3.0
3.5
4.0
4.5
5.0
24
26
28
30
32
34
36
500
500
1000
1000
1500
1500
2000
2000
2500
2500
3000
3000
3500
3500
epoch 1
epoch 1
★ best 3.2814
Training & Validation Loss
Cross-Entropy Loss
Train loss
Eval loss
★ best 26.61
25.96
Validation Perplexity
Perplexity (↓ better)
Perplexity
Training Step