Publications

Not All Data Are Unlearned Equally

Machine unlearning is concerned with the task of removing knowledge learned from particular data points from a trained model. In the context… (see more) of large language models (LLMs), unlearning has recently received increased attention, particularly for removing knowledge about named entities from models for privacy purposes. While various approaches have been proposed to address the unlearning problem, most existing approaches treat all data points to be unlearned equally, i.e., unlearning that Montreal is a city in Canada is treated exactly the same as unlearning the phone number of the first author of this paper. In this work, we show that this all data is equal assumption does not hold for LLM unlearning. We study how the success of unlearning depends on the frequency of the knowledge we want to unlearn in the pre-training data of a model and find that frequency strongly affects unlearning, i.e., more frequent knowledge is harder to unlearn. Additionally, we uncover a misalignment between probability and generation-based evaluations of unlearning and show that this problem worsens as models become larger. Overall, our experiments highlight the need for better evaluation practices and novel methods for LLM unlearning that take the training data of models into account.

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

doi.org

openreview.net

Partial Perspectives: How LLMs Handle Logically Inconsistent Knowledge in Reasoning Tasks

Zichao Li

Ines Arous

Jackie Cheung

Most natural language reasoning tasks in the research community assume consistent input knowledge. Nevertheless, real-world scenarios often … (see more)involve inconsistent information, which might lead to divergent conclusions and are typically associated with varying levels of uncertainty. This raises a key research question: can large language models (LLMs) effectively handle uncertainty in their reasoning process to maximize knowledge consistency? In this paper, we propose a framework for evaluating reasoning over inconsistent knowledge. Our approach models uncertainty via weights of logical rules, leveraging Markov logic networks (MLN), which integrate probabilistic reasoning with first-order logic. This enables us to quantify inconsistencies in knowledge bases, and hence rigorously evaluate LLM reasoning. We introduce two tasks using this framework: 1) QA, which involves answering questions by integrating inconsistent knowledge; and 2) knowledge rectification, where we aim to rectify language models' acquired knowledge to improve consistency. We curate a dataset of 3,000 MLN-formatted knowledge bases to implement these tasks. We evaluate state-of-the-art LLMs on these tasks and highlight their limitations in uncertainty-aware reasoning over inconsistent logical knowledge.

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

openreview.net

Putting the Value Back in RL: Better Test-Time Scaling by Unifying LLM Reasoners With Verifiers

Morgane M Moss

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

doi.org

openreview.net

Rethinking Safety in LLM Fine-tuning: An Optimization Perspective

Minseon Kim

Jin Myung Kwak

Lama Alssum

Bernard Ghanem

Philip Torr

David Scott Krueger

Fazl Barez

Adel Bibi

Fine-tuning language models is commonly believed to inevitably harm their safety, i.e., refusing to respond to harmful user requests, even w… (see more)hen using harmless datasets, thus requiring additional safety measures. We challenge this belief through systematic testing, showing that poor optimization choices, rather than inherent trade-offs, often cause safety problems, measured as harmful responses to adversarial prompts. By properly selecting key training hyper-parameters, e.g., learning rate, batch size, and gradient steps, we reduce unsafe model responses from 16\% to approximately 5\%, as measured by keyword matching, while maintaining utility performance. Based on this observation, we propose a simple exponential moving average (EMA) momentum technique in parameter space that preserves safety performance by creating a stable optimization path and retains the original pre-trained model's safety properties. Our experiments on the Llama families across multiple datasets (Dolly, Alpaca, ORCA) demonstrate that safety problems during fine-tuning can largely be avoided without specialized interventions, outperforming existing approaches that require additional safety data while offering practical guidelines for maintaining both model performance and safety during adaptation.

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

openreview.net

Steering Large Language Model Activations in Sparse Spaces

Reza Bayat

Ali Rahimi-Kalahroudi

Mohammad Pezeshki

Sarath Chandar

Pascal Vincent

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

doi.org

openreview.net

Training Plug-and-Play Knowledge Modules with Deep Context Distillation

Lucas Caccia

Alan Ansell

Edoardo Ponti

Ivan Vulić

Alessandro Sordoni

Dynamically integrating new or rapidly evolving information after (Large) Language Model pre-training remains challenging, particularly in l… (see more)ow-data scenarios or when dealing with private and specialized documents. In-context learning and retrieval-augmented generation (RAG) face limitations, including their high inference costs and their inability to capture global document information. In this paper, we propose a way of modularizing knowledge by training document-level Knowledge Modules (KMs). KMs are lightweight components implemented as parameter-efficient LoRA modules, which are trained to store information about new documents and can be easily plugged into models on demand. We show that next-token prediction performs poorly as the training objective for KMs. We instead propose Deep Context Distillation: we learn KMs parameters such as to simulate hidden states and logits of a teacher that takes the document in context. Our method outperforms standard next-token prediction and pre-instruction training techniques, across two datasets. Finally, we highlight synergies between KMs and retrieval-augmented generation.

