smolm-autoreg-bpe-counterfactual-babylm-only_other_det_removal-seed_211-1e-3

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

• Loss: 3.4143
• Accuracy: 0.4110

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
3.6004	1.0	18597	3.8219	0.3575
3.3852	2.0	37194	3.6092	0.3797
3.2597	3.0	55791	3.4837	0.3910
3.1758	4.0	74388	3.4364	0.3981
3.1197	5.0	92985	3.4116	0.4017
3.08	6.0	111582	3.3782	0.4040
3.0418	7.0	130179	3.3885	0.4055
3.0088	8.0	148776	3.3884	0.4062
2.9856	9.0	167373	3.3548	0.4077
2.9598	10.0	185970	3.3782	0.4090
2.9364	11.0	204567	3.3851	0.4093
2.9156	12.0	223164	3.3803	0.4097
2.8949	13.0	241761	3.3869	0.4100
2.8719	14.0	260358	3.3813	0.4104
2.8526	15.0	278955	3.3859	0.4108
2.8289	16.0	297552	3.3980	0.4103
2.8104	17.0	316149	3.3981	0.4109
2.7958	18.0	334746	3.4054	0.4110
2.781	19.0	353343	3.4057	0.4110
2.7571	20.0	371940	3.4143	0.4110

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1