nikolamilosevic/SCIFACT_xlm_roberta_large

SCIFACT_inference_model

This model is a fine-tuned version of xlm-roberta-large on the SciFact dataset. It achieves the following results on the evaluation set:

• Loss: 1.2496
• Accuracy: 0.8819

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: 3
• eval_batch_size: 3
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	378	1.0485	0.4724
1.0382	2.0	756	1.3964	0.6063
0.835	3.0	1134	0.9168	0.8268
0.6801	4.0	1512	0.7524	0.8425
0.6801	5.0	1890	1.0672	0.8346
0.4291	6.0	2268	0.9599	0.8425
0.2604	7.0	2646	0.8691	0.8661
0.1932	8.0	3024	1.3162	0.8268
0.1932	9.0	3402	1.3200	0.8583
0.0974	10.0	3780	1.1566	0.8740
0.1051	11.0	4158	1.1568	0.8819
0.0433	12.0	4536	1.2013	0.8661
0.0433	13.0	4914	1.1557	0.8819
0.034	14.0	5292	1.3044	0.8661
0.0303	15.0	5670	1.2496	0.8819

Framework versions

• Transformers 4.34.1
• Pytorch 1.13.1+cu116
• Datasets 2.14.6
• Tokenizers 0.14.1