bert-large-uncased-finetuned-ner-geocorpus

This model is a fine-tuned version of google-bert/bert-large-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1293
• Precision: 0.8171
• Recall: 0.8806
• F1: 0.8476
• Accuracy: 0.9721

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	0.9955	137	0.2292	0.4527	0.3450	0.3916	0.9378
No log	1.9982	275	0.1339	0.6814	0.7175	0.6990	0.9606
No log	2.9936	412	0.1147	0.7385	0.8057	0.7706	0.9647
0.2052	3.9964	550	0.1217	0.7099	0.8607	0.7781	0.9611
0.2052	4.9991	688	0.1076	0.7705	0.8531	0.8097	0.9674
0.2052	5.9946	825	0.1130	0.7970	0.8483	0.8219	0.9701
0.2052	6.9973	963	0.1332	0.7357	0.8758	0.7997	0.9637
0.0384	8.0	1101	0.1241	0.7798	0.8929	0.8325	0.9690
0.0384	8.9955	1238	0.1241	0.8303	0.8720	0.8507	0.9728
0.0384	9.9546	1370	0.1293	0.8171	0.8806	0.8476	0.9721

Framework versions

• Transformers 4.41.1
• Pytorch 2.1.2
• Datasets 2.19.1
• Tokenizers 0.19.1