metadata
license: mit
language:
- cs
- de
- en
- hu
- pl
metrics:
- f1
- precision
- recall
- accuracy
library_name: transformers
pipeline_tag: text-classification
tags:
- multilabel
- history
- holocaust
- heritage
base_model:
- sentence-transformers/LaBSE
LaBSE-Malach-Multilabel
A multilabel text classification model fine-tuned on a the Visual History Archive in 6 languages. Input text segments consisted of ~350 words on average.
Given an input string, the model predicts probablites for 2800 subject keyword IDs from the VHA ontology.