The intersection of chemistry and Artificial Intelligence (AI) is an active area of research focused on accelerating scientific discovery. While using large language models (LLMs) with scientific modalities has shown potential, there are significant challenges to address, such as improving training efficiency and dealing with the out-of-distribution problem. Focussing on the task of automated language-molecule translation, we are the first to use state-of-the art (SOTA) human-centric optimisation algorithms in the cross-modal setting, successfully aligning cross-language-molecule modals. We empirically show that we can augment the capabilities of scientific LLMs without the need for extensive data or large models. We conduct experiments using only 10% of the available data to mitigate memorisation effects associated with training large models on extensive datasets. We achieve significant performance gains, surpassing the best benchmark model trained on extensive in-distribution data by a large margin and reach new SOTA levels. Additionally we are the first to propose employing non-linear fusion for mixing cross-modal LLMs which further boosts performance gains without increasing training costs or data needs. Finally, we introduce a fine-grained, domain-agnostic evaluation method to assess hallucination in LLMs and promote responsible use.

Entity set expansion, taxonomy expansion, and seed-guided taxonomy construction are three representative tasks that aim to automatically populate an existing taxonomy with new concepts. Previous studies view them as three separate tasks, and the proposed methods usually only work for one specific task, which lack generalizability and a holistic perspective across different tasks. In this paper, we aim to discover a unified solution to all three tasks. To be specific, we identify two common skills needed for entity set expansion, taxonomy expansion, and seed-guided taxonomy construction: finding "siblings" and finding "parents". We introduce a taxonomy-guided instruction tuning framework to teach a large language model to generate siblings and parents for query entities, where the joint pre-training process facilitates the mutual enhancement of these two skills. Extensive experiments on multiple benchmark datasets demonstrate the efficacy of our proposed TaxoInstruct framework, which outperforms task-specific baselines across all three tasks.

BERT-based models for Electronic Health Records (EHR) have surged in popularity following the release of BEHRT and Med-BERT. Subsequent models have largely built on these foundations despite the fundamental design choices of these pioneering models remaining underexplored. To address this issue, we introduce CORE-BEHRT, a Carefully Optimized and Rigorously Evaluated BEHRT. Through incremental optimization, we isolate the sources of improvement for key design choices, giving us insights into the effect of data representation and individual technical components on performance. Evaluating this across a set of generic tasks (death, pain treatment, and general infection), we showed that improving data representation can increase the average downstream performance from 0.785 to 0.797 AUROC, primarily when including medication and timestamps. Improving the architecture and training protocol on top of this increased average downstream performance to 0.801 AUROC. We then demonstrated the consistency of our optimization through a rigorous evaluation across 25 diverse clinical prediction tasks. We observed significant performance increases in 17 out of 25 tasks and improvements in 24 tasks, highlighting the generalizability of our findings. Our findings provide a strong foundation for future work and aim to increase the trustworthiness of BERT-based EHR models.

Defect reduction planning plays a vital role in enhancing software quality and minimizing software maintenance costs. By training a black box machine learning model and "explaining" its predictions, explainable AI for software engineering aims to identify the code characteristics that impact maintenance risks. However, post-hoc explanations do not always faithfully reflect what the original model computes. In this paper, we introduce CounterACT, a Counterfactual ACTion rule mining approach that can generate defect reduction plans without black-box models. By leveraging action rules, CounterACT provides a course of action that can be considered as a counterfactual explanation for the class (e.g., buggy or not buggy) assigned to a piece of code. We compare the effectiveness of CounterACT with the original action rule mining algorithm and six established defect reduction approaches on 9 software projects. Our evaluation is based on (a) overlap scores between proposed code changes and actual developer modifications; (b) improvement scores in future releases; and (c) the precision, recall, and F1-score of the plans. Our results show that, compared to competing approaches, CounterACT's explainable plans achieve higher overlap scores at the release level (median 95%) and commit level (median 85.97%), and they offer better trade-off between precision and recall (median F1-score 88.12%). Finally, we venture beyond planning and explore leveraging Large Language models (LLM) for generating code edits from our generated plans. Our results show that suggested LLM code edits supported by our plans are actionable and are more likely to pass relevant test cases than vanilla LLM code recommendations.

Training large Deep Neural Network (DNN) models requires thousands of GPUs for days or weeks at a time. At these scales, failures are frequent and can have a big impact on training throughput. Restoring performance using spare GPU servers becomes increasingly expensive as models grow. SlipStream is a system for efficient DNN training in the presence of failures, without using spare servers. It exploits the functional redundancy inherent in distributed training systems -- servers hold the same model parameters across data-parallel groups -- as well as the bubbles in the pipeline schedule within each data-parallel group. SlipStream dynamically re-routes the work of a failed server to its data-parallel peers, ensuring continuous training despite multiple failures. However, re-routing work leads to imbalances across pipeline stages that degrades training throughput. SlipStream introduces two optimizations that allow re-routed work to execute within bubbles of the original pipeline schedule. First, it decouples the backward pass computation into two phases. Second, it staggers the execution of the optimizer step across pipeline stages. Combined, these optimizations enable schedules that minimize or even eliminate training throughput degradation during failures. We describe a prototype for SlipStream and show that it achieves high training throughput under multiple failures, outperforming recent proposals for fault-tolerant training such as Oobleck and Bamboo by up to 1.46x and 1.64x, respectively.

