metadata

language:
  - en
license: apache-2.0
size_categories:
  - 100K<n<1M
task_categories:
  - translation
  - text2text-generation
tags:
  - cybersecurity
dataset_info:
  features:
    - name: input_text
      dtype: string
    - name: output_text
      dtype: string
    - name: instruction
      dtype: string
  splits:
    - name: train
      num_bytes: 300289090
      num_examples: 155304
  download_size: 33462385
  dataset_size: 300289090
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

syslog-to-artifact Dataset

The dataset is constructed of 100k+ syslog messages from 150+ message families. Syslog sources include linux, windows, Cisco, Checkpoint, Palo Alto, and more than 100 other products. The purpose of the instruction dataset is to train models to translate the meaning of syslog messages to English.

Instructions

The following instructions are used in the dataset:

Create a JSON artifact from the message
Identify this syslog message
Explain this syslog message

Data Creation

Data was generated from 500,000,000+ WitFoo Precinct Artifacts. Data was sanitized and organized using the ArtiFish toolkit: https://ArtiFish.dev