder-dim 768 \
|
|
--decoder-dim 768 \
|
|
--joiner-dim 768 \
|
|
--use-fp16 1 \
|
|
--accum-grads 16 \
|
|
--encoder-type d2v \
|
|
--additional-block True \
|
|
--peak-dec-lr 0.04175 \
|
|
--peak-enc-lr 0.0003859 \
|
|
--update-ema True \
|
|
--layer-average False
|
|
fi |