The paper introduces MWE‑ECL, a bilingual diagnostic framework that tests whether distant discourse anchors can override local lexical priors in multi‑word expression interpretation. It evaluates models on a 0‑128K context grid, finding that while retrieval of anchors is near perfect, the ability to change locally preferred readings varies, especially when the model’s default conflicts with the anchor. The study shows that explicit recoverability does not always translate into behavioral influence, with gaps differing across models and languages.
Long-context evaluations often test whether a model can recover distant evidence, but recoverability does not guarantee behavioral influence. We test the prediction that a distant discourse anchor can...
arXiv:2608. 03297v1 Announce Type: new Abstract: A standard claim in the literature on retrieval-augmented and memory-augmented language models is that shorter context is better when the relevant information is preserved.
The paper introduces a three-level evaluation framework—behavioral deployment, LM-head readout, and probe recoverability—to distinguish whether a language model fails a syntactic test by not encoding structure or by failing to use it. Using a trilingual control-dependency benchmark, the authors find that probe recoverability consistently exceeds LM-head readout, which in turn exceeds behavioral deployment across seven models and three languages, with the largest gap observed in Qwen3-0.6B Instruct. Layer-localized activation patching shows that instruction tuning shifts the decoded layer later, suggesting decoding favors surface shortcuts and that behavioral evaluation understates what models encode while probing alone overstates what they deploy.
arXiv:2608. 04021v1 Announce Type: cross Abstract: Cloze-style probes that vary how often a target token appears implicitly assume that more copies of a target affect prediction the same way regardless of where the readout slot sits.
The paper investigates how multilingual large language models can be guided to reason more reliably in low- to mid-resource languages by selecting appropriate language modes during inference. Experiments with LLaMA and Qwen models show that using English context can correct errors from non‑English comprehension, but adding redundant bilingual context can cause interference. To balance this trade‑off, the authors propose Reliability‑Aware Adaptive Inference (RAAI), a training‑free test‑time framework that routes prompts based on Expected Calibration Error and gates reasoning with a mid‑layer Risk Index, achieving up to 37.7% accuracy gains and reduced calibration error on low‑resource languages.
arXiv:2602.22453v4 Announce Type: replace Abstract: Retrieval heads, a subset of attention heads in Transformers, were studied in English, showing its crucial role in retrieving information from the...
arXiv:2606. 07555v1 Announce Type: cross Abstract: Glossaries, technical specifications, and system prompts routinely ask language models to use familiar words in unfamiliar ways.
arXiv:2607. 04223v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) reduces but does not eliminate hallucination, and existing detectors return a single answer-level score that does not indicate which sentence is unsupported, or why.
arXiv:2606.16661v2 Announce Type: replace-cross Abstract: Fixed-length chunking in Retrieval-Augmented Generation (RAG) often leads to boundary fragmentation, where critical evidence is split across...
arXiv:2605. 14473v4 Announce Type: replace-cross Abstract: Retrieval-Augmented Generation (RAG) is usually evaluated by whether the final answer is correct.
When a tool-using agent is given the same task in a different language, does it still take the same steps? Multilingual evaluation rarely asks: it compares final answers and discards the actions.
MGAL is a new multilingual benchmark for evaluating long‑context large language models, built from United Nations reports in six official UN languages and covering 8K to 128K tokens. It tests four linguistic granularities—word, sentence, paragraph, and document—while also stratifying examples by their position within the document (begin, middle, end). Experiments show that models excel at word‑level tasks but struggle with coarser granularity, and that closed‑source models outperform others in lower‑resource languages, revealing challenges such as local semantic crowding and a fluency‑consistency gap.