bert-finetuned-ner

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6370
• Precision: 0.5313
• Recall: 0.4530
• F1: 0.4891
• Accuracy: 0.9290

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	125	0.5387	0.2190	0.0552	0.0882	0.8991
No log	2.0	250	0.4241	0.3430	0.1750	0.2317	0.9117
No log	3.0	375	0.4721	0.3502	0.1786	0.2366	0.9088
0.1529	4.0	500	0.6204	0.4300	0.2320	0.3014	0.9134
0.1529	5.0	625	0.6479	0.4470	0.2486	0.3195	0.9104
0.1529	6.0	750	0.4640	0.4532	0.4015	0.4258	0.9220
0.1529	7.0	875	0.5170	0.4288	0.4217	0.4253	0.9224
0.0229	8.0	1000	0.5846	0.5524	0.4273	0.4818	0.9233
0.0229	9.0	1125	0.5569	0.4644	0.4328	0.4480	0.9234
0.0229	10.0	1250	0.5818	0.5502	0.4438	0.4913	0.9258
0.0229	11.0	1375	0.6183	0.5607	0.4254	0.4838	0.9231
0.0048	12.0	1500	0.6148	0.5385	0.4254	0.4753	0.9250
0.0048	13.0	1625	0.6271	0.4896	0.4328	0.4594	0.9255
0.0048	14.0	1750	0.6475	0.5668	0.4217	0.4836	0.9267
0.0048	15.0	1875	0.6428	0.5704	0.4328	0.4921	0.9282
0.0016	16.0	2000	0.6577	0.5487	0.4254	0.4793	0.9270
0.0016	17.0	2125	0.6688	0.5556	0.4144	0.4747	0.9262
0.0016	18.0	2250	0.6481	0.5434	0.4383	0.4852	0.9282
0.0016	19.0	2375	0.6432	0.5428	0.4438	0.4883	0.9289
0.0007	20.0	2500	0.6370	0.5313	0.4530	0.4891	0.9290

Framework versions

• Transformers 4.23.1
• Pytorch 1.8.0
• Datasets 2.6.1
• Tokenizers 0.13.1