This model is the baseline higher-order values Openness to Change and Conservation detector from the papers:
Human Values in a Single Sentence: Moral Presence, Hierarchies, and Transformer Ensembles on the Schwartz Continuum
Víctor Yeste, Paolo Rosso (2026), arXiv:2601.14172
Do Schwartz Higher-Order Values Help Sentence-Level Human Value Detection? When Hard Gating Hurts
Víctor Yeste, Paolo Rosso (2026), arXiv:2602.00913
It is a multi-label classifier over the higher-order values Openness to Change and Conservation, trained on the English, machine-translated portion of the ValueEval'24 / ValuesML corpus.
This is the text-only DeBERTa-base baseline used in the paper.
The model was not trained or audited for safety-critical or high-stakes decision-making.
The 19 labels follow the refined Schwartz value continuum:
Self-direction: thought Self-direction: action Stimulation Hedonism Achievement Power: dominance Power: resources Face Security: personal Security: societal Tradition Conformity: rules Conformity: interpersonal Humility Benevolence: caring Benevolence: dependability Universalism: concern Universalism: nature Universalism: toleranceThe values categorized under the higher-order values Openness to Change and Conservation are collapsed into that binary labels:;
{
"Openness to Change": [
"Self-direction: thought attained", "Self-direction: thought constrained",
"Self-direction: action attained", "Self-direction: action constrained",
"Stimulation attained", "Stimulation constrained",
"Hedonism attained", "Hedonism constrained"
],
"Conservation": [
"Face attained", "Face constrained",
"Security: personal attained", "Security: personal constrained",
"Security: societal attained", "Security: societal constrained",
"Tradition attained", "Tradition constrained",
"Conformity: rules attained", "Conformity: rules constrained",
"Conformity: interpersonal attained", "Conformity: interpersonal constrained",
"Humility attained", "Humility constrained"
]
}
Because this model uses a custom architecture (EnhancedDebertaForSequenceClassification with extra feature inputs), it is loaded via AutoModelForSequenceClassification(..., trust_remote_code=True) rather than the generic pipeline("text-classification"), which only supports a fixed list of built-in model classes.
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
model_id = "VictorYeste/openness-conservation-deberta-baseline"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(
model_id,
trust_remote_code=True, # important for custom model code
)
values = [
"Openness to Change",
"Conservation"
]
id2label = {i: label for i, label in enumerate(values)}
def predict_values(text, threshold=0.50):
enc = tokenizer(text, return_tensors="pt", truncation=True)
with torch.no_grad():
outputs = model(**enc)
logits = outputs.logits.squeeze(0) # (2,)
probs = torch.sigmoid(logits) # tensor of shape (2,)
probs = probs.cpu().numpy()
active = probs >= threshold
active_labels = [id2label[i] for i, is_on in enumerate(active) if is_on]
return {
"probs": {id2label[i]: float(p) for i, p in enumerate(probs)},
"labels": active_labels,
}
example = "We must be brave enough to change the world and leave it better for our children."
print(predict_values(example))
This will return something like:
{
'probs': {
'Openness to Change': 0.8884779214859009,
'Conservation': 0.010850731283426285
},
'labels': ['Openness to Change']
}
Note: this is a multi-label model that returns all labels with scores in [0,1]. You still need to choose a threshold to decide what counts as “present”.
For research use, you will often want to:
The predict_values function above already takes a threshold argument, so you can simply do:
predict_values(example, threshold=0.30)
and tune the threshold on your own validation set depending on your precision/recall preferences.
The model was trained on the English, machine-translated portion of the ValueEval’24 / ValuesML dataset:
Important: the original dataset is distributed under a restricted Data Usage Agreement. You must obtain the data separately from the ValueEval/ValuesML organisers (e.g. via Zenodo) and respect their license.
This is the text-only baseline configuration in the paper (no LIWC or context features).
On the English ValueEval’24 sentence-level splits used in the paper, this baseline DeBERTa model achieves:
The paper reports comparisons with:
For full details, please refer to the papers:
Human Values in a Single Sentence: Moral Presence, Hierarchies, and Transformer Ensembles on the Schwartz Continuum Víctor Yeste, Paolo Rosso (2026), arXiv:2601.14172
Do Schwartz Higher-Order Values Help Sentence-Level Human Value Detection? When Hard Gating Hurts Víctor Yeste, Paolo Rosso (2026), arXiv:2602.00913
If you use this model, please:
The model weights and code in this repository are released under the Apache License 2.0.
See the LICENSE file (or the license field in this model card) for details.
Note: This does not grant you any rights over the underlying training data (ValueEval/ValuesML). Please obtain and use that data under its own license and Data Usage Agreement.
If you use this model or the associated code in your research, please cite:
@misc{yeste2026humanvaluessinglesentence,
title={Human Values in a Single Sentence: Moral Presence, Hierarchies, and Transformer Ensembles on the Schwartz Continuum},
author={Víctor Yeste and Paolo Rosso},
year={2026},
eprint={2601.14172},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2601.14172},
}
@misc{yeste2026schwartzhigherordervalueshelp,
title={Do Schwartz Higher-Order Values Help Sentence-Level Human Value Detection? When Hard Gating Hurts},
author={Víctor Yeste and Paolo Rosso},
year={2026},
eprint={2602.00913},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2602.00913},
}
You may also want to cite the ValueEval / ValuesML dataset:
@misc{ValueEval24Zenodo,
author = {{The ValuesML Team}},
title = {Touch{\'e}24{-}ValueEval},
year = {2024},
month = {8},
version = {2024-08-09},
publisher = {Zenodo},
doi = {10.5281/zenodo.13283288},
url = {https://doi.org/10.5281/zenodo.13283288}
}
model.safetensors 554 MBtraining_args.bin 5.2 kBmicrosoft/deberta-base