AdvPicker Extension for Cross-Lingual Zero-Shot NER with Multilingual Pretrained Models

Multilingual BERT (mBERT), a language model pre-trained on large multilingual corpora, has impressive zero-shot cross-lingual transfer capabilities and performs surprisingly well on zero-shot POS tagging and Named Entity Recognition (NER), as well as on cross-lingual model transfer. At present, the mainstream methods to solve the cross-lingual downstream tasks are always using the last transformer layer's output of mBERT as the representation of linguistic information. In this work, we explore the complementary property of lower layers to the last transformer layer of mBERT. A feature aggregat

Research goal: Can AdvPicker's approach be extended to improve cross-lingual zero-shot NER performance on the WNUT-17 or CoNLL-2003 benchmarks when using multilingual pretrained models (e.g., Multilingual BERT)?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.