This model was trained from scratch on the kanishka/counterfactual-babylm-random_removal dataset. It achieves the following results on the evaluation set:

Loss: 3.4298
Accuracy: 0.4089

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
3.6092	1.0	18586	3.7513	0.3603
3.3896	2.0	37172	3.5882	0.3803
3.2599	3.0	55758	3.4709	0.3917
3.177	4.0	74344	3.4183	0.3981
3.1287	5.0	92930	3.3982	0.4017
3.0841	6.0	111516	3.3841	0.4031
3.0483	7.0	130102	3.3494	0.4065
3.0156	8.0	148688	3.3597	0.4078
2.9911	9.0	167274	3.3719	0.4067
2.9616	10.0	185860	3.3717	0.4078
2.9384	11.0	204446	3.3679	0.4091
2.9133	12.0	223032	3.3673	0.4097
2.8923	13.0	241618	3.3885	0.4088
2.8781	14.0	260204	3.3873	0.4090
2.8563	15.0	278790	3.3848	0.4092
2.836	16.0	297376	3.3956	0.4094
2.8162	17.0	315962	3.4023	0.4091
2.7997	18.0	334548	3.4101	0.4093
2.7779	19.0	353134	3.4237	0.4090
2.7645	20.0	371720	3.4298	0.4089

Framework versions

Transformers 4.37.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.1

kanishka/smolm-autoreg-bpe-counterfactual-babylm-random_removal-1e-3

smolm-autoreg-bpe-counterfactual-babylm-random_removal-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-random_removal-1e-3