Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

standardized inference and number of iterations for mSuperb single lang track #4905

Merged
merged 7 commits into from
Feb 3, 2023
Merged

standardized inference and number of iterations for mSuperb single lang track #4905

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
a8122fa
standardizing iterations and inference model
Feb 1, 2023
c0b1d3a
standardizing iterations and inference model
DanBerrebbi Feb 1, 2023
d719ec6
applied black
DanBerrebbi Feb 2, 2023
b1d7f98
applied black
DanBerrebbi Feb 2, 2023
a113f3f
fixing minor bugs
DanBerrebbi Feb 2, 2023
1de4a28
applied blacked and solved inference model problem
DanBerrebbi Feb 2, 2023
75bc56e
black new version
DanBerrebbi Feb 3, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
22 changes: 10 additions & 12 deletions egs2/msuperb/asr1/conf/tuning/train_asr_fbank_single.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
encoder: conformer
encoder: transformer
encoder_conf:
output_size: 256
attention_heads: 8
Expand All @@ -9,13 +9,7 @@ encoder_conf:
attention_dropout_rate: 0.1
input_layer: conv2d2
normalize_before: true
macaron_style: true
rel_pos_type: latest
pos_enc_layer_type: rel_pos
selfattention_layer_type: rel_selfattn
activation_type: swish
use_cnn_module: true
cnn_module_kernel: 15


decoder: none

Expand All @@ -30,14 +24,14 @@ num_workers: 4
batch_type: sorted
batch_size: 8
accum_grad: 4
max_epoch: 50
max_epoch: 200
ftshijt marked this conversation as resolved.
Show resolved Hide resolved
patience: none
init: none
best_model_criterion:
- - valid
- acc
- max
keep_nbest_models: 10
- loss
- min
keep_nbest_models: 5

optim: adam
optim_conf:
Expand All @@ -59,3 +53,7 @@ specaug_conf:
- 0.
- 0.05
num_time_mask: 10


num_iters_per_epoch: 500 # number of iterations per epoch
max_epoch: 30
21 changes: 9 additions & 12 deletions egs2/msuperb/asr1/conf/tuning/train_asr_s3prl_single.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
encoder: conformer
encoder: transformer
encoder_conf:
output_size: 256
attention_heads: 8
Expand All @@ -9,13 +9,6 @@ encoder_conf:
attention_dropout_rate: 0.1
input_layer: conv2d2
normalize_before: true
macaron_style: true
rel_pos_type: latest
pos_enc_layer_type: rel_pos
selfattention_layer_type: rel_selfattn
activation_type: swish
use_cnn_module: true
cnn_module_kernel: 15

decoder: none

Expand Down Expand Up @@ -43,14 +36,14 @@ num_workers: 4
batch_type: sorted
batch_size: 8
accum_grad: 4
max_epoch: 50
max_epoch: 200
ftshijt marked this conversation as resolved.
Show resolved Hide resolved
patience: none
init: none
best_model_criterion:
- - valid
- acc
- max
keep_nbest_models: 10
- loss
- min
keep_nbest_models: 5

optim: adam
optim_conf:
Expand All @@ -72,3 +65,7 @@ specaug_conf:
- 0.
- 0.05
num_time_mask: 10


num_iters_per_epoch: 500 # number of iterations per epoch
max_epoch: 30
1 change: 1 addition & 0 deletions egs2/msuperb/asr1/run_single_lang.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,7 @@ for duration in 10min 1h; do
--feats_type raw \
--asr_config "${asr_config}" \
--inference_config "${inference_config}" \
--inference_asr_model "valid.loss.ave_5best.pth" \
ftshijt marked this conversation as resolved.
Show resolved Hide resolved
--train_set "${train_set}" \
--valid_set "${train_dev}" \
--test_sets "${test_set}" \
Expand Down