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

Loss: 3.4130
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: 0.0003
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
3.7346	1.0	18600	3.8972	0.3462
3.4287	2.0	37200	3.6409	0.3755
3.2901	3.0	55800	3.4779	0.3900
3.2043	4.0	74400	3.4166	0.3964
3.1401	5.0	93000	3.3858	0.4011
3.0938	6.0	111600	3.3652	0.4039
3.0576	7.0	130200	3.3590	0.4050
3.0235	8.0	148800	3.3599	0.4059
2.9953	9.0	167400	3.3596	0.4070
2.9671	10.0	186000	3.3670	0.4079
2.9471	11.0	204600	3.3677	0.4085
2.9196	12.0	223200	3.3675	0.4088
2.8999	13.0	241800	3.3742	0.4088
2.8818	14.0	260400	3.3657	0.4102
2.8608	15.0	279000	3.3741	0.4094
2.8388	16.0	297600	3.3850	0.4092
2.8207	17.0	316200	3.3959	0.4094
2.7965	18.0	334800	3.3924	0.4097
2.786	19.0	353400	3.4080	0.4092
2.7601	20.0	372000	3.4130	0.4092

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-only_measure_nps_as_singular_removal-3e-4

smolm-autoreg-bpe-counterfactual-babylm-only_measure_nps_as_singular_removal-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-only_measure_nps_as_singular_removal-3e-4