Model Card for MET-D-Qwen3-4B-es-only

MET-D-Qwen3-4B-es-only is a Spanish-only moral reasoning model fine-tuned from Qwen3-4B. Given a moral dilemma, a character description, and a candidate action, it judges the action from that character's perspective and explains its judgment with an explicit chain-of-thought before answering. Moral dilemmas rarely have a single correct answer, which makes reasoning traces hard to verify. We address this by introducing a character perspective that yields a ground-truth answer, which is used for rejection-sampling the model's own reasoning traces, conditioned on a per-language, per-situation selection of theoretical grounds. Both the reasoning trace and the final answer are generated in Spanish.

Model Details

• Base model: Qwen/Qwen3-4B
• Task: for a given (situation, character description, action) triple, answer two questions from the character's perspective:
  1. Is the action acceptable? (Yes / No / Ambiguous)
  2. Would (not) doing it be emotionally/mentally discomforting? (Yes / No)
• Training data: self-generated reasoning traces, rejection-sampled against the ground truth per character perspective, in Spanish only
• Language coverage: this checkpoint is trained on Spanish only. The multilingual model and other single-language variants are also available (see Model Variants below).
• License: Apache 2.0 (inherited from the base model)

How to Get Started

from transformers import AutoTokenizer
from vllm import LLM, SamplingParams

model_id = "launch/MET-D-Qwen3-4B-es-only"
tokenizer = AutoTokenizer.from_pretrained(model_id)
llm = LLM(model=model_id, max_model_len=16384)

situation = "..."            # the moral dilemma narrative
character = "..."            # the character description
action = "..."                # the candidate action to judge

prompt = f"""Te propondré un dilema moral y te presentaré una descripción del personaje y preguntas.
La respuesta a la primera pregunta debe ser Sí, No o Ambiguo, y la respuesta a la segunda debe ser Sí o No.
Debes respetar estrictamente el formato de salida, ya que se analizará automáticamente.
Asegúrate de mostrar tanto la "Razón:" como la "Respuesta:" para las dos preguntas.

Dilema moral:
{situation}

Descripción del personaje:
{character}

Pregunta:
1. Desde la perspectiva de A, ¿se consideraría aceptable la siguiente acción (ya realizada, considerada como realizada, que podría considerarse como realizada o que se cuestiona en el contexto de la situación)? "{action}"
2. Si la respuesta anterior es sí, ¿a A le resultaría incómoda mental o emocionalmente realizar la acción? Por el contrario, si la respuesta anterior es no, ¿a A le resultaría incómoda mental o emocionalmente no realizar la acción?

Tu respuesta:
1. Razón: {{Razón}} Respuesta: {{Sí/No/Ambiguo}}
2. Razón: {{Razón}} Respuesta: {{Sí/No}}
"""

chat_prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": prompt}],
    tokenize=False,
    add_generation_prompt=True,
)

sampling_params = SamplingParams(temperature=0.0, max_tokens=2048)
outputs = llm.generate(chat_prompt, sampling_params)
print(outputs[0].outputs[0].text)

Model Variants

This checkpoint is part of the MET collection, which includes the same task across base models and language subsets:

Repo	Base model	Language(s)
launch/MET-D-Qwen3-4B	Qwen3-4B	all 6 (mixed)
launch/MET-D-Qwen3-4B-en-only	Qwen3-4B	English only
launch/MET-D-Qwen3-4B-es-only	Qwen3-4B	Spanish only
launch/MET-D-Qwen3-4B-hi-only	Qwen3-4B	Hindi only
launch/MET-D-Qwen3-4B-ko-only	Qwen3-4B	Korean only
launch/MET-D-Qwen3-4B-ms-only	Qwen3-4B	Malay only
launch/MET-D-Qwen3-4B-zh-only	Qwen3-4B	Chinese only
launch/MET-D-Qwen3-8B	Qwen3-8B	all 6 (mixed)
launch/MET-D-Qwen3-8B-en-only	Qwen3-8B	English only
launch/MET-D-Gemma3-4B	Gemma-3-4B-it	all 6 (mixed)
launch/MET-D-Gemma3-4B-en-only	Gemma-3-4B-it	English only

Citation

If you use this, please cite:

@article{lee2026met,
  title={MET: Theory-Grounded and Culture-Aware Multilingual Moral Reasoning},
  author={Lee, Ayoung and Kwon, Ryan and Zhang, Yunxiang and Liu, Yuxuan and Railton, Peter and Wang, Lu},
  journal={arXiv preprint arXiv:2607.11736},
  year={2026}
}