learning_rate: 0.001
train_batch_size: 32
eval_batch_size: 64
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 256
optimizer: Use OptimizerNames.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: 32000
num_epochs: 20.0
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Accuracy
4.0907	1.0	2264	3.8071	0.3612
3.4383	2.0	4528	3.2980	0.4094
3.1227	3.0	6792	3.0874	0.4300
2.9193	4.0	9056	2.9818	0.4408
2.8404	5.0	11320	2.9211	0.4472
2.7821	6.0	13584	2.8807	0.4514
2.7389	7.0	15848	2.8561	0.4542
2.7066	8.0	18112	2.8369	0.4564
2.6832	9.0	20376	2.8227	0.4578
2.6626	10.0	22640	2.8118	0.4593
2.6456	11.0	24904	2.8031	0.4604
2.6441	12.0	27168	2.7938	0.4614
2.6321	13.0	29432	2.7913	0.4618
2.6219	14.0	31696	2.7878	0.4621
2.6034	15.0	33960	2.7630	0.4654
2.5609	16.0	36224	2.7376	0.4685
2.514	17.0	38488	2.7165	0.4716
2.4567	18.0	40752	2.6971	0.4747
2.3927	19.0	43016	2.6812	0.4775
2.3214	19.9914	45260	2.6796	0.4787

kanishka/opt-babylm2-clean-spacy-earlystop-bpe_seed-42_1e-3

opt-babylm2-clean-spacy-earlystop-bpe_seed-42_1e-3