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: 3.4270
Accuracy: 0.4094

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: 211
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
3.6072	1.0	18602	3.7687	0.3592
3.3848	2.0	37204	3.5595	0.3802
3.2576	3.0	55806	3.4654	0.3927
3.177	4.0	74408	3.4207	0.3982
3.1212	5.0	93010	3.4026	0.4006
3.0724	6.0	111612	3.3763	0.4035
3.0373	7.0	130214	3.3708	0.4051
3.0102	8.0	148816	3.3649	0.4063
2.9818	9.0	167418	3.3810	0.4072
2.9526	10.0	186020	3.3640	0.4078
2.9332	11.0	204622	3.3817	0.4081
2.9076	12.0	223224	3.3767	0.4087
2.8857	13.0	241826	3.3850	0.4089
2.8653	14.0	260428	3.3919	0.4093
2.8483	15.0	279030	3.3888	0.4091
2.828	16.0	297632	3.4040	0.4093
2.8069	17.0	316234	3.4020	0.4094
2.7906	18.0	334836	3.4096	0.4096
2.7701	19.0	353438	3.4215	0.4093
2.7515	20.0	372040	3.4270	0.4094

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_211-1e-3

smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-seed_211-1e-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_211-1e-3