Quantitative Certification of Bias in Large Language Models

Large Language Models (LLMs) can produce responses that exhibit social biases and support stereotypes. However, conventional benchmarking is insufficient to thoroughly evaluate LLM bias, as it can not scale to large sets of prompts and provides no guarantees. Therefore, we propose a novel certification framework QuaCer-B (Quantitative Certification of Bias) that provides formal guarantees on obtaining unbiased responses from target LLMs under large sets of prompts. A certificate consists of high-confidence bounds on the probability of obtaining biased responses from the LLM for any set of prompts containing sensitive attributes, sampled from a distribution. We illustrate the bias certification in LLMs for prompts with various prefixes drawn from given distributions. We consider distributions of random token sequences, mixtures of manual jailbreaks, and jailbreaks in the LLM's embedding space to certify its bias. We certify popular LLMs with QuaCer-B and present novel insights into their biases.

Toward Informal Language Processing: Knowledge of Slang in Large Language Models

Recent advancement in large language models (LLMs) has offered a strong potential for natural language systems to process informal language. A representative form of informal language is slang, used commonly in daily conversations and online social media. To date, slang has not been comprehensively evaluated in LLMs due partly to the absence of a carefully designed and publicly accessible benchmark. Using movie subtitles, we construct a dataset that supports evaluation on a diverse set of tasks pertaining to automatic processing of slang. For both evaluation and finetuning, we show the effectiveness of our dataset on two core applications: 1) slang detection, and 2) identification of regional and historical sources of slang from natural sentences. We also show how our dataset can be used to probe the output distributions of LLMs for interpretive insights. We find that while LLMs such as GPT-4 achieve good performance in a zero-shot setting, smaller BERT-like models finetuned on our dataset achieve comparable performance. Furthermore, we show that our dataset enables finetuning of LLMs such as GPT-3.5 that achieve substantially better performance than strong zero-shot baselines. Our work offers a comprehensive evaluation and a high-quality benchmark on English slang based on the OpenSubtitles corpus, serving both as a publicly accessible resource and a platform for applying tools for informal language processing.

Net 835-Gb/s/λ Carrier- and LO-Free 100-km Transmission Using Channel-Aware Phase Retrieval Reception

We experimentally demonstrate the first carrier- and LO-free 800G/{\lambda} receiver enabling direct compatibility with standard coherent transmitters via phase retrieval, achieving net 835-Gb/s transmission over 100-km SMF and record 8.27-b/s/Hz net optical spectral efficiency.

Can Small Language Models Help Large Language Models Reason Better?: LM-Guided Chain-of-Thought

We introduce a novel framework, LM-Guided CoT, that leverages a lightweight (i.e., <1B) language model (LM) for guiding a black-box large (i.e., >10B) LM in reasoning tasks. Specifically, the lightweight LM first generates a rationale for each input instance. The Frozen large LM is then prompted to predict a task output based on the rationale generated by the lightweight LM. Our approach is resource-efficient in the sense that it only requires training the lightweight LM. We optimize the model through 1) knowledge distillation and 2) reinforcement learning from rationale-oriented and task-oriented reward signals. We assess our method with multi-hop extractive question answering (QA) benchmarks, HotpotQA, and 2WikiMultiHopQA. Experimental results show that our approach outperforms all baselines regarding answer prediction accuracy. We also find that reinforcement learning helps the model to produce higher-quality rationales with improved QA performance.

Faithful Model Evaluation for Model-Based Metrics

Statistical significance testing is used in natural language processing (NLP) to determine whether the results of a study or experiment are likely to be due to chance or if they reflect a genuine relationship. A key step in significance testing is the estimation of confidence interval which is a function of sample variance. Sample variance calculation is straightforward when evaluating against ground truth. However, in many cases, a metric model is often used for evaluation. For example, to compare toxicity of two large language models, a toxicity classifier is used for evaluation. Existing works usually do not consider the variance change due to metric model errors, which can lead to wrong conclusions. In this work, we establish the mathematical foundation of significance testing for model-based metrics. With experiments on public benchmark datasets and a production system, we show that considering metric model errors to calculate sample variances for model-based metrics changes the conclusions in certain experiments.