The process of instruction tuning aligns pre-trained large language models (LLMs) with open-domain instructions and human-preferred responses. While several studies have explored autonomous approaches to distilling and annotating instructions from more powerful proprietary LLMs, such as ChatGPT, they often neglect the impact of task distributions and the varying difficulty of instructions of the training sets. This oversight can lead to imbalanced knowledge capabilities and poor generalization powers of small student LLMs. To address this challenge, we introduce Task-Aware Curriculum Planning for Instruction Refinement (TAPIR), a multi-round distillation framework with balanced task distributions and dynamic difficulty adjustment. This approach utilizes an oracle LLM to select instructions that are difficult for a student LLM to follow and distill instructions with balanced task distributions. By incorporating curriculum planning, our approach systematically escalates the difficulty levels, progressively enhancing the student LLM's capabilities. We rigorously evaluate TAPIR using two widely recognized benchmarks, including AlpacaEval 2.0 and MT-Bench. The empirical results demonstrate that the student LLMs, trained with our method and less training data, outperform larger instruction-tuned models and strong distillation baselines. The improvement is particularly notable in complex tasks, such as logical reasoning and code generation.

Blockchain technology and Artificial Intelligence (AI) have emerged as transformative forces in their respective domains. This paper explores synergies and challenges between these two technologies. Our research analyses the biggest projects combining blockchain and AI, based on market capitalization, and derives a novel framework to categorize contemporary and future use cases. Despite the theoretical compatibility, current real-world applications combining blockchain and AI remain in their infancy.

Researchers and innovators have made enormous efforts in developing ideation methods, such as morphological analysis and design-by-analogy, to aid engineering design ideation for problem solving and innovation. Among these, the Theory of Inventive Problem Solving (TRIZ) stands out as one of the most well-known approaches, widely applied for systematic innovation. However, the complexity of TRIZ resources and concepts, coupled with its reliance on users' knowledge, experience, and reasoning capabilities, limits its practicality. Therefore, we explore the recent advances of large language models (LLMs) for a generative approach to bridge this gap. This paper proposes AutoTRIZ, an artificial ideation tool that uses LLMs to automate and enhance the TRIZ methodology. By leveraging the broad knowledge and advanced reasoning capabilities of LLMs, AutoTRIZ offers a novel approach for design automation and interpretable ideation with artificial intelligence. AutoTRIZ takes a problem statement from the user as its initial input, and automatically generates a solution report after the reasoning process. We demonstrate and evaluate the effectiveness of AutoTRIZ through consistency experiments in contradiction detection, and a case study comparing solutions generated by AutoTRIZ with the experts' analyses from the textbook. Moreover, the proposed LLM-based framework holds the potential for extension to automate other knowledge-based ideation methods, including SCAMPER, Design Heuristics, and Design-by-Analogy, paving the way for a new era of artificial ideation for design innovation.

With the increasing use of large language models (LLMs) in daily life, concerns have emerged regarding their potential misuse and societal impact. Watermarking is proposed to trace the usage of specific models by injecting patterns into their generated texts. An ideal watermark should produce outputs that are nearly indistinguishable from those of the original LLM (imperceptibility), while ensuring a high detection rate (efficacy), even when the text is partially altered (robustness). Despite many methods having been proposed, none have simultaneously achieved all three properties, revealing an inherent trade-off. This paper utilizes a key-centered scheme to unify existing watermarking techniques by decomposing a watermark into two distinct modules: a key module and a mark module. Through this decomposition, we demonstrate for the first time that the key module significantly contributes to the trade-off issues observed in prior methods. Specifically, this reflects the conflict between the scale of the key sampling space during generation and the complexity of key restoration during detection. To this end, we introduce \textbf{WaterPool}, a simple yet effective key module that preserves a complete key sampling space required by imperceptibility while utilizing semantics-based search to improve the key restoration process. WaterPool can integrate with most watermarks, acting as a plug-in. Our experiments with three well-known watermarking techniques show that WaterPool significantly enhances their performance, achieving near-optimal imperceptibility and markedly improving efficacy and robustness (+12.73\% for KGW, +20.27\% for EXP, +7.27\% for ITS).

Recent large language models (LLMs) have witnessed significant advancement in various tasks, including mathematical reasoning and theorem proving. As these two tasks require strict and formal multi-step inference, they are appealing domains for exploring the reasoning ability of LLMs but still face important challenges. Previous studies such as Chain-of-Thought (CoT) have revealed the effectiveness of intermediate steps guidance. However, such step-wise annotation requires heavy labor, leading to insufficient training steps for current benchmarks. To fill this gap, this work introduces MUSTARD, a data generation framework that masters uniform synthesis of theorem and proof data of high quality and diversity. MUSTARD synthesizes data in three stages: (1) It samples a few mathematical concept seeds as the problem category. (2) Then, it prompts a generative language model with the sampled concepts to obtain both the problems and their step-wise formal solutions. (3) Lastly, the framework utilizes a proof assistant (e.g., Lean Prover) to filter the valid proofs. With the proposed MUSTARD, we present a theorem-and-proof benchmark MUSTARDSAUCE with 5,866 valid data points. Each data point contains an informal statement, an informal proof, and a translated formal proof that passes the prover validation. We perform extensive analysis and demonstrate that MUSTARD generates validated high-quality step-by-step data. We further apply the MUSTARDSAUCE for fine-tuning smaller language models. The fine-tuned Llama 2-7B achieves a 15.41% average relative performance gain in automated theorem proving, and 8.18% in math word problems. Codes and data are available at https://github.com/Eleanor-H/MUSTARD.