Recent advances in Large Language Models (LLM) have led to substantial interest in their application to commonsense reasoning tasks. Despite their potential, LLMs are susceptible to reasoning errors and hallucinations that may be harmful in use cases where accurate reasoning is critical. This challenge underscores the need for verifiable, debuggable, and repairable LLM reasoning. Recent works have made progress toward verifiable reasoning with LLMs by using them as either (i) a reasoner over an axiomatic knowledge base, or (ii) a semantic parser for use in existing logical inference systems. However, both settings are unable to extract commonsense axioms from the LLM that are not already formalized in the knowledge base, and also lack a reliable method to repair missed commonsense inferences. In this work, we present LLM-TRes, a logical reasoning framework based on the notion of “theory resolution” that allows for seamless integration of the commonsense knowledge from LLMs with a verifiable logical reasoning framework that mitigates hallucinations and facilitates debugging of the reasoning procedure as well as repair. We crucially prove that repaired axioms are theoretically guaranteed to be given precedence over flawed ones in our theory resolution inference process. We conclude by evaluating on three diverse language-based reasoning tasks—preference reasoning, deductive reasoning, and causal commonsense reasoning—and demonstrate the superior performance of LLM-TRes vs. state-of-the-art LLM-based reasoning methods in terms of both accuracy and reasoning correctness.

Due to emergent capabilities, large language models (LLMs) have been utilized as language-based agents to perform a variety of tasks and make decisions with an increasing degree of autonomy. These autonomous agents can understand high-level instructions, interact with their environments, and execute complex tasks using a selection of tools available to them. As the capabilities of the agents expand, ensuring their safety and trustworthiness becomes more imperative. In this study, we introduce the Athena framework which leverages the concept of verbal contrastive learning where past safe and unsafe trajectories are used as in-context (contrastive) examples to guide the agent towards safety while fulfilling a given task. The framework also incorporates a critiquing mechanism to guide the agent to prevent risky actions at every step. Furthermore, due to the lack of existing benchmarks on the safety reasoning ability of LLM-based agents, we curate a set of 80 toolkits across 8 categories with 180 scenarios to provide a safety evaluation benchmark. Our experimental evaluation, with both closed- and open-source LLMs, indicates verbal contrastive learning and interaction-level critiquing improve the safety rate significantly.

In multi-objective text generation, we aim to optimize over multiple weighted aspects (e.g., toxicity, semantic preservation, fluency) of the generated text. However, multi-objective weighting schemes may change dynamically in practice according to deployment requirements, evolving business needs, personalization requirements on edge devices, or the availability of new language models and/or objective requirements. Ideally, we need an efficient method to adapt to the dynamic requirements of the overall objective. To address these requirements, we propose a linear combination of objective-specific language models to efficiently adapt the decoding process and optimize for the desired objective without the significant computational overhead of retraining one or more language models. We show empirically that we can leverage Gaussian Process black box optimization to adapt the language model decoder weights to outperform other fixed weighting schemes and standard baselines of the task in only a few iterations of decoding. Overall this approach enables highly efficient adaptation of controllable language models via multi-objective weighting schemes that may evolve dynamically in practical deployment situations.

Text detoxification is a conditional text generation task aiming to remove offensive content from toxic text. It is highly useful for online forums and social media, where offensive content is frequently encountered. Intuitively, there are diverse ways to detoxify sentences while preserving their meanings, and we can select from detoxified sentences before displaying text to users. Conditional diffusion models are particularly suitable for this task given their demonstrated higher generative diversity than existing conditional text generation models based on language models. Nonetheless, text fluency declines when they are trained with insufficient data, which is the case for this task. In this work, we propose DiffuDetox, a mixed conditional and unconditional diffusion model for text detoxification. The conditional model takes toxic text as the condition and reduces its toxicity, yielding a diverse set of detoxified sentences. The unconditional model is trained to recover the input text, which allows the introduction of additional fluent text for training and thus ensures text fluency. Extensive experimental results and in-depth analysis demonstrate the effectiveness of our proposed DiffuDetox.

Offensive and toxic text on social media platforms can lead to polarization and divisiveness within online communities and hinders constructive dialogue. Text detoxification is a crucial task in natural language processing to ensure the generation of non-toxic and safe text. Text detoxification is a special case of the Text Style Transfer (TST) problem, where an input text is rephrased to an output text that preserves its content while modifying the style (in this case to a more neutral, non-toxic style). State-of-the-art methods for detoxification use supervised training of encoder-decoder models to produce gold-standard outputs with a standard likelihood-based objective. However, it can be hard for these models to deviate from their pretrained auto-encoder identity mapping. While previous methods have used unlikelihood-based losses to penalize input-to-output copying of toxic content, these methods also unfortunately penalize non-toxic content in the input that would be fine to preserve in the output. To address these issues, we introduce a novel contrastive unlikelihood objective (COUNT) that directly contrasts the gold standard rephrasing with the identity input-to-output mapping to effectively isolate and focus learning on non-toxic style transfer. We benchmark COUNT on two parallel datasets, ParaDetox and APPDIA, showing that it achieves significant improvements in jointly combined fluency, content preservation, and detoxification (i.e., the highest “J” score).