Prior work has shown that the ordering in which concepts are shown to a commonsense generator plays an important role, affecting the quality of the generated sentence. However, it remains a challenge to determine the optimal ordering of a given set of concepts such that a natural sentence covering all the concepts could be generated from a pretrained generator. To understand the relationship between the ordering of the input concepts and the quality of the generated sentences, we conduct a systematic study considering multiple language models (LMs) and concept ordering strategies. We find that BART-large model consistently outperforms all other LMs considered in this study when fine-tuned using the ordering of concepts as they appear in CommonGen training data as measured using multiple evaluation metrics. Moreover, the larger GPT3-based large language models (LLMs) variants do not necessarily outperform much smaller LMs on this task, even when fine-tuned on task-specific training data. Interestingly, human annotators significantly reorder input concept sets when manually writing sentences covering those concepts, and this ordering provides the best sentence generations independently of the LM used for the generation, outperforming a probabilistic concept ordering baseline

Reasoning presents a significant and challenging issue for Large Language Models (LLMs). The predominant focus of research has revolved around developing diverse prompting strategies to guide and structure the reasoning processes of LLMs. However, these approaches based on decoder-only causal language models often operate the input question in a single forward pass, potentially missing the rich, back-and-forth interactions inherent in human reasoning. Scant attention has been paid to a critical dimension, i.e., the input question itself embedded within the prompts. In response, we introduce a deceptively simple yet highly effective prompting strategy, termed question "re-reading". Drawing inspiration from human learning and problem-solving, re-reading entails revisiting the question information embedded within input prompts. This approach aligns seamlessly with the cognitive principle of reinforcement, enabling LLMs to extract deeper insights, identify intricate patterns, establish more nuanced connections, and ultimately enhance their reasoning capabilities across various tasks. Experiments conducted on a series of reasoning benchmarks serve to underscore the effectiveness and generality of our method. Moreover, our findings demonstrate that our approach seamlessly integrates with various language models, though-eliciting prompting methods, and ensemble techniques, further underscoring its versatility and compatibility in the realm of LLMs.

High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4$\times$ with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm

Large language models excel in many human-language tasks but often falter in highly specialized domains like scholarly astronomy. To bridge this gap, we introduce AstroLLaMA, a 7-billion-parameter model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv. Optimized for traditional causal language modeling, AstroLLaMA achieves a 30% lower perplexity than Llama-2, showing marked domain adaptation. Our model generates more insightful and scientifically relevant text completions and embedding extraction than state-of-the-arts foundation models despite having significantly fewer parameters. AstroLLaMA serves as a robust, domain-specific model with broad fine-tuning potential. Its public release aims to spur astronomy-focused research, including automatic paper summarization and conversational agent development.

The cross-lingual transfer is a promising technique to solve tasks in less-resourced languages. In this empirical study, we compare two fine-tuning approaches combined with zero-shot and full-shot learning approaches for large language models in a cross-lingual setting. As fine-tuning strategies, we compare parameter-efficient adapter methods with fine-tuning of all parameters. As cross-lingual transfer strategies, we compare the intermediate-training (\textit{IT}) that uses each language sequentially and cross-lingual validation (\textit{CLV}) that uses a target language already in the validation phase of fine-tuning. We assess the success of transfer and the extent of catastrophic forgetting in a source language due to cross-lingual transfer, i.e., how much previously acquired knowledge is lost when we learn new information in a different language. The results on two different classification problems, hate speech detection and product reviews, each containing datasets in several languages, show that the \textit{IT} cross-lingual strategy outperforms \textit{CLV} for the target language. Our findings indicate that, in the majority of cases, the \textit{CLV} strategy demonstrates superior retention of knowledge in the base language (English) compared to the \textit{IT} strategy, when evaluating catastrophic forgetting in multiple cross-lingual transfers.

As social media becomes increasingly popular, more and more public health activities emerge, which is worth noting for pandemic monitoring and government decision-making. Current techniques for public health analysis involve popular models such as BERT and large language models (LLMs). Although recent progress in LLMs has shown a strong ability to comprehend knowledge by being fine-tuned on specific domain datasets, the costs of training an in-domain LLM for every specific public health task are especially expensive. Furthermore, such kinds of in-domain datasets from social media are generally highly imbalanced, which will hinder the efficiency of LLMs tuning. To tackle these challenges, the data imbalance issue can be overcome by sophisticated data augmentation methods for social media datasets. In addition, the ability of the LLMs can be effectively utilised by prompting the model properly. In light of the above discussion, in this paper, a novel ALEX framework is proposed for social media analysis on public health. Specifically, an augmentation pipeline is developed to resolve the data imbalance issue. Furthermore, an LLMs explanation mechanism is proposed by prompting an LLM with the predicted results from BERT models. Extensive experiments conducted on three tasks at the Social Media Mining for Health 2023 (SMM4H) competition with the first ranking in two tasks demonstrate the superior performance of the proposed ALEX method. Our code has been released in https://github.com/YanJiangJerry/ALEX.

