We describe the CUNI submissions to the WMT26 Automated MT Evaluation Shared Task. We take part in error-span prediction (Task~1) and segment-level quality score prediction (Task~2). Our systems use a single open-weight Gemma 4 31B instruction-tuned model as a backbone. For Task~1, we decompose error prediction into semantic error detection, self-review, and a separate span-alignment step that maps free-form error descriptions to source and target character spans. We submit two variants of this approach, one additionally predicting MQM-style error categories. For Task~2, we explore two complementary approaches: regression over features extracted from the Task~1 predictions and LLM-based direct assessment on a 0–100 scale.

We describe the Charles University CUNI-AR submission to the WMT26 General Translation shared task for English into Egyptian Arabic. Our pipeline is built around GEMBA (LLM-as-judge quality) estimator that scores translations independently for accuracy and dialect authenticity. We use it throughout our pipeline: to filter training data, to construct preference pairs for DPO, to select best checkpoints, and to pick the best hypothesis per segment at inference time. Our primary translation model is Gemma-4-12B-it finetuned via QLoRA, we also finetune Jais-2-8B-Chat and Aya-expanse-8B. Gemma-4-12B-it provides a high-quality Egyptian Arabic translation, so SFT yields only marginal quality gains in automatic metrics; DPO is able to improve the metrics it optimizes for, but the real gain in translation quality is to be assessed by human evaluation. The final submission selects the best-scoring translation per segment from 31 candidate checkpoints, with the unmodified base model contributing 31\% of the output.

We describe our constrained CUNI-MH-v3 (based on Qwen3-14B) and unconstrained CUNI-EdUKate (based on Granite 4.1 8B) submissions for the WMT26 General MT shared task. Both submissions use a three-stage training procedure: QLoRA supervised fine-tuning on synthetic sentence pairs, continued supervised fine-tuning on a smaller mixture of paragraph-level and instruction-oriented examples, and iterative preference optimization with automatically generated and scored preference pairs. We document the data construction, training setup, checkpoint selection, and evaluation. We also share the training data and model checkpoints for our constrained system.

This paper describes linguistic and technological challenges encountered within an applied project aimed at expanding a large e-learning portal from its original Czech to three other languages: Ukrainian, English and German. Although there seems to be a general belief that machine translation is a solved task in 2026, we show that translating educational content, which in our case is highly terminological, multimodal, interactive and encoded in XML, brings along many challenges of different types, some easily solvable and some not. We also compare our results from the early phase of the project (Transformer-based machine translation) with those after the switch to the LLM-based translation methods. We show that both MT methods are prone to different types of errors, some of which are quite new (such as the undesired correction of counterfactual statements) and require new ways of handling them. The resulting four-language edition of the educational web portal will be freely available to educators, students and researchers by the end of 2026.

This paper describes the joint effort of Phrase a.s. and Charles University’sInstitute of Formal and Applied Linguistics (CUNI/UFAL) on the WMT25Automated Translation Quality Evaluation Systems Shared Task. Both teamsparticipated both in a collaborative and competitive manner, i.e. they eachsubmitted a system of their own as well as a contrastive joint system ensemble.In Task~1, we show that such an ensembling—if chosen in a clever way—canlead to a performance boost. We present the analysis of various kinds ofsystems comprising both “traditional” NN-based approach, as well as differentflavours of LLMs—off-the-shelf commercial models, their fine-tuned versions,but also in-house, custom-trained alternative models. In Tasks~2 and~3 we showPhrase’s approach to tackling the tasks via various GPT models: Error SpanAnnotation via the complete MQM solution using non-reasoning models (includingfine-tuned versions) in Task~2, and using reasoning models in Task~3.

This paper describes the CUNI submissions to the WMT25 General Translation task, namely for the English to Czech, English to Serbian, Czech to German and Czech to Ukrainian language pairs. We worked in multiple teams, each with a different approach, spanning from traditional, smaller Transformer NMT models trained on both sentence and document level, to fine-tuning LLMs using LoRA and CPO. We show that these methods are effective in improving automatic MT evaluation scores compared to the base pretrained models.

This paper presents the contributions of Charles University teams to the WMT24 General Translation task (English to Czech, German and Russian, and Czech to Ukrainian), and the WMT24 Translation into Low-Resource Languages of Spain task.Our most elaborate submission, CUNI-MH for en2cs, is the result of fine-tuning Mistral 7B v0.1 for translation using a three-stage process: Supervised fine-tuning using QLoRA, Contrastive Preference Optimization, and merging of model checkpoints. We also describe the CUNI-GA, CUNI-Transformer and CUNI-DocTransformer submissions, which are based on our systems from the previous year.Our en2ru system CUNI-DS uses a similar first stage as CUNI-MH (QLoRA for en2cs) and follows with transferring to en2ru.For en2de (CUNI-NL), we experimented with a LLM-based speech translation system, to translate without the speech input.For the Translation into Low-Resource Languages of Spain task, we performed QLoRA fine-tuning of a large LLM on a small amount of synthetic (backtranslated) data.