Latxa-Qwen3-8B-Clinical-v1-ca-eu is a LoRA adapter for Catalan-to-Basque clinical translation. It was trained directly from the base model on a back-translated clinical corpus built from Basque medical texts.

Model details

Developed by: Paula Guerrero and Iker Gutierrez
Affiliation: University of the Basque Country (EHU)
Model type: LoRA adapter for HiTZ/Latxa-Qwen3-VL-8B-Instruct
Languages: Catalan (ca), Basque (eu)
Domain: Clinical translation
Direction: ca->eu only
Base model: HiTZ/Latxa-Qwen3-VL-8B-Instruct
Repository: pguerrero-igutierrez/Latxa-Qwen3-8B-Clinical-v1-ca-eu
Collection: pguerrero-igutierrez/mt-domain-adaptation-ca-eu

Sources

Hugging Face repository: https://huggingface.co/pguerrero-igutierrez/Latxa-Qwen3-8B-Clinical-v1-ca-eu
Hugging Face collection: https://huggingface.co/collections/pguerrero-igutierrez/mt-domain-adaptation-ca-eu
Project repository: https://github.com/pguerrero-igutierrez/MT-domain-adaptation
Paper source: https://github.com/pguerrero-igutierrez/MT-domain-adaptation/tree/main/paper

Intended use

This model is intended for research on Catalan-to-Basque clinical translation in low-resource settings.

Supported prompting direction:

ca->eu: Tradueix aquest text clínic del català al basc:\n\n{source}

Out-of-scope use

Medical decision-making
Clinical deployment without expert review
Any reverse direction (eu->ca)
Translation outside the clinical domain

Training data

The adapter was trained on backtranslated-corpus/eu-clinical_backtranslated.json, where synthetic Catalan (ca) is used as source and original Basque (eu) as target.

The corpus was built from Basque clinical documents in the E3C corpus using back-translation.

Training procedure

LoRA rank: 16
LoRA alpha: 32
LoRA dropout: 0.05
Target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Quantization: 4-bit NF4
Max sequence length: 768
Epochs: 3
Batch size: 4
Gradient accumulation: 8
Learning rate: 5e-5
Scheduler: cosine
Warmup ratio: 0.05
Seed: 42
Checkpoint selection: best validation BLEU

Evaluation

Results on the clinical held-out test set:

Direction	chrF++	BLEU	TER	COMET
`ca->eu`	40.20	19.43	101.09	76.25

This was the strongest model in the project on the clinical domain by chrF++, BLEU, and COMET.

Limitations

Only supports ca->eu
Trained on synthetic-source data
Automatic metrics do not replace expert clinical validation
Must not be used for diagnosis or patient care without human oversight

Usage

python

import torch
from peft import PeftModel
from transformers import AutoTokenizer, Qwen3VLForConditionalGeneration

base_id = "HiTZ/Latxa-Qwen3-VL-8B-Instruct"
adapter_id = "pguerrero-igutierrez/Latxa-Qwen3-8B-Clinical-v1-ca-eu"

tokenizer = AutoTokenizer.from_pretrained(base_id, trust_remote_code=True)
base_model = Qwen3VLForConditionalGeneration.from_pretrained(
    base_id,
    device_map="auto",
    torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
    trust_remote_code=True,
)
model = PeftModel.from_pretrained(base_model, adapter_id)

prompt = "Tradueix aquest text clínic del català al basc:\n\nEl pacient presenta febre alta."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Citation

bibtex

@misc{guerrero-gutierrez-2026-caeu-mt,
  title        = {Domain Adaptation for Catalan-Basque Machine Translation via Synthetic Data and Continued Fine-Tuning},
  author       = {Guerrero, Paula and Gutierrez, Iker},
  year         = {2026},
  note         = {Unpublished manuscript}
}

Contact

Paula Guerrero: pguerrero005@ikasle.ehu.eus
Iker Gutierrez: igutierrez134@ikasle.ehu.eus