LEIA/LEIA-multilingual

1---2license: mit3language:4- multilingual5library_name: transformers6pipeline_tag: text-classification7widget:8- text: "You wont believe what happened to me today"9- text: "You wont believe what happened to me today!"10- text: "You wont believe what happened to me today..."11- text: "You wont believe what happened to me today <3"12- text: "You wont believe what happened to me today :)"13- text: "You wont believe what happened to me today :("14---15This is an emotion classification model based on fine-tuning of a Bernice model, which is a pre-trained model trained on multilingual Twitter data.16The fine-tuning dataset is a subset of the self-labeled emotion dataset (Lykousas et al., 2019) in English that corresponds to Anger, Fear, Sadness, Joy, and Affection. 17See the paper, [LEIA: Linguistic Embeddings for the Identification of Affect](https://doi.org/10.1140/epjds/s13688-023-00427-0) for further details. 18 19## Evaluation20We evaluated LEIA-multilingual on posts with self-annotated emotion labels identified as non-English using an ensemble of language identification tools. 21The table below shows the macro-F1 scores aggregated across emotion categories for each language:22 23 24  25|Language|Macro-F1|26|:---:|:---:|27|ar |44.18[43.07,45.29]|28|da |65.44[60.96,69.83] |29|de |60.47[57.58,63.38] |30|es |61.67[60.79,62.55] |31|fi |45.1[40.96,49.14] |32|fr |65.78[63.19,68.36] |33|it |63.37[59.67,67.1] |34|pt |57.27[55.15,59.4] |35|tl |58.37[55.51,61.23] |36|tr |45.42[41.17,49.79]|37 38 39 40## Citation41Please cite the following paper if you find the model useful for your work:42```bibtex43@article{aroyehun2023leia,44  title={LEIA: Linguistic Embeddings for the Identification of Affect},45  author={Aroyehun, Segun Taofeek and Malik, Lukas and Metzler, Hannah and Haimerl, Nikolas and Di Natale, Anna and Garcia, David},46  journal={EPJ Data Science},47  volume={12},48  year={2023},49  publisher={Springer}50}51```