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

Loss: 3.3909
Accuracy: 0.4102

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: 0.0003
train_batch_size: 32
eval_batch_size: 64
seed: 1024
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
3.7399	1.0	18586	3.9066	0.3474
3.4368	2.0	37172	3.6279	0.3750
3.294	3.0	55758	3.4854	0.3884
3.2094	4.0	74344	3.4178	0.3968
3.1515	5.0	92930	3.3861	0.4009
3.1023	6.0	111516	3.3600	0.4041
3.0643	7.0	130102	3.3565	0.4047
3.0294	8.0	148688	3.3575	0.4059
3.0007	9.0	167274	3.3660	0.4068
2.9771	10.0	185860	3.3513	0.4075
2.9526	11.0	204446	3.3433	0.4092
2.9307	12.0	223032	3.3542	0.4094
2.91	13.0	241618	3.3446	0.4095
2.888	14.0	260204	3.3463	0.4100
2.862	15.0	278790	3.3530	0.4103
2.8465	16.0	297376	3.3666	0.4098
2.8291	17.0	315962	3.3780	0.4099
2.8072	18.0	334548	3.3858	0.4099
2.786	19.0	353134	3.3847	0.4104
2.773	20.0	371720	3.3909	0.4102

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-seed_1024-3e-4

smolm-autoreg-bpe-counterfactual-babylm-random_removal-seed_1024-3e-4

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-seed_1024-3e-4