Edit model card

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

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

  • Loss: 3.3871
  • Accuracy: 0.4118

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
3.6086 1.0 18593 3.7866 0.3570
3.3876 2.0 37186 3.5895 0.3814
3.2566 3.0 55779 3.4826 0.3920
3.1766 4.0 74372 3.4033 0.3996
3.1188 5.0 92965 3.4073 0.4015
3.0796 6.0 111558 3.3821 0.4045
3.0412 7.0 130151 3.3505 0.4082
3.0092 8.0 148744 3.3533 0.4078
2.9817 9.0 167337 3.3516 0.4085
2.9541 10.0 185930 3.3482 0.4095
2.9354 11.0 204523 3.3638 0.4100
2.9126 12.0 223116 3.3272 0.4119
2.8898 13.0 241709 3.3513 0.4110
2.8735 14.0 260302 3.3416 0.4124
2.8536 15.0 278895 3.3536 0.4122
2.8328 16.0 297488 3.3505 0.4125
2.8111 17.0 316081 3.3719 0.4116
2.7953 18.0 334674 3.3815 0.4117
2.7733 19.0 353267 3.3844 0.4118
2.7618 20.0 371860 3.3871 0.4118

Framework versions

  • Transformers 4.35.0
  • Pytorch 2.1.0+cu121
  • Datasets 2.12.0
  • Tokenizers 0.14.1
Downloads last month
1
Safetensors
Model size
97.8M params
Tensor type
F32
·