MORES_emotions9
Model description
An xlm-roberta-large model finetuned on sentence-level multilingual training data hand-annotated using a nine-category emotion codebook.
Codebook
The model assigns the following labels:
- 0: "Anger"
- 1: "Fear"
- 2: "Disgust"
- 3: "Sadness"
- 4: "Joy"
- 5: "None of Them"
- 6: "Enthusiasm"
- 7: "Hope"
- 8: "Pride"
The training data we used was augmented with translated texts. It covers 7 languages (English, German, French, Polish, Slovak, Czech and Hungarian) with nearly identical shares.
How to use the model
from transformers import AutoTokenizer, pipeline
tokenizer = AutoTokenizer.from_pretrained("xlm-roberta-large")
pipe = pipeline(
model="MORES-horizon/MORES_emotions9",
task="text-classification",
tokenizer=tokenizer,
use_fast=False,
token="<your_hf_read_only_token>"
)
text = "We will place an immediate 6-month halt on the finance driven closure of beds and wards, and set up an independent audit of needs and facilities."
pipe(text)
Gated access
Due to the gated access, you must pass the token parameter when loading the model. In earlier versions of the Transformers package, you may need to use the use_auth_token parameter instead.
Model performance
Overall performance
- Accuracy: 0.72
- Macro avg: Precision 0.72, Recall 0.73, F1-score 0.72
- Weighted avg: Precision 0.72, Recall 0.72, F1-score 0.72
Per-class metrics
| Label | Precision | Recall | F1-score | Support |
|---|---|---|---|---|
| Anger | 0.55 | 0.54 | 0.55 | 5439 |
| Fear | 0.76 | 0.80 | 0.78 | 5432 |
| Disgust | 0.95 | 0.92 | 0.94 | 5432 |
| Sadness | 0.84 | 0.84 | 0.84 | 5425 |
| Joy | 0.80 | 0.81 | 0.80 | 5152 |
| None of Them | 0.69 | 0.65 | 0.67 | 11158 |
| Enthusiasm | 0.64 | 0.60 | 0.62 | 5432 |
| Hope | 0.50 | 0.59 | 0.54 | 5439 |
| Pride | 0.78 | 0.77 | 0.78 | 5432 |
Inference platform
This model is used by the Babel Machine, an open-source and free natural language processing tool, designed to simplify and speed up projects for comparative research.
Debugging and issues
This architecture uses the sentencepiece tokenizer. In order to use the model before transformers==4.27 you need to install it manually.
If you encounter a RuntimeError when loading the model using the from_pretrained() method, adding ignore_mismatched_sizes=True should solve the issue.
Citation
If you use this model in your research, please cite:
Szabó, G., Ring, O., Lipiński, A., & Beichelt, T. (2026). Protecting collective worth: pride in European parliament campaigns in Germany, Hungary, and Poland. Frontiers in Political Science, 8, 1828880. https://doi.org/10.3389/fpos.2026.1828880
@article{szabo2026pride,
author = {Szab{\'o}, Gabriella and Ring, Orsolya and Lipi{\'n}ski, Artur and Beichelt, Timm},
title = {Protecting collective worth: pride in European parliament campaigns in {Germany}, {Hungary}, and {Poland}},
journal = {Frontiers in Political Science},
year = {2026},
volume = {8},
pages = {1828880},
doi = {10.3389/fpos.2026.1828880},
url = {https://doi.org/10.3389/fpos.2026.1828880}
}
Funding
This project was funded by MORES (Moral Emotions in Politics. How they Divide, how they Unite) which received funding from the European Union under grant agreement No 101132601. However, views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.