LHRS-UM-FERI/MENTHOS-dataset-rootcause

MENTHOS-dataset-rootcause

English

About

MENTHOS-dataset-rootcause is a binary log classification dataset built for root-cause identification experiments.

Source Data

• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/training-data/root-cause-training-data.csv
• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/validation-data/root-cause-validation-data-input.jsonlines
• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/training-data/root-cause-unseen-errors.csv

Processing and Balancing

• —Input sources are merged into a single dataframe (label, log).
• —Split into 70% train, 15% validation, 15% test.
• —Balancing: each split is balanced by downsampling both classes to the same count.

Splits and Class Distribution

The prepared train, validation, and test splits are included with the dataset release. | split | rows | label 0 | label 1 | | ---------- | ---: | ------: | ------: | | train | 1266 | 633 | 633 | | validation | 296 | 148 | 148 | | test | 296 | 148 | 148 |

Citation

@misc{borovic_li-dobnik_kranjec_ferme_2026,
  title        = {MENTHOS-dataset-rootcause},
  author       = {Borovic, Li Dobnik, Kranjec, Ferme},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/LHRS-UM-FERI/MENTHOS-dataset-rootcause}}
}

Slovenščina

O datasetu

MENTHOS-dataset-rootcause je binarni dataset zapisov za eksperimente zaznavanja koreninskih vzrokov.

Izvorni podatki

• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/training-data/root-cause-training-data.csv
• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/validation-data/root-cause-validation-data-input.jsonlines
• —https://github.com/nv-morpheus/Morpheus/raw/refs/heads/branch-25.10/models/datasets/training-data/root-cause-unseen-errors.csv

Obdelava in uravnoteženje

• —Vhodni viri se združijo v enoten format (label, log).
• —Razdelitev 70% train, 15% validation, 15% test.
• —Uravnoteženje: vsak split je uravnotežen z downsamplingom obeh razredov.

Delitve in porazdelitev razredov

Pripravljene train, validation in test delitve so vključene v izdajo nabora podatkov.

Citiranje

@misc{borovic_li-dobnik_kranjec_ferme_2026,
  title        = {MENTHOS-dataset-rootcause},
  author       = {Borovic, Li Dobnik, Kranjec, Ferme},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/LHRS-UM-FERI/MENTHOS-dataset-rootcause}}
}