asun17904/anliR1-t5-base-alum

Knowledge Continuity Regularized Network

Dataset: ANLI Round: None

Trainer Hyperparameters:

• lr = 5e-05
• per_device_batch_size = 2
• gradient_accumulation_steps = 4
• weight_decay = 0.0
• seed = 42

Regularization Hyperparameters

• numerical stability denominator constant = 0.01
• lambda = 0.0001
• alpha = 2.0
• beta = 2.0

Extended Logs:

eval_loss	eval_accuracy	epoch
1.070	0.413	1.0
1.110	0.407	2.0
1.115	0.416	3.0
1.108	0.431	4.0
1.108	0.428	5.0
1.119	0.413	6.0
1.102	0.438	7.0
1.107	0.429	8.0
1.101	0.439	9.0
1.101	0.434	10.0
1.110	0.428	11.0
1.102	0.442	12.0
1.110	0.430	13.0
1.093	0.455	14.0
1.105	0.434	15.0
1.106	0.435	16.0
1.105	0.439	17.0
1.099	0.441	18.0
1.099	0.443	19.0

Test Accuracy: 0.445