opt-babylm2-rewritten-clean-spacy-earlystop_hierarchical_211_age-origin_adj2-bpe_seed-211_1e-3

This model was trained from scratch on the kanishka/babylm2-rewritten-clean-spacy_hierarchical-adj_211_age-origin_adj2-ablation dataset. It achieves the following results on the evaluation set:

• Loss: 2.6823
• Accuracy: 0.4786

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
4.0947	1.0	2256	3.8134	0.3605
3.4536	2.0	4512	3.3000	0.4096
3.1288	3.0	6768	3.0869	0.4310
2.9233	4.0	9024	2.9804	0.4415
2.8452	5.0	11280	2.9195	0.4474
2.7873	6.0	13536	2.8806	0.4517
2.7423	7.0	15792	2.8551	0.4546
2.7103	8.0	18048	2.8367	0.4568
2.6867	9.0	20304	2.8202	0.4584
2.6677	10.0	22560	2.8088	0.4599
2.6494	11.0	24816	2.8028	0.4610
2.6473	12.0	27072	2.7959	0.4613
2.6367	13.0	29328	2.7940	0.4618
2.6273	14.0	31584	2.7863	0.4626
2.6111	15.0	33840	2.7641	0.4654
2.5684	16.0	36096	2.7397	0.4687
2.5175	17.0	38352	2.7189	0.4717
2.4637	18.0	40608	2.7000	0.4748
2.3982	19.0	42864	2.6855	0.4772
2.3239	20.0	45120	2.6823	0.4786

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.4