Evaluating Large Language Models on Controlled Generation Tasks

While recent studies have looked into the abilities of large language models in various benchmark tasks, including question generation, reading comprehension, multilingual and etc, there have been few studies looking into the controllability of large language models on generation tasks. We present an extensive analysis of various benchmarks including a sentence planning benchmark with different granularities. After comparing large language models against state-of-the-start finetuned smaller models, we present a spectrum showing large language models falling behind, are comparable, or exceed the ability of smaller models. We conclude that **large language models struggle at meeting fine-grained hard constraints**.

FLIRT: Feedback Loop In-context Red Teaming

Warning: this paper contains content that may be inappropriate or offensive. As generative models become available for public use in various applications, testing and analyzing vulnerabilities of these models has become a priority. Here we propose an automatic red teaming framework that evaluates a given model and exposes its vulnerabilities against unsafe and inappropriate content generation. Our framework uses in-context learning in a feedback loop to red team models and trigger them into unsafe content generation. We propose different in-context attack strategies to automatically learn effective and diverse adversarial prompts for text-to-image models. Our experiments demonstrate that compared to baseline approaches, our proposed strategy is significantly more effective in exposing vulnerabilities in Stable Diffusion (SD) model, even when the latter is enhanced with safety features. Furthermore, we demonstrate that the proposed framework is effective for red teaming text-to-text models, resulting in significantly higher toxic response generation rate compared to previously reported numbers.

Is the Elephant Flying? Resolving Ambiguities in Text-to-Image Generative Models

Natural language often contains ambiguities that can lead to misinterpretation and miscommunication. While humans can handle ambiguities effectively by asking clarifying questions and/or relying on contextual cues and common-sense knowledge, resolving ambiguities can be notoriously hard for machines. In this work, we study ambiguities that arise in text-to-image generative models. We curate a benchmark dataset covering different types of ambiguities that occur in these systems. We then propose a framework to mitigate ambiguities in the prompts given to the systems by soliciting clarifications from the user. Through automatic and human evaluations, we show the effectiveness of our framework in generating more faithful images aligned with human intention in the presence of ambiguities.

Balancing the trade-off between cost and reliability for wireless sensor networks: a multi-objective optimized deployment method

The deployment of the sensor nodes (SNs) always plays a decisive role in the system performance of wireless sensor networks (WSNs). In this work, we propose an optimal deployment method for practical heterogeneous WSNs which gives a deep insight into the trade-off between the reliability and deployment cost. Specifically, this work aims to provide the optimal deployment of SNs to maximize the coverage degree and connection degree, and meanwhile minimize the overall deployment cost. In addition, this work fully considers the heterogeneity of SNs (i.e. differentiated sensing range and deployment cost) and three-dimensional (3-D) deployment scenarios. This is a multi-objective optimization problem, non-convex, multimodal and NP-hard. To solve it, we develop a novel swarm-based multi-objective optimization algorithm, known as the competitive multi-objective marine predators algorithm (CMOMPA) whose performance is verified by comprehensive comparative experiments with ten other stateof-the-art multi-objective optimization algorithms. The computational results demonstrate that CMOMPA is superior to others in terms of convergence and accuracy and shows excellent performance on multimodal multiobjective optimization problems. Sufficient simulations are also conducted to evaluate the effectiveness of the CMOMPA based optimal SNs deployment method. The results show that the optimized deployment can balance the trade-off among deployment cost, sensing reliability and network reliability. The source code is available on https://github.com/iNet-WZU/CMOMPA.

Predicting vacant parking space availability zone-wisely: a graph based spatio-temporal prediction approach

Vacant parking space (VPS) prediction is one of the key issues of intelligent parking guidance systems. Accurately predicting VPS information plays a crucial role in intelligent parking guidance systems, which can help drivers find parking space quickly, reducing unnecessary waste of time and excessive environmental pollution. Through the simple analysis of historical data, we found that there not only exists a obvious temporal correlation in each parking lot, but also a clear spatial correlation between different parking lots. In view of this, this paper proposed a graph data-based model ST-GBGRU (Spatial-Temporal Graph Based Gated Recurrent Unit), the number of VPSs can be predicted both in short-term (i.e., within 30 min) and in long-term (i.e., over 30min). On the one hand, the temporal correlation of historical VPS data is extracted by GRU, on the other hand, the spatial correlation of historical VPS data is extracted by GCN inside GRU. Two prediction methods, namely direct prediction and iterative prediction, are combined with the proposed model. Finally, the prediction model is applied to predict the number VPSs of 8 public parking lots in Santa Monica. The results show that in the short-term and long-term prediction tasks, ST-GBGRU model can achieve high accuracy and have good application prospects.