Models for parsing SNACS datasets. See Schneider et al. (2018).
See model finetuning runs on Weights & Biases.
English
| Model | W&B | Dataset | Precision | Recall | F1 |
|---|---|---|---|---|---|
bert-base-cased + Linear |
polished-plasma-24 |
en-streusle |
70.9 | 72.3 | 71.6 |
roberta-base + Linear |
feasible-dragon-20 |
en-streusle |
77.0 | 79.6 | 78.2 |
| Liu et al. (2021) | en-streusle |
70.9 | |||
| Schneider et al. (2018) | en-streusle |
55.7 | |||
bert-base-cased + Linear |
confused-elevator-22 |
en-lp |
67.4 | 70.1 | 68.7 |
roberta-base + Linear |
pleasant-salad-21 |
en-lp |
66.8 | 69.4 | 68.1 |
roberta-base + Linear |
youthful-frog-30 |
en-pastrie |
53.9 | 57.7 | 55.7 |
Hindi
| Model | W&B | Dataset | Precision | Recall | F1 |
|---|---|---|---|---|---|
neuralspace-reverie/indic-transformers-hi-roberta + Linear |
soft-butterfly-33 |
hi-lp |
61.3 | 63.3 | 62.3 |