xlm-roberta-base-finetuned_60kURL

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3703
• Accuracy: 0.9514
• F1: 0.9515

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.3136	1.0	704	0.1763	0.9341	0.9341
0.1735	2.0	1408	0.1522	0.9429	0.9431
0.1389	3.0	2112	0.1497	0.9475	0.9478
0.1155	4.0	2816	0.1724	0.9462	0.9467
0.0979	5.0	3520	0.1662	0.9514	0.9517
0.0828	6.0	4224	0.1598	0.9474	0.9480
0.0695	7.0	4928	0.1715	0.9511	0.9513
0.0574	8.0	5632	0.2141	0.9482	0.9485
0.051	9.0	6336	0.2399	0.9497	0.9500
0.0418	10.0	7040	0.2441	0.9488	0.9491
0.0359	11.0	7744	0.3052	0.9495	0.9497
0.0314	12.0	8448	0.2626	0.9523	0.9524
0.0288	13.0	9152	0.3100	0.9514	0.9516
0.0227	14.0	9856	0.3394	0.9502	0.9503
0.0214	15.0	10560	0.3331	0.9498	0.9499
0.0184	16.0	11264	0.3550	0.9510	0.9512
0.0168	17.0	11968	0.3506	0.9507	0.9508
0.0142	18.0	12672	0.3667	0.9524	0.9525
0.0135	19.0	13376	0.3689	0.9514	0.9515
0.0136	20.0	14080	0.3703	0.9514	0.9515

Framework versions

• Transformers 4.41.1
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.19.1