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

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.3932
• Accuracy: 0.4111

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
3.6069	1.0	18595	3.7979	0.3582
3.385	2.0	37190	3.5575	0.3809
3.2606	3.0	55785	3.4287	0.3945
3.1764	4.0	74380	3.4149	0.3985
3.1251	5.0	92975	3.3929	0.4017
3.0795	6.0	111570	3.3834	0.4039
3.0454	7.0	130165	3.3527	0.4071
3.0185	8.0	148760	3.3456	0.4078
2.9879	9.0	167355	3.3603	0.4079
2.9599	10.0	185950	3.3428	0.4097
2.9382	11.0	204545	3.3425	0.4117
2.9124	12.0	223140	3.3537	0.4100
2.8915	13.0	241735	3.3543	0.4104
2.8713	14.0	260330	3.3478	0.4116
2.8545	15.0	278925	3.3561	0.4110
2.8306	16.0	297520	3.3703	0.4106
2.817	17.0	316115	3.3776	0.4109
2.7938	18.0	334710	3.3861	0.4107
2.7768	19.0	353305	3.3823	0.4116
2.759	20.0	371900	3.3932	0.4111

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu121
• Datasets 2.12.0
• Tokenizers 0.14.1