asun17904/anliR2-t5-base-alum

Knowledge Continuity Regularized Network

Dataset: ANLI Round: None

Trainer Hyperparameters:

• lr = 5e-05
• per_device_batch_size = 4
• gradient_accumulation_steps = 4
• weight_decay = 1e-09
• seed = 42

Regularization Hyperparameters

• numerical stability denominator constant = 1.0
• lambda = 1.0
• alpha = 1.0
• beta = 1.0

Extended Logs:

eval_loss	eval_accuracy	epoch
1.140	0.389	1.0
1.127	0.407	2.0
1.126	0.409	3.0
1.130	0.401	4.0
1.122	0.414	5.0
1.110	0.431	6.0
1.114	0.427	7.0
1.109	0.433	8.0
1.102	0.440	9.0
1.093	0.451	10.0
1.085	0.459	11.0
1.096	0.448	12.0
1.092	0.449	13.0
1.094	0.449	14.0

Test Accuracy: 0.448