Training Progress
Step 650 of 1000
65.0%
~35 min left
Current Loss
0.150
2.3%
Current Reward
1.000
3.7%
KL Divergence
0.021
1.2%
Tokens/Sec
1326
0.5%
Training Loss
How well the model is learning
Reward Score
Model performance score
KL Divergence
Difference from original model
Learning Rate
Learning rate schedule