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

Loss: 3.4280
Accuracy: 0.4092

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: 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
3.5991	1.0	18602	3.7770	0.3585
3.3808	2.0	37204	3.5764	0.3802
3.2544	3.0	55806	3.4559	0.3923
3.168	4.0	74408	3.4507	0.3977
3.1158	5.0	93010	3.4035	0.4024
3.0757	6.0	111612	3.3938	0.4040
3.0394	7.0	130214	3.3852	0.4048
3.0038	8.0	148816	3.3858	0.4069
2.9731	9.0	167418	3.3887	0.4073
2.9471	10.0	186020	3.3704	0.4084
2.9246	11.0	204622	3.3703	0.4090
2.904	12.0	223224	3.3839	0.4087
2.8815	13.0	241826	3.3820	0.4095
2.8618	14.0	260428	3.3779	0.4094
2.8451	15.0	279030	3.4014	0.4096
2.8223	16.0	297632	3.4042	0.4095
2.8049	17.0	316234	3.4128	0.4093
2.7864	18.0	334836	3.4110	0.4092
2.7622	19.0	353438	3.4278	0.4091
2.7455	20.0	372040	3.4280	0.4092

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_measure_nps_as_singular_new-seed_1024-1e-3

smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-seed_1024-1e-3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

kanishka/
smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-seed_1024-1e-3