United States

Supervised fine-tuning (SFT) is fundamental to adapting large language models, yet training on complete datasets incurs prohibitive costs with diminishing returns. Existing data selection methods suffer from severe domain specificity: techniques optimized for general instruction-following fail on reasoning tasks, and vice versa. We observe that measuring contrastive entropy between base models and minimally instruction-tuned calibrated models reveals a pattern—samples with the lowest contrastive entropy consistently yield optimal performance across domains, yet this principle manifests domain-adaptively: reasoning tasks favor entropy increase (cognitive expansion), while general tasks favor entropy decrease (cognitive compression). We introduce InstructDiff, a unified framework that operationalizes contrastive entropy as a domain-adaptive selection criterion through warmup calibration, bi-directional NLL filtering, and entropy-based ranking. Extensive experiments show that InstructDiff achieves 17\% relative improvement over full data training on mathematical reasoning and 52% for general instruction-following, outperforming prior baselines while using only 10% of the data.

ACL 2026

InstructDiff: Domain-Adaptive Data Selection via Contrastive Entropy for Efficient LLM Fine-Tuning

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. <br>
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) <br><br><br>
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>
<br><br><br><br>

To access this event page, you need to log in with the **email address you registered with**. <br>Access credentials will be sent to your email from Underline -  subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

Registration Required

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

poster

Large Language Models (LLMs), despite its impressive capabilities across domains, have been shown to be vulnerable to backdoor attacks. Prior backdoor strategies predominantly operate at the token level, where an injected trigger causes the model to generate a specific target word, choice, or class (depending on the task). Recent advances, however, exploit the long-form reasoning tendencies of modern LLMs to conduct reasoning-level backdoors: once triggered, the victim model inserts one or more malicious reasoning steps into its chain-of-thought (CoT). These attacks are substantially harder to detect, as the backdoored answer remains plausible and consistent with the poisoned reasoning trajectory. Yet, defenses tailored to this type of backdoor remain largely unexplored. To bridge this gap, we propose Critical-CoT, a novel defense mechanism that conducts a two-stage fine-tuning (FT) on LLMs to develop critical thinking behaviors, enabling them to automatically identify potential backdoors and refuse to generate malicious reasoning steps. Extensive experiments across multiple LLMs and datasets demonstrate that Critical-CoT provides strong robustness against both in-context learning-based and FT-based backdoor attacks. Notably, Critical-CoT exhibits strong cross-domain and cross-task generalization. Our code is available at \url{hthttps://github.com/tuanvu171/Critical-CoT}.

Critical-CoT: A Robust Defense Framework against Reasoning-Level Backdoor Attacks in Large Language Models

LLMs are highly sensitive to prompt design, but handcrafting effective prompts is difficult and often requires intricate crafting of few-shot examples. We propose a fast automatic prompt construction algorithm that augments human instructions by generating a small set of few shot examples. Our method iteratively replaces/drops/keeps few-shot examples using Monte Carlo Shapley estimation of example utility. For faster execution, we use aggressive subsampling and a replay buffer for faster evaluations. Our method can be run using different compute time budgets. Under a limited budget, it outperforms prior automatic prompting methods on text simplification and mathematical reasoning (GSM8K, DeepMath, Math500), while achieving second-best results on classification and summarization and third-best on MedQA. With an extended, yet still modest budget, PIAST sets a new state of the art among automatic prompting methods on classification, simplification, GSM8K, DeepMath, and Math500. Overall, our results suggest that optimizing in-context examples, rather than exhaustively searching over instruction rewrites is the dominant lever for fast and data-efficient prompt engineering. We will release code and data upon acceptance.

PIAST: Rapid Prompting with In-context Augmentation for Scarce Training data

Web applications (web apps) have become a key arena for large language models (LLMs) to demonstrate their code generation capabilities and commercial potential. However, building a benchmark for LLM-generated web apps remains challenging due to the need for real-world user requirements, generalizable evaluation metrics without relying on ground-truth implementations or test cases, and interpretable evaluation results. To address these challenges, we introduce WebCoderBench, the first real-world-collected, generalizable, and interpretable benchmark for web app generation. WebCoderBench comprises 1,572 user requirements, covering diverse modalities and expression styles that reflect realistic user intentions. WebCoderBench provides 24 fine-grained evaluation metrics across 9 perspectives, combining the rule-based and LLM-as-a-judge paradigms for fully automated, objective, and general evaluation. Moreover, WebCoderBench adopts human-preference-aligned weights over metrics to yield interpretable overall scores. Experiments across 12 representative LLMs and 2 LLM-based agents show that there exists no dominant model across all evaluation metrics, offering an opportunity for LLM developers to optimize their models in a targeted manner for a more powerful version.

