Fine-tuning Large Language Models (LLMs) on some task-specific datasets has been a primary use of LLMs. However, it has been empirically observed that this approach to enhancing capability inevitably compromises safety, a phenomenon also known as the safety-capability trade-off in LLM fine-tuning. This paper presents a theoretical framework for understanding the interplay between safety and capability in two primary safety-aware LLM fine-tuning strategies, providing new insights into the effects of data similarity, context overlap, and alignment loss landscape. Our theoretical results characterize the fundamental limits of the safety-capability trade-off in LLM fine-tuning, which are also validated by numerical experiments.

Large language models often expose their brittleness in reasoning tasks, especially while executing long chains of reasoning over context. We propose MemReasoner, a new and simple memory-augmented LLM architecture, in which the memory learns the relative order of facts in context, and enables hopping over them, while the decoder selectively attends to the memory. MemReasoner is trained end-to-end, with optional supporting fact supervision of varying degrees. We train MemReasoner, along with existing memory-augmented transformer models and a state-space model, on two distinct synthetic multi-hop reasoning tasks. Experiments performed under a variety of challenging scenarios, including the presence of long distractor text or target answer changes in test set, show strong generalization of MemReasoner on both single- and two-hop tasks. This generalization of MemReasoner is achieved using none-to-weak supporting fact supervision (using none and 1\% of supporting facts for one- and two-hop tasks, respectively). In contrast, baseline models overall struggle to generalize and benefit far less from using full supporting fact supervision. The results highlight the importance of explicit memory mechanisms, combined with additional weak supervision, for improving large language model's context processing ability toward reasoning tasks.

Recent advances in Large Language Models (LLMs) have yielded impressive successes on many language tasks. However, efficient processing of long contexts using LLMs remains a significant challenge. We introduce \textbf{EpMAN} -- a method for processing long contexts in an \textit{episodic memory} module while \textit{holistically attending to} semantically relevant context chunks. The output of \textit{episodic attention} is then used to reweigh the decoder's self-attention to the stored KV cache of the context during training and generation. When an LLM decoder is trained using \textbf{EpMAN}, its performance on multiple challenging single-hop long-context recall and question-answering benchmarks is found to be stronger and more robust across the range from 16k to 256k tokens than baseline decoders trained with self-attention, and popular retrieval-augmented generation frameworks.

As the research community aims to build better AI assistants that are more dynamic and personalized to the diversity of humans that they interact with, there is increased interest in evaluating the theory of mind capabilities of large language models (LLMs). Indeed, several recent studies suggest that LLM theory of mind capabilities are quite impressive, approximating human-level performance. Our paper aims to rebuke this narrative and argues instead that past studies were not directly measuring agent performance, potentially leading to findings that are illusory in nature as a result. We draw a strong distinction between what we call literal theory of mind i.e. measuring the agent's ability to predict the behavior of others and functional theory of mind i.e. adapting to agents in-context based on a rational response to predictions of their behavior. We find that top performing open source LLMs may display strong capabilities in literal theory of mind, depending on how they are prompted, but seem to struggle with functional theory of mind -- even when partner policies are exceedingly simple. Our work serves to highlight the double sided nature of inductive bias in LLMs when adapting to new situations. While this bias can lead to strong performance over limited horizons, it often hinders convergence to optimal long-term behavior.

Designing novel functional proteins crucially depends on accurately modeling their fitness landscape. Given the limited availability of functional annotations from wet-lab experiments, previous methods have primarily relied on self-supervised models trained on vast, unlabeled protein sequence or structure datasets. While initial protein representation learning studies solely focused on either sequence or structural features, recent hybrid architectures have sought to merge these modalities to harness their respective strengths. However, these sequence-structure models have so far achieved only incremental improvements when compared to the leading sequence-only approaches, highlighting unresolved challenges effectively leveraging these modalities together. Moreover, the function of certain proteins is highly dependent on the granular aspects of their surface topology, which have been overlooked by prior models. To address these limitations, we introduce the Sequence-Structure-Surface Fitness (S3F) model -- a novel multimodal representation learning framework that integrates protein features across several scales. Our approach combines sequence representations from a protein language model with Geometric Vector Perceptron networks encoding protein backbone and detailed surface topology. The proposed method achieves state-of-the-art fitness prediction on the ProteinGym benchmark encompassing 217 substitution deep mutational scanning assays, and provides insights into the determinants of protein function.

There is a growing interest in training domain-expert LLMs that excel in specific technical fields compared to their general-purpose instruction-tuned counterparts. However, these expert models often experience a loss in their safety abilities in the process, making them capable of generating harmful content. As a solution, we introduce an efficient and effective merging-based alignment method called \textsc{MergeAlign} that interpolates the domain and alignment vectors, creating safer domain-specific models while preserving their utility. We apply \textsc{MergeAlign} on Llama3 variants that are experts in medicine and finance, obtaining substantial alignment improvements with minimal to no degradation on domain-specific benchmarks. We study the impact of model merging through model similarity metrics and contributions of individual models being merged. We hope our findings open new research avenues and inspire more efficient development of safe expert LLMs.

