Safety without alignment

no code implementations • 27 Feb 2023 • András Kornai, Michael Bukatin, Zsolt Zombori

Currently, the dominant paradigm in AI safety is alignment with human values.

Ethics

Evaluating Transferability of BERT Models on Uralic Languages

1 code implementation • ACL (IWCLUL) 2021 • Judit Ács, Dániel Lévai, András Kornai

Transformer-based language models such as BERT have outperformed previous models on a large number of English benchmarks, but their evaluation is often limited to English or a small number of well-resourced languages.

Hyperparameter Optimization NER +1

Subword Pooling Makes a Difference

1 code implementation • 22 Feb 2021 • Judit Ács, Ákos Kádár, András Kornai

For POS tagging both of these strategies perform poorly and the best choice is to use a small LSTM over the subwords.

NER POS +1

Evaluating Contextualized Language Models for Hungarian

1 code implementation • 22 Feb 2021 • Judit Ács, Dániel Lévai, Dávid Márk Nemeskey, András Kornai

We present an extended comparison of contextualized language models for Hungarian.

NER POS +1

Sentence Length

1 code implementation • 22 May 2019 • Gábor Borbély, András Kornai

The distribution of sentence length in ordinary language is not well captured by the existing models.

Sentence

Edit wars in Wikipedia

1 code implementation • 19 Jul 2011 • Róbert Sumi, Taha Yasseri, András Rung, András Kornai, János Kertész

We present a new, efficient method for automatically detecting severe conflicts `edit wars' in Wikipedia and evaluate this method on six different language WPs.