smolm-autoreg-bpe-counterfactual-babylm-indef-removal-seed_1024-1e-4

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

• Loss: 3.4350
• Accuracy: 0.4063

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 64
• seed: 1024
• 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
4.0569	1.0	18595	4.2925	0.3078
3.5613	2.0	37190	3.7467	0.3633
3.3894	3.0	55785	3.5963	0.3802
3.2856	4.0	74380	3.5303	0.3880
3.2222	5.0	92975	3.4793	0.3935
3.1693	6.0	111570	3.4552	0.3967
3.13	7.0	130165	3.4350	0.3992
3.0982	8.0	148760	3.4201	0.4011
3.0659	9.0	167355	3.4224	0.4025
3.0359	10.0	185950	3.4172	0.4028
3.0157	11.0	204545	3.4154	0.4036
2.9936	12.0	223140	3.4075	0.4043
2.9676	13.0	241735	3.4182	0.4048
2.9444	14.0	260330	3.4119	0.4057
2.9265	15.0	278925	3.4089	0.4056
2.9105	16.0	297520	3.4230	0.4058
2.8892	17.0	316115	3.4260	0.4060
2.8774	18.0	334710	3.4290	0.4063
2.8592	19.0	353305	3.4288	0.4065
2.8472	20.0	371900	3.4350	0.4063

Framework versions

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