While large language models excel in a variety of natural language processing (NLP) tasks, to perform well on spoken language understanding (SLU) tasks, they must either rely on off-the-shelf automatic speech recognition (ASR) systems for transcription, or be equipped with an in-built speech modality. This work focuses on the former scenario, where LLM's accuracy on SLU tasks is constrained by the accuracy of a fixed ASR system on the spoken input. Specifically, we tackle speech-intent classification task, where a high word-error-rate can limit the LLM's ability to understand the spoken intent. Instead of chasing a high accuracy by designing complex or specialized architectures regardless of deployment costs, we seek to answer how far we can go without substantially changing the underlying ASR and LLM, which can potentially be shared by multiple unrelated tasks. To this end, we propose prompting the LLM with an n-best list of ASR hypotheses instead of only the error-prone 1-best hypothesis. We explore prompt-engineering to explain the concept of n-best lists to the LLM; followed by the finetuning of Low-Rank Adapters on the downstream tasks. Our approach using n-best lists proves to be effective on a device-directed speech detection task as well as on a keyword spotting task, where systems using n-best list prompts outperform those using 1-best ASR hypothesis; thus paving the way for an efficient method to exploit ASR uncertainty via LLMs for speech-based applications.

As social media becomes increasingly popular, more and more activities related to public health emerge. Current techniques for public health analysis involve popular models such as BERT and large language models (LLMs). However, the costs of training in-domain LLMs for public health are especially expensive. Furthermore, such kinds of in-domain datasets from social media are generally imbalanced. To tackle these challenges, the data imbalance issue can be overcome by data augmentation and balanced training. Moreover, the ability of the LLMs can be effectively utilized by prompting the model properly. In this paper, a novel ALEX framework is proposed to improve the performance of public health analysis on social media by adopting an LLMs explanation mechanism. Results show that our ALEX model got the best performance among all submissions in both Task 2 and Task 4 with a high score in Task 1 in Social Media Mining for Health 2023 (SMM4H)[1]. Our code has been released at https:// github.com/YanJiangJerry/ALEX.

The application of Large Language Models (LLMs) to the medical domain has stimulated the interest of researchers. Recent studies have focused on constructing Instruction Fine-Tuning (IFT) data through medical knowledge graphs to enrich the interactive medical knowledge of LLMs. However, the medical literature serving as a rich source of medical knowledge remains unexplored. Our work introduces the CALLA dataset to probe LLMs' interactive knowledge acquisition from Chinese medical literature. It assesses the proficiency of LLMs in mastering medical knowledge through a free-dialogue fact-checking task. We identify a phenomenon called the ``fact-following response``, where LLMs tend to affirm facts mentioned in questions and display a reluctance to challenge them. To eliminate the inaccurate evaluation caused by this phenomenon, for the golden fact, we artificially construct test data from two perspectives: one consistent with the fact and one inconsistent with the fact. Drawing from the probing experiment on the CALLA dataset, we conclude that IFT data highly correlated with the medical literature corpus serves as a potent catalyst for LLMs, enabling themselves to skillfully employ the medical knowledge acquired during the pre-training phase within interactive scenarios, enhancing accuracy. Furthermore, we design a framework for automatically constructing IFT data based on medical literature and discuss some real-world applications.

Previous studies have typically assumed that large language models are unable to accurately perform arithmetic operations, particularly multiplication of >8 digits, and operations involving decimals and fractions, without the use of calculator tools. This paper aims to challenge this misconception. With sufficient training data, a 2 billion-parameter language model can accurately perform multi-digit arithmetic operations with almost 100% accuracy without data leakage, significantly surpassing GPT-4 (whose multi-digit multiplication accuracy is only 4.3%). We also demonstrate that our MathGLM, fine-tuned from GLM-10B on a dataset with additional multi-step arithmetic operations and math problems described in text, achieves similar performance to GPT-4 on a 5,000-samples Chinese math problem test set. Our code and data are public at https://github.com/THUDM/MathGLM.