SAE-bert-base-uncased

This model is a fine-tuned version of bert-base-uncased on the jgammack/SAE-door-abstracts dataset.

It achieves the following results on the evaluation set:

• Loss: 2.1256

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: 2e-05
• train_batch_size: 7
• eval_batch_size: 7
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.5967	1.0	80	2.3409
2.4881	2.0	160	2.2707
2.3567	3.0	240	2.3134
2.3413	4.0	320	2.2592
2.3006	5.0	400	2.2351
2.2568	6.0	480	2.2556
2.2303	7.0	560	2.2546
2.1892	8.0	640	2.1868
2.1851	9.0	720	2.2073
2.1738	10.0	800	2.1344
2.1673	11.0	880	2.1927
2.1518	12.0	960	2.1844
2.1142	13.0	1040	2.1466
2.1343	14.0	1120	2.2024
2.1332	15.0	1200	2.1035

Framework versions

• Transformers 4.16.2
• Pytorch 1.10.0+cu111
• Datasets 1.18.3
• Tokenizers 0.11.0