opt-babylm2-rewritten-clean-spacy-earlystop_hierarchical_211_age-origin_adj2-bpe_seed-1024_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: 2.6826
• Accuracy: 0.4788

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: 1024
• 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
4.0969	1.0	2256	3.8123	0.3609
3.4532	2.0	4512	3.3014	0.4096
3.1308	3.0	6768	3.0884	0.4305
2.9231	4.0	9024	2.9812	0.4414
2.8438	5.0	11280	2.9205	0.4477
2.7858	6.0	13536	2.8798	0.4520
2.7413	7.0	15792	2.8560	0.4547
2.7094	8.0	18048	2.8348	0.4568
2.6876	9.0	20304	2.8210	0.4587
2.6663	10.0	22560	2.8111	0.4599
2.6472	11.0	24816	2.8042	0.4609
2.6465	12.0	27072	2.7948	0.4615
2.6369	13.0	29328	2.7909	0.4623
2.6257	14.0	31584	2.7868	0.4622
2.6109	15.0	33840	2.7652	0.4655
2.5667	16.0	36096	2.7395	0.4687
2.5162	17.0	38352	2.7178	0.4717
2.4611	18.0	40608	2.7003	0.4746
2.3984	19.0	42864	2.6857	0.4773
2.325	20.0	45120	2.6826	0.4788

Framework versions

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