learning_rate: 0.0006
train_batch_size: 32
eval_batch_size: 32
seed: 41
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 1000
num_epochs: 10.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1382	1.0	1035	4.3726	0.3182
1.0291	2.0	2070	4.3597	0.3146
1.0267	3.0	3105	4.3334	0.3179
1.0192	4.0	4140	4.2829	0.3216
1.0042	5.0	5175	4.2311	0.3267
0.9873	6.0	6210	4.1560	0.3320
0.9662	7.0	7245	4.0627	0.3410
0.9422	8.0	8280	3.9768	0.3493
0.9207	9.0	9315	3.9174	0.3542
0.9058	10.0	10350	3.8891	0.3569

xiulinyang/GPT2_BABYLM_20000

BABYLM_20000_41