File size: 716 Bytes
3e8baa2 5a76ef4 3e8baa2 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
---
datasets:
- universal_dependencies
language:
- tl
metrics:
- f1
pipeline_tag: token-classification
---
## Model Specification
- Model: XLM-RoBERTa (base-sized model)
- Training Data:
- Combined Afrikaans, Hebrew, Bulgarian, Vietnamese, Norwegian, Urdu, & Czech corpora (Top 7 Languages)
- Training Details:
- Base configurations with learning rate 5e-5
## Evaluation
- Evaluation Dataset: Universal Dependencies Tagalog Ugnayan (Testing Set)
- Tested in a zero-shot cross-lingual scenario on a Universal Dependencies Tagalog Ugnayan testing dataset (with 78.81\% Accuracy)
## POS Tags
- ADJ – ADP – ADV – CCONJ – DET – INTJ – NOUN – NUM – PART – PRON – PROPN – PUNCT – SCONJ – VERB |