BertWhyCommitOriginal

This model is a fine-tuned version of prajjwal1/bert-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4881
• Accuracy: 0.8788

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: 1e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 200

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	31	0.5058	0.7394
No log	2.0	62	0.4463	0.7758
No log	3.0	93	0.4260	0.7758
No log	4.0	124	0.3954	0.8061
No log	5.0	155	0.3745	0.8061
No log	6.0	186	0.3653	0.8303
No log	7.0	217	0.3533	0.8424
No log	8.0	248	0.3500	0.8364
No log	9.0	279	0.3416	0.8606
No log	10.0	310	0.3546	0.8424
No log	11.0	341	0.3469	0.8485
No log	12.0	372	0.3511	0.8606
No log	13.0	403	0.3883	0.8545
No log	14.0	434	0.4090	0.8485
No log	15.0	465	0.4301	0.8485
No log	16.0	496	0.4415	0.8606
0.2667	17.0	527	0.4732	0.8545
0.2667	18.0	558	0.4849	0.8727
0.2667	19.0	589	0.4881	0.8788

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3