UNER_subword_tk_en_lora_alpha_32_drop_0.3_rank_16_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: 0.0580
• Precision: 0.7562
• Recall: 0.8251
• F1: 0.7891
• Accuracy: 0.9834

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	0.1118	0.3833	0.5611	0.4555	0.9625
0.2147	2.0	784	0.0715	0.6765	0.7857	0.7270	0.9791
0.062	3.0	1176	0.0660	0.6973	0.8157	0.7519	0.9811
0.0474	4.0	1568	0.0591	0.7283	0.7909	0.7583	0.9821
0.0474	5.0	1960	0.0606	0.7078	0.8251	0.7620	0.9809
0.0424	6.0	2352	0.0573	0.7428	0.8251	0.7818	0.9830
0.0379	7.0	2744	0.0572	0.7248	0.8178	0.7685	0.9823
0.0354	8.0	3136	0.0571	0.7524	0.8209	0.7851	0.9836
0.0335	9.0	3528	0.0548	0.7481	0.8209	0.7828	0.9835
0.0335	10.0	3920	0.0585	0.7479	0.8137	0.7794	0.9833
0.0309	11.0	4312	0.0561	0.7474	0.8209	0.7824	0.9834
0.0295	12.0	4704	0.0567	0.7512	0.8188	0.7836	0.9837
0.0292	13.0	5096	0.0552	0.7708	0.8147	0.7921	0.9843
0.0292	14.0	5488	0.0565	0.7421	0.8282	0.7828	0.9835
0.0276	15.0	5880	0.0572	0.7545	0.8178	0.7849	0.9837
0.0266	16.0	6272	0.0585	0.7567	0.8240	0.7889	0.9837
0.0251	17.0	6664	0.0576	0.7601	0.8199	0.7888	0.9837
0.0257	18.0	7056	0.0580	0.7567	0.8240	0.7889	0.9835
0.0257	19.0	7448	0.0582	0.7526	0.8313	0.7900	0.9836
0.0251	20.0	7840	0.0580	0.7562	0.8251	0.7891	0.9834

Framework versions

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