In Switzerland legal translation is uniquely important due to the country's four official languages and requirements for multilingual legal documentation. However, this process traditionally relies on professionals who must be both legal experts and skilled translators -- creating bottlenecks and impacting effective access to justice. To address this challenge, we introduce SwiLTra-Bench, a comprehensive multilingual benchmark of over 180K aligned Swiss legal translation pairs comprising laws, headnotes, and press releases across all Swiss languages along with English, designed to evaluate LLM-based translation systems. Our systematic evaluation reveals that frontier models achieve superior translation performance across all document types, while specialized translation systems excel specifically in laws but under-perform in headnotes. Through rigorous testing and human expert validation, we demonstrate that while fine-tuning open SLMs significantly improves their translation quality, they still lag behind the best zero-shot prompted frontier models such as Claude-3.5-Sonnet. Additionally, we present SwiLTra-Judge, a specialized LLM evaluation system that aligns best with human expert assessments.

This paper investigates the challenges in building Swiss German speech translation systems, specifically focusing on the impact of dialect diversity and differences between Swiss German and Standard German. Swiss German is a spoken language with no formal writing system, it comprises many diverse dialects and is a low-resource language with only around 5 million speakers. The study is guided by two key research questions: how does the inclusion and exclusion of dialects during the training of speech translation models for Swiss German impact the performance on specific dialects, and how do the differences between Swiss German and Standard German impact the performance of the systems? We show that dialect diversity and linguistic differences pose significant challenges to Swiss German speech translation, which is in line with linguistic hypotheses derived from empirical investigations.

Humans possess an extraordinary ability to create and utilize tools, allowing them to overcome physical limitations and explore new frontiers. With the advent of foundation models, AI systems have the potential to be equally adept in tool use as humans. This paradigm, i.e., tool learning with foundation models, combines the strengths of specialized tools and foundation models to achieve enhanced accuracy, efficiency, and automation in problem-solving. Despite its immense potential, there is still a lack of a comprehensive understanding of key challenges, opportunities, and future endeavors in this field. To this end, we present a systematic investigation of tool learning in this paper. We first introduce the background of tool learning, including its cognitive origins, the paradigm shift of foundation models, and the complementary roles of tools and models. Then we recapitulate existing tool learning research into tool-augmented and tool-oriented learning. We formulate a general tool learning framework: starting from understanding the user instruction, models should learn to decompose a complex task into several subtasks, dynamically adjust their plan through reasoning, and effectively conquer each sub-task by selecting appropriate tools. We also discuss how to train models for improved tool-use capabilities and facilitate the generalization in tool learning. Considering the lack of a systematic tool learning evaluation in prior works, we experiment with 18 representative tools and show the potential of current foundation models in skillfully utilizing tools. Finally, we discuss several open problems that require further investigation for tool learning. Overall, we hope this paper could inspire future research in integrating tools with foundation models.

Neural machine translation (NMT) systems exhibit limited robustness in handling source-side linguistic variations. Their performance tends to degrade when faced with even slight deviations in language usage, such as different domains or variations introduced by second-language speakers. It is intuitive to extend this observation to encompass dialectal variations as well, but the work allowing the community to evaluate MT systems on this dimension is limited. To alleviate this issue, we compile and release \dataset, a contrastive dialectal benchmark encompassing 882 different variations from nine different languages. We also quantitatively demonstrate the challenges large MT models face in effectively translating dialectal variants. We are releasing all code and data.

Accurate estimates of the surface solar radiation (SSR) are a prerequisite for intra-day forecasts of solar resources and photovoltaic power generation. Intra-day SSR forecasts are of interest to power traders and to operators of solar plants and power grids who seek to optimize their revenues and maintain the grid stability by matching power supply and demand. Our study analyzes systematic biases and the uncertainty of SSR estimates derived from Meteosat with the SARAH-2 and HelioMont algorithms at intra-hour and intra-day time scales. The satellite SSR estimates are analyzed based on 136 ground stations across altitudes from 200 m to 3570 m Switzerland in 2018. We find major biases and uncertainties in the instantaneous, hourly and daily-mean SSR. In peak daytime periods, the instantaneous satellite SSR deviates from the ground-measured SSR by a mean absolute deviation (MAD) of 110.4 and 99.6 W/m2 for SARAH-2 and HelioMont, respectively. For the daytime SSR, the instantaneous, hourly and daily-mean MADs amount to 91.7, 81.1, 50.8 and 82.5, 66.7, 42.9 W/m2 for SARAH-2 and HelioMont, respectively. Further, the SARAH-2 instantaneous SSR drastically underestimates the solar resources at altitudes above 1000 m in the winter half year. A possible explanation in line with the seasonality of the bias is that snow cover may be misinterpreted as clouds at higher altitudes.

As you may know, I've been playing around with AI lately. While these are humorous and can sometimes show the model's strengths and weaknesses, I felt the realm of pre-pubescent humor had had its time. I instead wanted to see if I could ask the AI questions and have a conversation-style interaction much like this old program I used to mess around with back in the day called Eliza (example in link). It was supposed to be kind of a therapist and you could ask questions and it would respond. It was super basic but it felt like an early AI to me. Even if it was limited in responses, it was kind of fun to use, sometimes to humorous effect.