2025-07-07

colmweb.org/COLM/2025/Conference (accepted)

openreview.net

RAT: Bridging RNN Efficiency and Attention Accuracy via Chunk-based Sequence Modeling

Xiuying Wei

Anunay Yadav

Razvan Pascanu

Caglar Gulcehre

Transformers have become the cornerstone of modern large-scale language models, but their reliance on softmax attention poses a computationa… (see more)l bottleneck at both training and inference. Recurrent models offer high efficiency, but compressing the full sequence into a fixed-size and holistic representation suffers from memory degradation in long contexts and limits fine-grained retrieval. To address this, we propose RAT, an intermediate design that bridges the efficiency of RNNs and capacity of attention. RAT partitions the input into chunks, applies recurrence within each chunk for local dependencies, and softmax-based attention across chunks for long-range interactions. This design mitigates memory degradation and enables direct access to distant tokens, while retaining computational efficiency. Empirically, with a chunk size of 16, the RAT block achieves a 7x improvement in training speed with 100K token sequences and 9x in generation at the 4K position, while maintaining similar performance compared to standard attention. We demonstrate this by training 1.3B parameter models from scratch and performing large-scale evaluations, including short- and long-context benchmarks, as well as supervised fine-tuning~(SFT). We further propose a hybrid architecture that interleaves RAT with local attention. By combining efficient long-range modeling with strong local interactions, this hybrid design not only improves inference speed and reduces cache memory usage, but also consistently enhances performance and shows the overall best results. Code is available at https://github.com/CLAIRE-Labo/RAT.

2025-07-06

ArXiv (preprint)

arxiv.org

DOLPHIN advances single-cell transcriptomics beyond gene level by leveraging exon and junction reads

Kailu Song

Yumin Zheng

Bowen Zhao

David H. Eidelman

Jian Tang

Jun Ding

2025-07-04

Nature Communications (published)

doi.org

Extracting and Following Paths for Robust Relational Reasoning with Large Language Models

Ge Zhang

Mohammad Alomrani

Hongjian Gu

Jiaming Zhou

Yaochen Hu

B. Wang

Qun Liu

Mark Coates

Yingxue Zhang

Jianye Hao

2025-07-04

KDD.org/2025/Workshop/SKnow-LLM (oral)

openreview.net

Assemblies, synapse clustering, and network topology interact with plasticity to explain structure-function relationships of the cortical connectome

András Ecker

Daniela Egas Santander

Marwan Abdellah

Jorge Blanco Alonso

Sirio Bolaños-Puchet

Giuseppe Chindemi

Dhuruva Priyan Gowri Mariyappan

James B. Isbister

James King

Pramod Kumbhar

Ioannis Magkanaris

Eilif Benjamin Muller

Michael W. Reimann

Synaptic plasticity underlies the brain’s ability to learn and adapt. While experiments in brain slices have revealed mechanisms and proto… (see more)cols for the induction of plasticity between pairs of neurons, how these synaptic changes are coordinated in biological neuronal networks to ensure the emergence of learning remains poorly understood. Simulation and modeling have emerged as important tools to study learning in plastic networks, but have yet to achieve a scale that incorporates realistic network structure, active dendrites, and multi-synapse interactions, key determinants of synaptic plasticity. To rise to this challenge, we endowed an existing large-scale cortical network model, incorporating data-constrained dendritic processing and multi-synaptic connections, with a calcium-based model of functional plasticity that captures the diversity of excitatory connections extrapolated to in vivo-like conditions. This allowed us to study how dendrites and network structure interact with plasticity to shape stimulus representations at the microcircuit level. In our exploratory simulations, plasticity acted sparsely and specifically, firing rates and weight distributions remained stable without additional homeostatic mechanisms. At the circuit level, we found plasticity was driven by co-firing stimulus-evoked functional assemblies, spatial clustering of synapses on dendrites, and the topology of the network connectivity. As a result of the plastic changes, the network became more reliable with more stimulus-specific responses. We confirmed our testable predictions in the MICrONS datasets, an openly available electron microscopic reconstruction of a large volume of cortical tissue. Our results quantify at a large scale how the dendritic architecture and higher-order structure of cortical microcircuits play a central role in functional plasticity and provide a foundation for elucidating their role in learning.

2025-07-03

eLife (published)

doi.org

Toward whole-genome inference of polygenic scores with fast and memory-efficient algorithms.

Shadi Zabad

Chirayu Anant Haryan

Simon Gravel

Sanchit Misra

Yue Li

2025-07-03

American Journal of Human Genetics (published)

doi.org