dhruvyadav89300/BERT-phishing-classifier

Model Card for Model ID

Model Details

Model Description

This is the model is a fine-tuned version of google-bert/bert-base-uncased on the phishing-site-classification dataset

Model Sources

• Repository: GitHub

Evaluation

Training Results

Epoch	Training Loss	Step	Validation Loss	Accuracy	AUC	Learning Rate
1	0.4932	263	0.4237	0.789	0.912	0.00019
2	0.3908	526	0.3761	0.824	0.932	0.00018
3	0.3787	789	0.3136	0.860	0.941	0.00017
4	0.3606	1052	0.4401	0.818	0.944	0.00016
5	0.3545	1315	0.2928	0.864	0.947	0.00015
6	0.3600	1578	0.3406	0.867	0.949	0.00014
7	0.3233	1841	0.2897	0.869	0.950	0.00013
8	0.3411	2104	0.3328	0.871	0.949	0.00012
9	0.3292	2367	0.3189	0.876	0.954	0.00011
10	0.3239	2630	0.3685	0.849	0.956	0.00010
11	0.3201	2893	0.3317	0.862	0.956	0.00009
12	0.3335	3156	0.2725	0.869	0.957	0.00008
13	0.3230	3419	0.2856	0.882	0.955	0.00007
14	0.3087	3682	0.2900	0.882	0.957	0.00006
15	0.3050	3945	0.2704	0.893	0.957	0.00005
16	0.3032	4208	0.2662	0.878	0.957	0.00004
17	0.3027	4471	0.2930	0.882	0.956	0.00003
18	0.2950	4734	0.2707	0.880	0.957	0.00002
19	0.2998	4997	0.2782	0.884	0.957	0.00001
20	0.2971	5260	0.2792	0.882	0.957	0.00000

Final Training Summary

• Total Training Runtime: 555.4381 seconds
• Final Training Loss: 0.3372
• Train Samples per Second: 75.616
• Eval Accuracy (Best Epoch): 0.893 (Epoch 15)
• Eval AUC (Best Epoch): 0.957 (Multiple Epochs)