File size: 4,166 Bytes
3eda1c6 e5e2561 3eda1c6 e5e2561 3eda1c6 e5e2561 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 |
---
license: apache-2.0
base_model: jackaduma/SecBERT
tags:
- generated_from_trainer
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: Cyber-ThreaD/SecBERT-AttackER
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Cyber-ThreaD/SecBERT-AttackER
This model is a fine-tuned version of [jackaduma/SecBERT](https://huggingface.co/jackaduma/SecBERT) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6932
- Precision: 0.3931
- Recall: 0.4987
- F1: 0.4397
- Accuracy: 0.7295
## 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: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 1.7927 | 0.4 | 500 | 1.5607 | 0.0956 | 0.0780 | 0.0859 | 0.6139 |
| 1.3551 | 0.81 | 1000 | 1.3530 | 0.2064 | 0.2381 | 0.2211 | 0.6495 |
| 1.0432 | 1.21 | 1500 | 1.3107 | 0.2269 | 0.3082 | 0.2614 | 0.6740 |
| 0.8468 | 1.61 | 2000 | 1.2497 | 0.2447 | 0.3373 | 0.2836 | 0.6767 |
| 0.7775 | 2.01 | 2500 | 1.2710 | 0.2895 | 0.3730 | 0.3260 | 0.6939 |
| 0.5374 | 2.42 | 3000 | 1.3020 | 0.3006 | 0.4048 | 0.3450 | 0.7044 |
| 0.5071 | 2.82 | 3500 | 1.2614 | 0.2959 | 0.4048 | 0.3419 | 0.7081 |
| 0.4237 | 3.22 | 4000 | 1.3251 | 0.3367 | 0.4405 | 0.3817 | 0.7166 |
| 0.3597 | 3.63 | 4500 | 1.3853 | 0.3423 | 0.4524 | 0.3897 | 0.7125 |
| 0.3632 | 4.03 | 5000 | 1.4156 | 0.3559 | 0.4524 | 0.3984 | 0.7127 |
| 0.2589 | 4.43 | 5500 | 1.4472 | 0.3504 | 0.4709 | 0.4018 | 0.7173 |
| 0.323 | 4.83 | 6000 | 1.3997 | 0.3452 | 0.4603 | 0.3946 | 0.7222 |
| 0.2167 | 5.24 | 6500 | 1.5194 | 0.3467 | 0.4590 | 0.3950 | 0.7233 |
| 0.2363 | 5.64 | 7000 | 1.5585 | 0.3507 | 0.4722 | 0.4025 | 0.7222 |
| 0.2721 | 6.04 | 7500 | 1.5420 | 0.3715 | 0.4854 | 0.4209 | 0.7210 |
| 0.2073 | 6.45 | 8000 | 1.5878 | 0.3536 | 0.4854 | 0.4091 | 0.7147 |
| 0.2021 | 6.85 | 8500 | 1.6637 | 0.3722 | 0.4854 | 0.4214 | 0.7197 |
| 0.1648 | 7.25 | 9000 | 1.6724 | 0.3795 | 0.4788 | 0.4234 | 0.7255 |
| 0.1927 | 7.66 | 9500 | 1.6891 | 0.3801 | 0.4947 | 0.4299 | 0.7245 |
| 0.1958 | 8.06 | 10000 | 1.6774 | 0.3937 | 0.4974 | 0.4395 | 0.7281 |
| 0.1508 | 8.46 | 10500 | 1.7379 | 0.3815 | 0.4854 | 0.4272 | 0.7259 |
| 0.184 | 8.86 | 11000 | 1.7001 | 0.3863 | 0.5013 | 0.4364 | 0.7277 |
| 0.1696 | 9.27 | 11500 | 1.6932 | 0.3931 | 0.4987 | 0.4397 | 0.7295 |
| 0.1425 | 9.67 | 12000 | 1.7137 | 0.3824 | 0.5013 | 0.4339 | 0.7276 |
### Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0
### Citing & Authors
If you use the model kindly cite the following work
```
@inproceedings{deka2024attacker,
title={AttackER: Towards Enhancing Cyber-Attack Attribution with a Named Entity Recognition Dataset},
author={Deka, Pritam and Rajapaksha, Sampath and Rani, Ruby and Almutairi, Amirah and Karafili, Erisa},
booktitle={International Conference on Web Information Systems Engineering},
pages={255--270},
year={2024},
organization={Springer}
}
```
|