Fine-tuning on task-specific data to boost downstream performance is a crucial step for leveraging Large Language Models (LLMs). However, previous studies have demonstrated that fine-tuning the models on several adversarial samples or even benign data can greatly comprise the model's pre-equipped alignment and safety capabilities. In this work, we propose SEAL, a novel framework to enhance safety in LLM fine-tuning. SEAL learns a data ranker based on the bilevel optimization to up rank the safe and high-quality fine-tuning data and down rank the unsafe or low-quality ones. Models trained with SEAL demonstrate superior quality over multiple baselines, with 8.5% and 9.7% win rate increase compared to random selection respectively on Llama-3-8b-Instruct and Merlinite-7b models. Our code is available on github https://github.com/hanshen95/SEAL.

This paper contains examples that may be considered offensive and inappropriate. Reducing the likelihood of generating harmful and toxic output is an essential task when aligning large language models (LLMs). Existing methods mainly rely on training an external reward model (i.e., another language model) or fine-tuning the LLM using self-generated data to influence the outcome. In this paper, we show that LLMs have the capability of self-detoxification without the use of an additional reward model or re-training. We propose Self-disciplined Autoregressive Sampling (SASA), a lightweight controlled decoding algorithm for toxicity reduction of LLMs. SASA leverages the contextual representations from an LLM to learn linear subspaces characterizing toxic v.s. When auto-completing a response token-by-token, SASA dynamically tracks the margin of the current output to steer the generation away from the toxic subspace, by adjusting the autoregressive sampling strategy. Recent advancements in large language models (LLMs) have dramatically enhanced their capabilities in textual understanding and reasoning (Brown et al., 2020; Kojima et al., 2022). Their capabilities in performing diverse linguistic tasks and producing coherent texts have catalyzed their adoption across a variety of applications (Rae et al., 2021; Hoffmann et al., 2022; Le Scao et al., 2023; Touvron et al., 2023a;b; Achiam et al., 2023). However, with the escalating size of models (Raffel et al., 2020; Brown et al., 2020; Achiam et al., 2023), there is a corresponding increase in the scale of the training datasets required to avert overfitting and to encapsulate extensive world knowledge. These extensive datasets, predominantly derived from internet crawls and merely subjected to basic filtering protocols (Raffel et al., 2020), often harbor biases that are problematic or directly detrimental for many applications and may not inherently align with these desirable attributes (Wallace et al., 2019; Gehman et al., 2020). In fact, it is known that language models trained on such data may not only mimic but also amplify these biases (Bolukbasi et al., 2016; Caliskan et al., 2017; Zhao et al., 2018; Sheng et al., 2019; Gehman et al., 2020; Hartvigsen et al., For example, an "aligned" LLM may be inadvertently or maliciously tricked into generating harmful or toxic output that causes usage violations and safety concerns (Sun et al., 2024).

Current large language models (LLMs) often perform poorly on simple fact retrieval tasks. Here we investigate if coupling a dynamically adaptable external memory to a LLM can alleviate this problem. For this purpose, we test Larimar, a recently proposed language model architecture which uses an external associative memory, on long-context recall tasks including passkey and needle-in-the-haystack tests. We demonstrate that the external memory of Larimar, which allows fast write and read of an episode of text samples, can be used at test time to handle contexts much longer than those seen during training. We further show that the latent readouts from the memory (to which long contexts are written) control the decoder towards generating correct outputs, with the memory stored off of the GPU. Compared to existing transformer-based LLM architectures for long-context recall tasks that use larger parameter counts or modified attention mechanisms, a relatively smaller size Larimar is able to maintain strong performance without any task-specific training or training on longer contexts.

Efficient and accurate updating of knowledge stored in Large Language Models (LLMs) is one of the most pressing research challenges today. This paper presents Larimar - a novel, brain-inspired architecture for enhancing LLMs with a distributed episodic memory. Larimar's memory allows for dynamic, one-shot updates of knowledge without the need for computationally expensive re-training or fine-tuning. Experimental results on multiple fact editing benchmarks demonstrate that Larimar attains accuracy comparable to most competitive baselines, even in the challenging sequential editing setup, but also excels in speed - yielding speed-ups of 8-10x depending on the base LLM - as well as flexibility due to the proposed architecture being simple, LLM-agnostic, and hence general. We further provide mechanisms for selective fact forgetting, information leakage prevention, and input context length generalization with Larimar and show their effectiveness. Our code is available at https://github.com/IBM/larimar