smolm-autoreg-bpe-counterfactual-babylm-pipps_10k-1e-3

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

• Loss: 3.3378
• Accuracy: 0.4121

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: 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
3.6013	1.0	18720	3.7465	0.3588
3.3867	2.0	37440	3.4883	0.3819
3.2589	3.0	56160	3.4083	0.3943
3.1811	4.0	74880	3.3900	0.3986
3.1273	5.0	93600	3.3351	0.4020
3.0824	6.0	112320	3.3293	0.4028
3.0466	7.0	131040	3.3131	0.4065
3.0117	8.0	149760	3.2990	0.4092
2.9899	9.0	168480	3.2979	0.4097
2.9556	10.0	187200	3.2939	0.4105
2.9395	11.0	205920	3.2948	0.4120
2.9155	12.0	224640	3.3005	0.4110
2.8971	13.0	243360	3.2907	0.4126
2.8777	14.0	262080	3.2993	0.4121
2.8546	15.0	280800	3.2985	0.4128
2.8337	16.0	299520	3.2980	0.4130
2.8153	17.0	318240	3.3254	0.4116
2.799	18.0	336960	3.3345	0.4114
2.7824	19.0	355680	3.3410	0.4117
2.7614	20.0	374400	3.3378	0.4121

Framework versions

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