Human mobility modeling from GPS-trajectories and synthetic trajectory generation are crucial for various applications, such as urban planning, disaster management and epidemiology. Both of these tasks often require filling gaps in a partially specified sequence of visits - a new problem that we call "controlled" synthetic trajectory generation. Existing methods for next-location prediction or synthetic trajectory generation cannot solve this problem as they lack the mechanisms needed to constrain the generated sequences of visits. Moreover, existing approaches (1) frequently treat space and time as independent factors, an assumption that fails to hold true in real-world scenarios, and (2) suffer from challenges in accuracy of temporal prediction as they fail to deal with mixed distributions and the inter-relationships of different modes with latent variables (e.g., day-of-the-week). These limitations become even more pronounced when the task involves filling gaps within sequences instead of solely predicting the next visit. We introduce TrajGPT, a transformer-based, multi-task, joint spatiotemporal generative model to address these issues. Taking inspiration from large language models, TrajGPT poses the problem of controlled trajectory generation as that of text infilling in natural language. TrajGPT integrates the spatial and temporal models in a transformer architecture through a Bayesian probability model that ensures that the gaps in a visit sequence are filled in a spatiotemporally consistent manner. Our experiments on public and private datasets demonstrate that TrajGPT not only excels in controlled synthetic visit generation but also outperforms competing models in next-location prediction tasks - Relatively, TrajGPT achieves a 26-fold improvement in temporal accuracy while retaining more than 98% of spatial accuracy on average.

This paper presents a novel benchmark where the large language model (LLM) must write code that computes integer sequences from the Online Encyclopedia of Integer Sequences (OEIS), a widely-used resource for mathematical sequences. The benchmark is designed to evaluate both the correctness of the generated code and its computational efficiency. Our benchmark reveals that the o1 series of models outperform other frontier models from OpenAI, Anthropic, Meta, and Google in accuracy and cheating rates across both easy and hard integer sequences. In order to ensure models do not exploit memorized sequence values, we introduce an automated cheating detection mechanism that flags the use of lookup tables and validated this automation against human cheating evaluations. This benchmark provides a meaningful challenge for current LLMs, offering insights into their mathematical reasoning and code writing capabilities, which can guide future research directions and model development in mathematical reasoning and code synthesis.

An open problem in artificial intelligence is how to train language models that produce responses that are factually correct. Current frontier models sometimes produce false outputs or answers that are not substantiated by evidence, a problem known as "hallucinations." Such hallucinations are one of the major barriers for broader adoption of general forms artificial intelligence like large language models. Factuality is a complicated topic because it is hard to measure--evaluating the factuality of any given arbitrary claim can be challenging, and language models often generate long completions that contain dozens of factual claims. In this work we will sidestep the open-endedness of language models by considering only short, fact-seeking questions with a single answer. This reduction of scope is important because it makes measuring factuality much more tractable, albeit at the cost of leaving open research questions such as whether improved behavior on short-form factuality generalizes to long-form factuality. We present a benchmark called SimpleQA, which contains 4,326 short, fact-seeking questions. SimpleQA was designed with a few important properties in mind: High correctness. Reference answers to questions are determined by two independent AI trainers, and questions were written in such a way that the predicted answers are easily gradable.

Large Language Models have become a valuable source of information for our daily inquiries. However, after training, its data source quickly becomes out-of-date, making RAG a useful tool for providing even more recent or pertinent data. In this work, we investigate how RAG pipelines, with the course materials serving as a data source, might help students in K-12 education. The initial research utilizes Reddit as a data source for up-to-date cybersecurity information. Chunk size is evaluated to determine the optimal amount of context needed to generate accurate answers. After running the experiment for different chunk sizes, answer correctness was evaluated using RAGAs with average answer correctness not exceeding 50 percent for any chunk size. This suggests that Reddit is not a good source to mine for data for questions about cybersecurity threats. The methodology was successful in evaluating the data source, which has implications for its use to evaluate educational resources for effectiveness.

Large Language Models (LLM) have brought numerous of new applications to Machine Learning (ML). In the context of tabular data (TD), recent studies show that TabLLM is a very powerful mechanism for few-shot-learning (FSL) applications, even if gradient boosting decisions trees (GBDT) have historically dominated the TD field. In this work we demonstrate that although LLMs are a viable alternative, the evidence suggests that baselines used to gauge performance can be improved. We replicated public benchmarks and our methodology improves LightGBM by 290%, this is mainly driven by forcing node splitting with few samples, a critical step in FSL with GBDT. Our results show an advantage to TabLLM for 8 or fewer shots, but as the number of samples increases GBDT provides competitive performance at a fraction of runtime. For other real-life applications with vast number of samples, we found FSL still useful to improve model diversity, and when combined with ExtraTrees it provides strong resilience to overfitting, our proposal was validated in a ML competition setting ranking first place.

Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capable large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.

South Africa and the Democratic Republic of Congo (DRC) present a complex linguistic landscape with languages such as Zulu, Sepedi, Afrikaans, French, English, and Tshiluba (Ciluba), which creates unique challenges for AI-driven translation and sentiment analysis systems due to a lack of accurately labeled data. This study seeks to address these challenges by developing a multilingual lexicon designed for French and Tshiluba, now expanded to include translations in English, Afrikaans, Sepedi, and Zulu. The lexicon enhances cultural relevance in sentiment classification by integrating language-specific sentiment scores. A comprehensive testing corpus is created to support translation and sentiment analysis tasks, with machine learning models such as Random Forest, Support Vector Machine (SVM), Decision Trees, and Gaussian Naive Bayes (GNB) trained to predict sentiment across low resource languages (LRLs). Among them, the Random Forest model performed particularly well, capturing sentiment polarity and handling language-specific nuances effectively. Furthermore, Bidirectional Encoder Representations from Transformers (BERT), a Large Language Model (LLM), is applied to predict context-based sentiment with high accuracy, achieving 99% accuracy and 98% precision, outperforming other models. The BERT predictions were clarified using Explainable AI (XAI), improving transparency and fostering confidence in sentiment classification. Overall, findings demonstrate that the proposed lexicon and machine learning models significantly enhance translation and sentiment analysis for LRLs in South Africa and the DRC, laying a foundation for future AI models that support underrepresented languages, with applications across education, governance, and business in multilingual contexts.

Large language models (LLMs) offer promise in generating educational content, providing instructor feedback, and reducing teacher workload on assessments. While prior studies have focused on studying LLM-powered learning analytics, limited research has examined how effective LLMs are in a bilingual context. In this paper, we study the effectiveness of multilingual large language models (MLLMs) across monolingual (English-only, Spanish-only) and bilingual (Spanglish) student writing. We present a learning analytics use case that details LLM performance in assessing acceptable and unacceptable explanations of Science and Social Science concepts. Our findings reveal a significant bias in the grading performance of pre-trained models for bilingual writing compared to English-only and Spanish-only writing. Following this, we fine-tune open-source MLLMs including Llama 3.1 and Mistral NeMo using synthetic datasets generated in English, Spanish, and Spanglish. Our experiments indicate that the models perform significantly better for all three languages after fine-tuning with bilingual data. This study highlights the potential of enhancing MLLM effectiveness to support authentic language practices amongst bilingual learners. It also aims to illustrate the value of incorporating non-English languages into the design and implementation of language models in education.

In moving from excluding the majority of the world's languages to blindly adopting what we make for English, we first risk importing the same harms we have at best mitigated and at least measured for English. For instance, Yong et al. [15] showed how prompting GPT-4 in low-resource languages circumvents guardrails that are effective in English. However, in evaluating and mitigating harms arising from adopting new technologies into such contexts, we often disregard (1) the actual community needs of Language Technologies, and (2) biases and fairness issues within the context of the communities. Here, we consider fairness, bias, and inclusion in Language Technologies through the lens of the Capabilities Approach [12]. The Capabilities Approach centers what people are capable of achieving, given their intersectional social, political, and economic contexts instead of what resources are (theoretically) available to them. In the following sections, we detail the Capabilities Approach, its relationship to multilingual and multicultural evaluation, and how the framework affords meaningful collaboration with community members in defining and measuring harms of Language Technologies. 2 THE CAPABILITIES APPROACH The Capabilities Approach is a framework in developmental economic studies proposed by Amartya Sen in a series of articles published as far back as 1974 [1]. It has been applied to varied fields including environmental justice [e.g.

Vision-language models (VLMs) have improved significantly in multi-modal tasks, but their more complex architecture makes their safety alignment more challenging than the alignment of large language models (LLMs). In this paper, we reveal an unfair distribution of safety across the layers of VLM's vision encoder, with earlier and middle layers being disproportionately vulnerable to malicious inputs compared to the more robust final layers. This 'cross-layer' vulnerability stems from the model's inability to generalize its safety training from the default architectural settings used during training to unseen or out-of-distribution scenarios, leaving certain layers exposed. We conduct a comprehensive analysis by projecting activations from various intermediate layers and demonstrate that these layers are more likely to generate harmful outputs when exposed to malicious inputs. Our experiments with LLaVA-1.5 and Llama 3.2 show discrepancies in attack success rates and toxicity scores across layers, indicating that current safety alignment strategies focused on a single default layer are insufficient.