RecordManager.md

RecordManager

• Init RecordManager
• Use RecordManager
• Plot

import random
from torchhk.rm import RecordManager

Init RecordManager

rm = RecordManager(['Epoch', 'Iter', 'loss_a', 'loss_b', 'train_acc', 'test_acc'])
rm

RecordManager(keys=[Epoch, Iter, loss_a, loss_b, train_acc, test_acc])

Use RecordManager

for epoch in range(5):
    
    for i in range(2) :

        #########
        # Train #
        #########
        
        rm.progress()
        
        rm.add([epoch+1, i,
                random.random(), random.random(),
                90+random.random()*(epoch+1), 90+random.random()*(epoch+1)])

rm.summary()

--------------------------------------------------------
Epoch   Iter   loss_a   loss_b   train_acc   test_acc   
========================================================
1       0      0.5122   0.8280   90.1486     90.2941    
--------------------------------------------------------
1       1      0.1822   0.4356   90.1765     90.3528    
--------------------------------------------------------
2       0      0.6448   0.4194   91.9486     90.3231    
--------------------------------------------------------
2       1      0.8943   0.9038   91.0642     91.4241    
--------------------------------------------------------
3       0      0.6358   0.7359   92.4341     90.5188    
--------------------------------------------------------
3       1      0.9082   0.7852   92.3704     92.4229    
--------------------------------------------------------
4       0      0.9486   0.1586   93.7000     91.9010    
--------------------------------------------------------
4       1      0.9043   0.4352   90.6886     92.4582    
--------------------------------------------------------
5       0      0.1235   0.1660   91.9936     91.1863    
--------------------------------------------------------
5       1      0.6256   0.3831   91.1582     91.0011    
--------------------------------------------------------
========================================================
Total Epoch: 5
Time Elapsed: 0:00:00.046384
Min(epoch)/Max(epoch): 
-loss_a: 0.1235(5)/0.9486(4)
-loss_b: 0.1586(4)/0.9038(2)
-train_acc: 90.1486(1)/93.7000(4)
-test_acc: 90.2941(1)/92.4582(4)
--------------------------------------------------------

Plot

rm.plot('Epoch', 'loss_a')

rm.plot('Epoch', [['loss_a', 'loss_b'], ['train_acc', 'test_acc']], 
        title="All Records", xlabel='epochs', ylabel="loss", ylabel_second="acc",
        ylim_second=(90,95),
        linestyles=['-', '-', '--', '--'],
        colors=['r', 'g', 'b', 'y'])