opt-babylm2-rewritten-clean-spacy_no-num-adj-earlystop-strictest-bpe_seed-211_1e-3

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

• Loss: 2.6457
• Accuracy: 0.4820

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
• 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
3.0909	1.0	17737	3.1568	0.4220
2.9687	2.0	35474	3.0528	0.4331
2.8949	3.0	53211	2.9890	0.4395
2.8595	4.0	70948	2.9572	0.4426
2.8349	5.0	88685	2.9384	0.4446
2.8075	6.0	106422	2.9204	0.4468
2.7923	7.0	124159	2.9035	0.4486
2.7795	8.0	141896	2.8861	0.4505
2.7572	9.0	159633	2.8730	0.4521
2.7432	10.0	177370	2.8544	0.4541
2.7224	11.0	195107	2.8417	0.4558
2.6996	12.0	212844	2.8219	0.4582
2.6725	13.0	230581	2.8028	0.4601
2.6502	14.0	248318	2.7859	0.4625
2.6235	15.0	266055	2.7627	0.4653
2.5883	16.0	283792	2.7409	0.4680
2.5425	17.0	301529	2.7152	0.4713
2.4883	18.0	319266	2.6881	0.4750
2.418	19.0	337003	2.6616	0.4790
2.3343	20.0	354740	2.6457	0.4820

Framework versions

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