Despite the advances in the abstractive summarization task using Large Language Models (LLM), there is a lack of research that asses their abilities to easily adapt to different domains. We evaluate the domain adaptation abilities of a wide range of LLMs on the summarization task across various domains in both fine-tuning and in-context learning settings. We also present AdaptEval, the first domain adaptation evaluation suite. AdaptEval includes a domain benchmark and a set of metrics to facilitate the analysis of domain adaptation. Our results demonstrate that LLMs exhibit comparable performance in the in-context learning setting, regardless of their parameter scale.

Security concerns for large language models (LLMs) have recently escalated, focusing on thwarting jailbreaking attempts in discrete prompts. However, the exploration of jailbreak vulnerabilities arising from continuous embeddings has been limited, as prior approaches primarily involved appending discrete or continuous suffixes to inputs. Our study presents a novel channel for conducting direct attacks on LLM inputs, eliminating the need for suffix addition or specific questions provided that the desired output is predefined. We additionally observe that extensive iterations often lead to overfitting, characterized by repetition in the output. To counteract this, we propose a simple yet effective strategy named CLIP. Our experiments show that for an input length of 40 at iteration 1000, applying CLIP improves the ASR from 62% to 83%

The semantic capabilities of language models (LMs) have the potential to enable rich analytics and reasoning over vast knowledge corpora. Unfortunately, existing systems lack high-level abstractions to perform semantic queries at scale. We introduce semantic operators, a declarative programming interface that extends the relational model with composable AI-based operations for semantic queries over datasets (e.g., sorting or aggregating records using natural language criteria). Each operator can be implemented and optimized in multiple ways, opening a rich space for execution plans similar to relational operators. We implement our operators and several optimizations for them in LOTUS, an open-source query engine with a Pandas-like API. We demonstrate LOTUS' effectiveness across a series of real applications, including fact-checking, extreme multi-label classification, and search. We find that LOTUS' programming model is highly expressive, capturing state-of-the-art query pipelines with low development overhead. Specifically, on the FEVER dataset, LOTUS' programs can reproduce FacTool, a recent state-of-the-art fact-checking pipeline, in few lines of code, and implement a new pipeline that improves accuracy by $9.5\%$, while offering $7-34\times$ lower execution time. In the extreme multi-label classification task on the BioDEX dataset, LOTUS reproduces state-of-the art result quality with its join operator, while providing an efficient algorithm that runs $800\times$ faster than a naive join. In the search and ranking application, LOTUS allows a simple composition of operators to achieve $5.9 - 49.4\%$ higher nDCG@10 than the vanilla retriever and re-ranker, while also providing query efficiency, with $1.67 - 10\times$ lower execution time than LM-based ranking methods used by prior works. LOTUS is publicly available at https://github.com/stanford-futuredata/lotus.

This study seeks to identify and quantify biases in simulating political samples with Large Language Models, specifically focusing on vote choice and public opinion. Using the GPT-3.5-Turbo model, we leverage data from the American National Election Studies, German Longitudinal Election Study, Zuobiao Dataset, and China Family Panel Studies to simulate voting behaviors and public opinions. This methodology enables us to examine three types of representation bias: disparities based on the the country's language, demographic groups, and political regime types. The findings reveal that simulation performance is generally better for vote choice than for public opinions, more accurate in English-speaking countries, more effective in bipartisan systems than in multi-partisan systems, and stronger in democratic settings than in authoritarian regimes. These results contribute to enhancing our understanding and developing strategies to mitigate biases in AI applications within the field of computational social science.

In recent years, researchers have proposed numerous benchmarks to evaluate the impressive coding capabilities of large language models (LLMs). However, existing benchmarks primarily focus on assessing the correctness of code generated by LLMs, while neglecting other critical dimensions that also significantly impact code quality. Therefore, this paper proposes the RACE benchmark, which comprehensively evaluates the quality of code generated by LLMs across 4 dimensions: Readability, mAintainability, Correctness, and Efficiency. Specifically, considering the demand-dependent nature of dimensions beyond correctness, we design various types of user requirements for each dimension to assess the model's ability to generate correct code that also meets user demands. We evaluate 18 representative LLMs on RACE and find that: 1) the current LLMs' ability to generate high-quality code on demand does not yet meet the requirements of software development; 2) readability serves as a critical indicator of the overall quality of generated code; 3) most LLMs exhibit an inherent preference for specific coding style. These findings can help researchers gain a deeper understanding of the coding capabilities of current LLMs and shed light on future directions for model improvement.