WebCoderBench: Benchmarking Web Application Generation with Comprehensive and Interpretable Evaluation Metrics

Tool-Integrated Reasoning (TIR) empowers large language models (LLMs) to tackle complex tasks by interleaving reasoning steps with external tool interactions. However, existing reinforcement learning methods typically rely on outcome- or trajectory-level rewards, assigning uniform advantages to all steps within a trajectory. This coarse-grained credit assignment fails to distinguish effective tool calls from redundant or erroneous ones, particularly in long-horizon multi-turn scenarios. To address this, we propose MatchTIR, a framework that introduces fine-grained supervision via bipartite matching-based turn-level reward assignment and dual-level advantage estimation. Specifically, we formulate credit assignment as a bipartite matching problem between predicted and ground-truth traces, utilizing two assignment strategies to derive dense turn-level rewards. Furthermore, to balance local step precision with global task success, we introduce a dual-level advantage estimation scheme that integrates turn-level and trajectory-level signals, assigning distinct advantage values to individual interaction turns. Extensive experiments on three benchmarks demonstrate the superiority of MatchTIR. Notably, our 4B model surpasses the majority of 8B competitors, particularly in long-horizon and multi-turn tasks. Our codes are available at https://anonymous.4open.science/r/MatchTIR.

MatchTIR: Fine-Grained Supervision for Tool-Integrated Reasoning via Bipartite Matching

Large-scale social simulators are essential for studying complex social patterns. Prior work explores hybrid methods to scale up simulations, combining large language models (LLM)-based agents with numerical agent-based models (ABM). However, this incurs high latency due to expensive memory retrieval and sequential ABM execution. To address this challenge, we propose GASim, a graph-accelerated hybrid multi-agent framework for large-scale social simulations. For core agents driven by LLM, GASim introduces Graph-Optimized Memory (GOM) to replace intensive LLM-based retrieval pipelines with lightweight propagation over a sparse memory graph. For the majority of ordinary agents, GASim employs Graph Message Passing (GMP), substituting sequential ABM execution with parallel updates by fine-grained feature aggregation and Graph Attention Network. We further introduce Entropy-Driven Grouping (EDG) that coordinates this hybrid partitioning, leveraging information entropy to dynamically identify emergent core agents situated in information-diverse neighborhoods. Extensive experiments show that GASim not only delivers a substantial 9.94× end-to-end speedup over the traditional hybrid framework but also consumes less than 20% of baseline tokens, significantly reducing costs while preserving strong alignment with real-world public opinion trends.

 GASim: A Graph-Accelerated Hybrid Framework for Social Simulation

Process Reward Models (PRMs), which assign fine-grained scores to intermediate reasoning steps within a solution trajectory, have emerged as a promising approach to enhance the reasoning quality of Large Language Models (LLMs).
However, most existing PRMs rely on a unidirectional left-to-right (L2R) evaluation scheme, which restricts their utilization of global context.
In light of this challenge, we propose a novel bidirectional evaluation paradigm, named $\mathbf{Bi}$directional $\mathbf{P}$rocess $\mathbf{R}$eward $\mathbf{M}$odel ($\mathbf{BiPRM}$).
BiPRM incorporates a parallel right-to-left (R2L) evaluation stream, implemented via prompt reversal, alongside the conventional L2R flow.
Then a gating mechanism is introduced to adaptively fuse the reward scores from both streams to yield a holistic quality assessment.
Remarkably, compared to the original PRM, BiPRM introduces only a 0.3\% parameter increase for the gating module, and the parallel execution of two streams incurs merely 5\% inference time latency. 
Our extensive empirical evaluations spanning diverse benchmarks, LLM backbones, PRM objectives and sampling policies demonstrate that BiPRM consistently surpasses unidirectional baselines, achieving an average relative gain of 10.6\% over 54 solution-level configurations and 37.7\% in 12 step-level error detection scenarios.
Generally, our results highlight the effectiveness, robustness and general applicability of BiPRM, offering a promising new direction for process-based reward modeling.

