Shipping translation systems without knowing where they break is how you end up with fluent output that still misses the job: discourse coherence slips in document-level simplification for Estonian, and a one-word phrasing change can swing vision-language-action success by tens of points. The papers here probe both sides of that problem, from five multilingual LLMs on document-level simplification in a low-resource, morphologically rich language to rephrase sensitivity in VLAs and a gold-set-driven pipeline for turning a handful of expert annotations into a usable task definition.

Document-Level Text Simplification in Estonian Using Large Language Models

For Estonian document-level simplification, use the pipeline or guideline-augmented prompting only if you can check discourse coherence, because Gemini-2.0 and LLaMA-3.3 are the only models here that stay near-native while preserving meaning.

Document-level text simplification involves transformations that go beyond sentence-internal edits, addressing discourse coherence, anaphora resolution, and cross-paragraph consistency. Despite advances in sentence-level simplification for high-resource languages, document-level simplification in morphologically rich, low-resource languages such as Estonian remains largely unexplored. This study presents a comprehensive evaluation of five state-of-the-art multilingual large language models (LLMs) for document-level simplification in Estonian. Three prompting strategies are examined: single-pass generation, pipeline-based modular agents, and guideline-augmented pipelines. The evaluation framework integrates automatic metrics assessing readability, semantic preservation, and discourse coherence, alongside a structured manual annotation protocol. The findings indicate that Gemini-2.0 and LLaMA-3.3 produce outputs with near-native fluency and strong meaning preservation, whereas other models display notable grammatical and semantic limitations. This work contributes novel document-level coherence metrics, evidence-based prompting strategies, and publicly available resources for reproducibility.

Rephrase Before You Act: Characterizing and Mitigating Language Sensitivity in Vision-Language-Action Models

Instruction wording matters enough that a one-word rephrase can swing VLA success by tens of points, so normalise prompts or apply rephrase rules before deployment instead of trusting finetuning alone.

Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-language models they are built on. A one-word edit can move success by tens of points: π_0.5 turns on a LIBERO stove 100% of the time for "switch on the stove" and 2% for "switch on the hot plate", and a π_0 checkpoint finetuned with rephrase augmentation still shows swings of up to 61 points. We characterize this sensitivity with statistically tested single-edit swings and an oracle phrase search, which shows that phrasing alone nearly closes the 21-point gap between in-distribution and out-of-distribution tasks. We then reduce it without modifying the policy. Because the sensitivity is systematic, it can be expressed as explicit rules: we score many phrasings of a few training tasks, have a large language model distill the evidence into ten to twenty rephrasing rules, and at deployment rewrite each incoming instruction once under these rules. The rules improve the frozen π_0 by 16 to 27% relative on twelve held-out tasks across adversarial, VLM-generated, and human-generated phrasings, with gains concentrated on out-of-distribution tasks. The pipeline replicates on π_0.5 and LIBERO, lifting in-finetune success from 93.6% to 97.8%. The method requires no retraining and no per-step verification, and applies zero-shot to unseen tasks and instructions. Project website: https://sttawm.github.io/rephrase-before-you-act

Goldsmith: Gold-Loss-Guided Definition Optimization with an Agentic Annotation Harness

Goldsmith turns a small expert gold set into a reusable annotation definition with loss-guided revisions, which is useful when you need to scale annotation before the guideline is stable.

Many annotation projects begin before experts have a stable guideline or enough labels to train a task-specific model. We present Goldsmith, an agentic pipeline that turns a small gold set---expert-annotated calibration examples representing the intended task boundaries---into a reusable structured annotation definition. Goldsmith treats this definition as a trainable textual object. Candidate definitions are run on the same gold examples and scored with an executable structured loss, while the output schema, formatting, retrieval, repair, judging, and human review remain in an external harness. A large language model (LLM) editor converts the highest-loss failures into textual-gradient revisions, which are accepted only when the measured loss decreases. In prompt-optimization comparisons, Goldsmith improves over direct rewriting, OPRO, APE, and PromptBreeder under matched evaluation protocols. The resulting definition also improves downstream annotation when combined with retrieval, score-based routing, and human review across typed span, pair-level relation, and fixed-trigger event-argument tasks. These results show that scarce expert supervision can support both task-definition learning and scalable annotation.

Papers announced by arXiv on Wednesday, 7 October 2026, covering submissions from 6 October to 7 October 2026.