-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_model.sh
30 lines (22 loc) · 3.1 KB
/
train_model.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
# Trying models with different configuration
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 10 --initial-learning-rate 1 --use-gru ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 15 --initial-learning-rate 1 --use-gru ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 --use-gru ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 15 --attn-num-layers 3 --initial-learning-rate 1 --use-gru ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 10 --initial-learning-rate 1 ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 15 --initial-learning-rate 1 ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 50 --batch-size 64 --target-embedding-size 15 --attn-num-layers 3 --initial-learning-rate 1 ./datasets/small-data.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 10 --initial-learning-rate 1 --use-gru ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 15 --initial-learning-rate 1 --use-gru ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 --use-gru ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 15 --attn-num-layers 3 --initial-learning-rate 1 --use-gru ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 10 --initial-learning-rate 1 ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 15 --initial-learning-rate 1 ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 ./datasets/email-training.tfrecords
# aocr train --no-resume --num-epoch 25 --batch-size 64 --target-embedding-size 15 --attn-num-layers 3 --initial-learning-rate 1 ./datasets/email-training.tfrecords
# Selecting aocr train --batch-size 64 --target-embedding-size 20 --use-gru for final training
# aocr train --no-resume --num-epoch 10 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 --use-gru ./datasets/new-train.tfrecords
# aocr train --num-epoch 5 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 1 --use-gru ./datasets/new-train.tfrecords
# aocr train --num-epoch 1 --batch-size 64 --target-embedding-size 20 --initial-learning-rate 0.005 --use-gru ./datasets/new-train.tfrecords
aocr train --num-epoch 10 --target-embedding-size 20 --use-gru --initial-learning-rate 0.01 ./datasets/train.tfrecords