The Bidirectional Process Reward Model

As large language models (LLMs) improve, many applications are moving from a single LLM call to multi-agent systems. These systems often rely on either hand-designed or automatically optimized workflows with multiple verification and testing steps. While those extra steps can improve accuracy, they also increase latency and token costs. In practice, many queries do not need such heavy processing and can be handled well by a single strong agent.

To address this inefficiency, we propose LLM-as-Scheduler (LAS), a system that dynamically chooses the right workflow for each query. LAS uses a two-stage cascade: first, a lightweight gate quickly evaluates each agent’s output; then, an LLM-based scheduler uses query features and gate signals to make more detailed routing decisions. Experiments show that LAS cuts token usage by 43% and reduces end-to-end latency by more than 36%, while causing at most a 1.4 percentage-point drop in accuracy compared with a strong fixed workflow.

LLM-as-Scheduler: Agentic Workflow Dynamic Scheduling

Large Language Models (LLMs) have demonstrated impressive capabilities across a wide range of NLP tasks, but they remain fundamentally stateless, constrained by limited context windows that hinder long-horizon reasoning. Recent efforts to address this limitation often augment LLMs with an external memory bank, yet most existing pipelines are static and heuristic-driven, lacking a learned mechanism for deciding what to store, update, or retrieve. We present Memory-R1, a reinforcement learning (RL) framework that equips LLMs with the ability to actively manage and utilize external memory through two specialized agents: a Memory Manager that learns structured operations, including ADD, UPDATE, DELETE, and NOOP; and an Answer Agent that pre-selects and reasons over relevant entries. Both agents are fine-tuned with outcome-driven RL (PPO and GRPO), enabling adaptive memory management with minimal supervision. With only 152 training QA pairs, Memory-R1 outperforms strong baselines and generalizes across diverse question types, three benchmarks (LoCoMo, MSC, LongMemEval), and multiple model scales (3B–14B).

Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning

Recent advances in large language models (LLMs) have highlighted the effectiveness of chain-of-thought reasoning in symbolic domains such as mathematics and programming. However, our study shows that directly transferring such text-based reasoning paradigms to protein function understanding is ineffective: reinforcement learning mainly amplifies superficial keyword patterns while failing to introduce new biological knowledge, resulting in limited generalization. We argue that protein function prediction is a knowledge-intensive scientific task that fundamentally relies on external biological priors and computational tools rather than purely internal reasoning. To address this gap, we propose Protein Function Understanding Agent (PFUA), a tool-augmented protein reasoning agent that unifies problem decomposition, tool invocation, and grounded answer generation. Instead of relying on long unconstrained reasoning traces, PFUA integrates domain-specific tools to produce verifiable intermediate evidence. Experiments on four benchmarks demonstrate that PFUA consistently outperforms text-only reasoning models with an average performance improvement of 103%. We believe PFUA has the potential to become a standard paradigm for agentic reasoning in knowledge-intensive life science domains.

Interleaved Tool-Call Reasoning for Protein Function Understanding

Reliable failure detection and causal reasoning are critical in robotic manipulation, as their absence risks robot damage and endangers human safety.
Although recent Vision–Language Models (VLMs) are employed to attempt failure detection and causality reasoning, they typically make retrospective assessment only after task completion, and their reasoning accuracy is often limited.
To address these issues, we introduce RoboFailRing, which enables timely failure detection during task execution and enhances the reasoning accuracy of VLMs.
It achieves rapid failure detection by retrieving a pre-constructed failure memory and returning a similarity-based decision.
In addition, by providing grounded failure report to VLMs, it improves the accuracy of their reasoning about the failure causes and repair strategies.
We evaluate RoboFailRing on two large-scale simulated datasets comprising over 6,000 failure trajectories and covering 81 distinct manipulation tasks.
The results show that the average success rate of out-of-distribution failure detection reaches 80%, while the mean detection time is cut to roughly 50% of the baseline.
Moreover, evaluations on real-world systems show an average 35% gain in VLM failure-reasoning accuracy.
We make our code publicly available at: https://github.com/DynamicPoet/RoboFailRing.

Downloads

Next from ACL 2026

Critical-CoT: A Robust Defense Framework against Reasoning-Level Backdoor Attacks in Large Language Models

Similar lecture

Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective