bert-large-cased-sigir-support-refute-no-label-40-2nd-test-LR10-40

This model is a fine-tuned version of jojoUla/bert-large-cased-sigir-support-refute-no-label-40 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1048

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: 4e-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: 20.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
4.9062	1.0	1	5.6418
4.5785	2.0	2	5.3839
6.6562	3.0	3	5.0883
4.0274	4.0	4	4.4272
2.9225	5.0	5	4.1994
1.9388	6.0	6	2.9638
2.6745	7.0	7	2.4477
2.0988	8.0	8	2.4030
2.3506	9.0	9	3.5475
1.734	10.0	10	0.1426
1.8435	11.0	11	2.2994
1.5274	12.0	12	1.5195
1.5668	13.0	13	1.3508
1.4771	14.0	14	1.5684
1.4649	15.0	15	0.0011
1.0896	16.0	16	2.2005
0.9002	17.0	17	0.0748
1.2433	18.0	18	0.4664
1.4224	19.0	19	1.5759
0.791	20.0	20	0.4863

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2