@InProceedings{jourdain-pradelles-semmar:2026:wmt,
  author    = {Jourdain, Louis  and  Pradelles, Aurélie  and  Semmar, Nasredine},
  title     = {ChapsVision (CHV) at WMT 2026 CreoleMT: Shifting Creole MT from Train-Time to Test-Time Compute for Haitian Creole},
  booktitle      = {Proceedings of the Eleventh Conference on Machine Translation},
  month          = {October},
  year           = {2026},
  address        = {Budapest, Hungary},
  publisher      = {Association for Computational Linguistics},
  pages     = {2420--2446},
  abstract  = {We describe a training-free system for the WMT26 Creole MT shared task, targeting Haitian Creole in four directions (eng↔hat, fra↔hat). Rather than fine-tune, we wrap a sin- gle fixed 8B instruction model (qwen3-8b) in a language-agnostic harness that retrieves lexi- cal, parallel-data and grammatical evidence and injects it directly into the translation prompt. On FLORES+ devtest the harness lifts the base model by up to +14 chrF++ and clears the shared task's primary baseline in all four di- rections, though a purpose-fine-tuned system (kreyòl-MT) and frontier LLM still lead the mean. On the task's blind test set that ordering reverses: we finish below both baselines, while the harness gain itself holds, staying large into Creole and small out of it. Two findings frame the paper. First, injected knowledge behaves as a reasoning equalizer: the same evidence that lifts a weak open model leaves a strong frontier model flat, so grounding substitutes for a capa- bility the model lacks rather than amplifying one it has. Second, how evidence is delivered dominates: forcing it into the prompt beats let- ting the model self-serve the same tools in an agentic (ReAct) loop, which costs about 6× the tokens.},
  url       = {https://aclanthology.org/2026.wmt-1.186}
}