scibert-base-uncased-ner-tuning

This model is a fine-tuned version of allenai/scibert_scivocab_uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0055
• Cases: {'precision': 1.0, 'recall': 0.9984615384615385, 'f1': 0.9992301770592764, 'number': 650}
• Country: {'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}
• Date: {'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}
• Deaths: {'precision': 0.9978260869565218, 'recall': 0.9956616052060737, 'f1': 0.996742671009772, 'number': 461}
• Virus: {'precision': 0.998995983935743, 'recall': 0.9979939819458375, 'f1': 0.9984947315604615, 'number': 997}
• Overall Precision: 0.9990
• Overall Recall: 0.9976
• Overall F1: 0.9983
• Overall Accuracy: 0.9992

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: 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Cases	Country	Date	Deaths	Virus	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
No log	1.0	291	0.0042	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 650}	{'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}	{'precision': 1.0, 'recall': 0.9978308026030369, 'f1': 0.998914223669924, 'number': 461}	{'precision': 0.9969727547931383, 'recall': 0.9909729187562688, 'f1': 0.993963782696177, 'number': 997}	0.9987	0.9960	0.9973	0.9990
0.0545	2.0	582	0.0055	{'precision': 1.0, 'recall': 0.9953846153846154, 'f1': 0.9976869699306091, 'number': 650}	{'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}	{'precision': 0.9935344827586207, 'recall': 1.0, 'f1': 0.9967567567567568, 'number': 461}	{'precision': 0.9979879275653923, 'recall': 0.9949849548645938, 'f1': 0.9964841788046208, 'number': 997}	0.9980	0.9966	0.9973	0.9990
0.0545	3.0	873	0.0071	{'precision': 1.0, 'recall': 0.9969230769230769, 'f1': 0.9984591679506933, 'number': 650}	{'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}	{'precision': 0.9956331877729258, 'recall': 0.9891540130151844, 'f1': 0.9923830250272034, 'number': 461}	{'precision': 0.9979879275653923, 'recall': 0.9949849548645938, 'f1': 0.9964841788046208, 'number': 997}	0.9983	0.9953	0.9968	0.9987
0.0035	4.0	1164	0.0051	{'precision': 1.0, 'recall': 0.9984615384615385, 'f1': 0.9992301770592764, 'number': 650}	{'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}	{'precision': 0.9978260869565218, 'recall': 0.9956616052060737, 'f1': 0.996742671009772, 'number': 461}	{'precision': 0.998995983935743, 'recall': 0.9979939819458375, 'f1': 0.9984947315604615, 'number': 997}	0.9990	0.9976	0.9983	0.9992
0.0035	5.0	1455	0.0055	{'precision': 1.0, 'recall': 0.9984615384615385, 'f1': 0.9992301770592764, 'number': 650}	{'precision': 0.9986824769433466, 'recall': 0.9973684210526316, 'f1': 0.9980250164581962, 'number': 760}	{'precision': 1.0, 'recall': 1.0, 'f1': 1.0, 'number': 109}	{'precision': 0.9978260869565218, 'recall': 0.9956616052060737, 'f1': 0.996742671009772, 'number': 461}	{'precision': 0.998995983935743, 'recall': 0.9979939819458375, 'f1': 0.9984947315604615, 'number': 997}	0.9990	0.9976	0.9983	0.9992

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu121
• Datasets 3.5.0
• Tokenizers 0.21.1