This model was trained from scratch on the kanishka/counterfactual_babylm_aann_high_variability_all dataset. It achieves the following results on the evaluation set:

Loss: 3.4047
Accuracy: 0.4097

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.001
train_batch_size: 32
eval_batch_size: 64
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 32000
num_epochs: 20.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.605	1.0	18595	3.7768	0.3598
3.3873	2.0	37190	3.5661	0.3794
3.2507	3.0	55785	3.4873	0.3898
3.1763	4.0	74380	3.4285	0.3965
3.1243	5.0	92975	3.3951	0.4005
3.0836	6.0	111570	3.3648	0.4042
3.0427	7.0	130165	3.3687	0.4051
3.0135	8.0	148760	3.3633	0.4059
2.9832	9.0	167355	3.3600	0.4066
2.9566	10.0	185950	3.3532	0.4086
2.9345	11.0	204545	3.3562	0.4085
2.9151	12.0	223140	3.3601	0.4088
2.8945	13.0	241735	3.3579	0.4089
2.8744	14.0	260330	3.3680	0.4093
2.8562	15.0	278925	3.3801	0.4087
2.8329	16.0	297520	3.3856	0.4096
2.8158	17.0	316115	3.3797	0.4095
2.7926	18.0	334710	3.3849	0.4105
2.7745	19.0	353305	3.3979	0.4098
2.7575	20.0	371900	3.4047	0.4097

Framework versions

Transformers 4.38.0
Pytorch 2.3.1+cu121
Datasets 2.16.1
Tokenizers 0.15.2

kanishka/smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-1e-3

smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-1e-3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

kanishka/
smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_all-1e-3