UNER_subword_tk_en_lora_alpha_16_drop_0.3_rank_8_seed_42

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

• Loss: 0.0558
• Precision: 0.7538
• Recall: 0.8178
• F1: 0.7845
• Accuracy: 0.9838

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	392	0.1379	0.3148	0.3571	0.3346	0.9554
0.2595	2.0	784	0.0790	0.6156	0.7412	0.6726	0.9759
0.0724	3.0	1176	0.0705	0.6757	0.7981	0.7318	0.9791
0.0532	4.0	1568	0.0617	0.7274	0.7899	0.7573	0.9818
0.0532	5.0	1960	0.0628	0.7094	0.8085	0.7557	0.9810
0.0472	6.0	2352	0.0578	0.7378	0.8157	0.7748	0.9827
0.0427	7.0	2744	0.0586	0.7314	0.8230	0.7745	0.9819
0.0398	8.0	3136	0.0586	0.7297	0.8188	0.7717	0.9823
0.038	9.0	3528	0.0571	0.7378	0.8271	0.7799	0.9825
0.038	10.0	3920	0.0578	0.7304	0.8106	0.7684	0.9829
0.0358	11.0	4312	0.0562	0.7380	0.8137	0.7740	0.9827
0.0344	12.0	4704	0.0559	0.7408	0.8168	0.7770	0.9833
0.0339	13.0	5096	0.0554	0.7465	0.8168	0.7800	0.9835
0.0339	14.0	5488	0.0567	0.7275	0.8209	0.7714	0.9827
0.0321	15.0	5880	0.0556	0.7533	0.8188	0.7847	0.9838
0.0318	16.0	6272	0.0562	0.7493	0.8199	0.7830	0.9838
0.0303	17.0	6664	0.0551	0.7569	0.8188	0.7867	0.9840
0.0307	18.0	7056	0.0563	0.7555	0.8157	0.7845	0.9837
0.0307	19.0	7448	0.0561	0.7479	0.8230	0.7836	0.9836
0.0303	20.0	7840	0.0558	0.7538	0.8178	0.7845	0.9838

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1