Despite their linguistic competence, Large Language models (LLMs) often exhibit limitations in their ability to reason reliably and flexibly. To address this, we propose a neurosymbolic approach that prompts LLMs to extract and encode all relevant information from a problem statement as logical code statements, and then use a logic programming language (Prolog) to conduct the iterative computations of explicit deductive reasoning. Our approach significantly enhances the performance of LLMs on the standard mathematical reasoning benchmark, GSM8k, and the Navigate dataset from the BIG-bench dataset. Additionally, we introduce a novel dataset, the Non-Linear Reasoning (NLR) dataset, consisting of 55 unique word problems that target the shortcomings of the next token prediction paradigm of LLMs and require complex non-linear reasoning but only basic arithmetic skills to solve. Our findings demonstrate that the integration of Prolog enables LLMs to achieve high performance on the NLR dataset, which even the most advanced language models (including GPT4) fail to solve using text only.

Recent advancements in multimodal foundation models have showcased impressive capabilities in understanding and reasoning with visual and textual information. Adapting these foundation models trained for general usage to specialized domains like biomedicine requires large-scale domain-specific instruction datasets. While existing works have explored curating such datasets automatically, the resultant datasets are not explicitly aligned with domain expertise. In this work, we propose a data-centric framework, Biomedical Visual Instruction Tuning with Clinician Preference Alignment (BioMed-VITAL), that incorporates clinician preferences into both stages of generating and selecting instruction data for tuning biomedical multimodal foundation models. First, during the generation stage, we prompt the GPT-4V generator with a diverse set of clinician-selected demonstrations for preference-aligned data candidate generation. Then, during the selection phase, we train a separate selection model, which explicitly distills clinician and policy-guided model preferences into a rating function to select high-quality data for medical instruction tuning. Results show that the model tuned with the instruction-following data from our method demonstrates a significant improvement in open visual chat (18.5% relatively) and medical VQA (win rate up to 81.73%). Our instruction-following data and models are available at BioMed-VITAL.github.io.

Evaluating modern ML models is hard. The strong incentive for researchers and companies to report a state-of-the-art result on some metric often leads to questionable research practices (QRPs): bad practices which fall short of outright research fraud. We describe 43 such practices which can undermine reported results, giving examples where possible. Our list emphasises the evaluation of large language models (LLMs) on public benchmarks. We also discuss "irreproducible research practices", i.e. decisions that make it difficult or impossible for other researchers to reproduce, build on or audit previous research.

Autoregressive language models demonstrate excellent performance in various scenarios. However, the inference efficiency is limited by its one-step-one-word generation mode, which has become a pressing problem recently as the models become increasingly larger. Speculative decoding employs a "draft and then verify" mechanism to allow multiple tokens to be generated in one step, realizing lossless acceleration. Existing methods mainly adopt fixed heuristic draft structures, which fail to adapt to different situations to maximize the acceptance length during verification. To alleviate this dilemma, we proposed OPT-Tree, an algorithm to construct adaptive and scalable draft trees. It searches the optimal tree structure that maximizes the mathematical expectation of the acceptance length in each decoding step. Experimental results reveal that OPT-Tree outperforms the existing draft structures and achieves a speed-up ratio of up to 3.2 compared with autoregressive decoding. If the draft model is powerful enough and the node budget is sufficient, it can generate more than ten tokens in a single step. Our code is available at https://github.com/Jikai0Wang/OPT-Tree.

Sentence embedding is a cornerstone in NLP. Whitening has been claimed to be an effective operation to improve embedding quality obtained from Large Language Models (LLMs). However, we find that the efficacy of whitening is model-dependent and task-dependent. In particular, whitening degenerates embeddings for classification tasks. The conclusion is supported by extensive experiments. We also explored a variety of whitening operations, including PCA, ZCA, PCA-Cor, ZCA-Cor and Cholesky whitenings. A by-product of our research is embedding evaluation platform for LLMs called SentEval+.