lilt_xlmroberta_lr2e-05_bs4_train30

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

• Loss: 0.1755
• Precision: 0.9016
• Recall: 0.8981
• F1: 0.8999
• Accuracy: 0.9525

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	15	1.6030	0.0	0.0	0.0	0.5666
No log	2.0	30	1.2972	1.0	0.0005	0.0010	0.5668
No log	3.0	45	1.1206	0.4578	0.0183	0.0353	0.5745
No log	4.0	60	0.9735	0.3883	0.2776	0.3238	0.6848
No log	5.0	75	0.8024	0.4795	0.4573	0.4681	0.7593
No log	6.0	90	0.6347	0.6115	0.5958	0.6036	0.8208
No log	7.0	105	0.5102	0.6957	0.6746	0.6850	0.8568
No log	8.0	120	0.4197	0.7380	0.7262	0.7321	0.8780
No log	9.0	135	0.3716	0.7815	0.7668	0.7741	0.8956
No log	10.0	150	0.3196	0.7982	0.8078	0.8030	0.9081
No log	11.0	165	0.2879	0.8407	0.8281	0.8343	0.9226
No log	12.0	180	0.2455	0.8720	0.8687	0.8703	0.9393
No log	13.0	195	0.2277	0.8726	0.8667	0.8697	0.9385
No log	14.0	210	0.2241	0.88	0.8711	0.8755	0.9412
No log	15.0	225	0.2029	0.8840	0.8865	0.8852	0.9458
No log	16.0	240	0.2097	0.8913	0.8788	0.8850	0.9452
No log	17.0	255	0.1949	0.8967	0.8923	0.8945	0.9496
No log	18.0	270	0.2004	0.9008	0.8856	0.8931	0.9483
No log	19.0	285	0.1843	0.9024	0.8972	0.8998	0.9523
No log	20.0	300	0.1866	0.9013	0.8952	0.8983	0.9514
No log	21.0	315	0.1755	0.9016	0.8981	0.8999	0.9525
No log	22.0	330	0.1829	0.8994	0.8981	0.8988	0.9523
No log	23.0	345	0.1768	0.9056	0.9034	0.9045	0.9546
No log	24.0	360	0.1785	0.9094	0.9068	0.9081	0.9560

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1