One paper argues for a two-stage generate-then-rank pipeline for symbolic goal interpretation in embodied agents, while another says the real gap is elsewhere: τ-bench-LILT extends τ^3-bench beyond English into German, Korean, Turkish, and Farsi for airline, retail, telecom, and banking tool use. On the MT side, a German–English study filters 45k OPUS Books sentence pairs down to 25k and fine-tunes NLLB-200 with consumer-GPU-friendly settings, reporting a statistically significant gain from the 600M model.

Generate-then-Rank: Learning Symbolic Goal Interpretation for Embodied Agents

This is an embodied-agent planning paper, but the generate-then-rank setup matters for MT because candidate generation plus output reranking is a useful pattern when the target format is rigid and errors cascade.

Goal interpretation is to translate a natural language instruction into a formal symbolic goal that a planner can execute. It is critical for embodied agents because any error here propagates to every downstream planning and action step. Doing it well requires both generating suitable candidates and reliably selecting among them. We present a two-stage generate-then-rank approach to Goal Interpretation in the Embodied Agent Interface (EAI). First, supervised fine-tuning adapts a Large Language Model (LLM) to EAI's output convention. Second, Candidate-Output Ranking from Errors (CORE) learns preferences over complete outputs from the model's own generated alternatives. We conducted experiments on the household scenes of the Embodied Agent Interface Benchmark. Multiple open source LLMs including DeepSeek-V3 chat, DeepSeek-Reasoner, and Qwen2.5-7B and Qwen2.5-72B were evaluated. The proposed method outperformed the baseline models including LLMs with simple prompting, chain-of-thought, and knowledge graph based approaches.

τ-bench-LILT: Evaluating Conversational Agents on Realistic Multilingual Tool-Use Tasks

Multilingual tool-use agents lose heavily outside English, with Korean and Farsi retail tasks dropping by up to 61 pass^4 points, so locale-specific benchmarks need native data, not translated English tasks.

Frontier evaluations of tool-using conversational agents are conducted almost exclusively in English, leaving a critical gap for organizations deploying agents in other languages and regions. We present τ-bench-LILT, a multilingual extension of τ^3-bench covering German, Korean, Turkish, and Farsi in the airline, retail, telecom, and banking domains. We natively localize every layer of the benchmark: customer databases are populated with culturally appropriate records, user-simulator prompts are refined iteratively through contrastive annotation by native-language experts, and tasks are updated according to the localized database and to reflect realistic tasks in the target region. Evaluation across eight agent models shows the largest drops compared to English in the Korean and Farsi retail domains, by up to 61 points in pass^4. Code and data released at https://github.com/lilt/tau-bench-lilt.

Dataset Collection and Fine-Tuning of NLLB for German--English Translation

A 45k-to-25k filtered OPUS Books pipeline plus consumer-GPU fine-tuning of NLLB-600M gives a reproducible German-English baseline that improves by 3.4 BLEU and integrates cleanly into document translation.

This paper describes the curation of a high-quality parallel corpus from OPUS Books and the fine-tuning of Meta's NLLB-200 models for German--English translation. We detail a dataset extraction pipeline that filters 45k raw OPUS sentence pairs to 25k high-quality examples through encoding validation, length filtering, language identification, deduplication, and quality scoring. We present a reproducible fine-tuning methodology optimized for consumer GPUs using mixed precision, gradient accumulation, and gradient checkpointing. Our fine-tuned NLLB-600M model achieves statistically significant improvements (+3.4 BLEU, +2.6 chrF++, -3.2 TER) over the base model on literary and technical German-English translation. We demonstrate practical deployment by integrating the fine-tuned model into LinguaHaru, an existing open-source document translation system, achieving format-preserving translation across PDF and Office documents. All dataset preparation scripts, training code, and evaluation tools are provided for reproducibility. The fine-tuned model is available at HuggingFace.

Papers announced by arXiv on Thursday, 5 November 2026, covering submissions from 4 November to 5 November 2026.