opt-babylm2-rewritten-clean-spacy-earlystop_adj-num-strict-swapped-bpe_seed-211_1e-3

This model was trained from scratch on the kanishka/babylm2-rewritten-clean-spacy_adj-num-strict-swapped dataset. It achieves the following results on the evaluation set:

• Loss: 2.6856
• Accuracy: 0.4784

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
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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
4.1006	1.0	2257	3.8182	0.3600
3.4504	2.0	4514	3.3030	0.4095
3.1303	3.0	6771	3.0936	0.4298
2.9241	4.0	9028	2.9881	0.4402
2.845	5.0	11285	2.9291	0.4463
2.7881	6.0	13542	2.8889	0.4510
2.7449	7.0	15799	2.8633	0.4539
2.7148	8.0	18056	2.8420	0.4563
2.691	9.0	20313	2.8247	0.4582
2.6698	10.0	22570	2.8165	0.4591
2.6533	11.0	24827	2.8093	0.4599
2.6522	12.0	27084	2.8027	0.4609
2.6386	13.0	29341	2.8010	0.4610
2.6282	14.0	31598	2.7923	0.4622
2.6129	15.0	33855	2.7684	0.4649
2.5701	16.0	36112	2.7443	0.4681
2.5196	17.0	38369	2.7217	0.4714
2.4633	18.0	40626	2.7019	0.4743
2.4002	19.0	42883	2.6888	0.4770
2.3266	19.9914	45120	2.6856	0.4784

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.1