opt-babylm2-rewritten-clean-spacy-earlystop_hierarchical_211_age-origin_adj1-bpe_seed-42_1e-3

This model was trained from scratch on the kanishka/babylm2-rewritten-clean-spacy_hierarchical-adj_211_age-origin_adj1-ablation dataset. It achieves the following results on the evaluation set:

• Loss: 2.6850
• Accuracy: 0.4783

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
• 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
4.0955	1.0	2256	3.8052	0.3618
3.4456	2.0	4512	3.2946	0.4099
3.125	3.0	6768	3.0869	0.4306
2.923	4.0	9024	2.9833	0.4411
2.8464	5.0	11280	2.9220	0.4471
2.7873	6.0	13536	2.8827	0.4511
2.7433	7.0	15792	2.8578	0.4545
2.7136	8.0	18048	2.8395	0.4564
2.6903	9.0	20304	2.8225	0.4581
2.668	10.0	22560	2.8125	0.4594
2.6502	11.0	24816	2.8054	0.4603
2.6501	12.0	27072	2.8014	0.4612
2.6387	13.0	29328	2.7939	0.4619
2.6284	14.0	31584	2.7923	0.4623
2.613	15.0	33840	2.7676	0.4650
2.5674	16.0	36096	2.7411	0.4682
2.5204	17.0	38352	2.7208	0.4714
2.4641	18.0	40608	2.7018	0.4743
2.3983	19.0	42864	2.6874	0.4769
2.3272	20.0	45120	2.6850	0.4783

Framework versions

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