For Spanish, we used TASS 2020 dataset. For English, we used SemEval 2017 Task 4 Subtask A.
Model | Lang | Neg f1 | Pos f1 | Neu f1 | Macro f1 |
---|---|---|---|---|---|
distilbert-en | en | 0.6228 | 0.6299 | 0.6725 | 0.6417 |
mbert-en | 0.6655 | 0.6291 | 0.6340 | 0.6428 | |
roberta-base | 0.7022 | 0.6657 | 0.6829 | 0.6836 | |
bert-base | 0.6932 | 0.6701 | 0.6885 | 0.6839 | |
bertweet-base | 0.7333 | 0.6784 | 0.6774 | 0.6964 | |
---- | ---- | ---- | ---- | ----- | ---- |
distilbert-es | es | 0.6492 | 0.6678 | 0.4799 | 0.5990 |
mbert-es | 0.6678 | 0.6694 | 0.4756 | 0.6043 | |
beto | 0.7289 | 0.7349 | 0.5356 | 0.6665 |
For both languages we used EmoEvent datasets.
Model | Lang | Others f1 | Joy f1 | Sadness f1 | Anger f1 | Surprise f1 | Disgust f1 | Fear f1 | Macro f1 |
---|---|---|---|---|---|---|---|---|---|
distilbert-en | en | 0.5094 | 0.6654 | 0.5021 | 0.3030 | 0.1475 | 0.3511 | 0.2018 | 0.3829 |
mbert-en | 0.5477 | 0.6506 | 0.5446 | 0.2970 | 0.1029 | 0.3569 | 0.2553 | 0.3936 | |
bert-base | 0.5939 | 0.6835 | 0.5414 | 0.3557 | 0.2384 | 0.4088 | 0.2526 | 0.4392 | |
roberta-base | 0.5738 | 0.6895 | 0.5479 | 0.3636 | 0.2123 | 0.4726 | 0.2558 | 0.4451 | |
bertweet-base | 0.6060 | 0.7110 | 0.6086 | 0.4338 | 0.2576 | 0.4524 | 0.2600 | 0.4756 | |
---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
distilbert-es | es | 0.6789 | 0.6022 | 0.7167 | 0.4343 | 0.2643 | 0.0845 | 0.4615 | 0.4632 |
mbert-es | 0.6911 | 0.5852 | 0.7104 | 0.4905 | 0.2732 | 0.1333 | 0.4313 | 0.4736 | |
beto | 0.7493 | 0.6719 | 0.7536 | 0.5736 | 0.3966 | 0.1538 | 0.5333 | 0.5475 |