scenario-KD-SCR-DIV2-data-glue-qnli-model-bert-base-uncased-run-3

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

• Loss: 1.8270
• Accuracy: 0.8596

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 45
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6969

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.5227	1.0	3273	1.6868	0.8596
1.6366	2.0	6547	1.6143	0.8539
1.1125	3.0	9820	1.7793	0.8521
0.8908	4.0	13094	1.6376	0.8642
0.7217	5.0	16367	1.7602	0.8545
0.673	6.0	19641	1.6412	0.8634
0.5869	7.0	22914	1.6810	0.8611
0.5455	8.0	26188	1.5553	0.8662
0.506	9.0	29461	1.6137	0.8666
0.4771	10.0	32735	1.6643	0.8616
0.448	11.0	36008	1.5750	0.8666
0.4227	12.0	39282	1.5342	0.8678
0.4109	13.0	42555	1.6987	0.8638
0.3761	14.0	45829	1.6144	0.8645
0.3548	15.0	49102	1.6813	0.8605
0.3471	16.0	52376	1.6060	0.8656
0.3511	17.0	55649	1.8270	0.8596

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2
• Datasets 2.16.0
• Tokenizers 0.15.0