United States

Large language models (LLMs) have shown great performance on complex reasoning tasks but often require generating long intermediate thoughts before reaching a final answer. During generation, LLMs rely on a key-value (KV) cache for autoregressive decoding. However, the memory footprint of the KV cache grows with output length. Prior work on KV cache optimization mostly focus on compressing the long input context, while retaining the full KV cache for decoding. For tasks requiring long output generation, this leads to increased computational and memory costs. In this paper, we introduce ZoomR, a novel approach that enables LLMs to adaptively compress verbose reasoning thoughts into summaries and uses a dynamic KV cache selection policy that leverages these summaries while also strategically &quot;zooming in&quot; on fine-grained details. By using summary keys as a coarse-grained index during decoding, ZoomR uses the query to retrieve details for only the most important thoughts. This hierarchical strategy significantly reduces memory usage by avoiding full-cache attention at each step. Experiments across math and reasoning tasks show that our approach achieves competitive performance compared to baselines, while reducing inference memory requirements by more than $4 \times$. These results demonstrate that a multi-granularity KV selection enables more memory efficient decoding, especially for long output generation.

ACL 2026

ZoomR: Memory Efficient Reasoning through Multi-Granularity Key Value Retrieval

### Welcome to 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics

Welcome to the 2025 meeting of the Nations of the Americas Chapter of the Association for Computational Linguistics! I am proud to help organize the first NAACL conference to carry the new name of our organization, one that emphasizes inclusion for all of the Americas. I am also pleased to welcome you to Albuquerque, New Mexico, a state whose unique blend of cultural influences will make for an excellent backdrop for NAACL 2025, especially with this year’s special theme on NLP in a Multicultural World. 
**[Continue reading...](https://drive.google.com/file/d/1jX-qGhqVSZZCIrAnJaz798pflu5Irrdn/view?usp=sharing)**

*- Colin Cherry, Google, NAACL 2025 General Chair* 

[![](https://assets.underline.io/markdown_image/1/image/b087f8a4dc5816d6e1a6514e59c59ac3.png)](https://drive.google.com/file/d/1T96GzPqObXrMln2BMByCSXTSizjTg69P/view?usp=sharing)

You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Please log in to explore this event.

To access NAACL 2025 event page you are required to register. Please follow [**this link**](https://2025.naacl.org/registration/registration/) to register. Access will depend on your registration type.

Please register!

NAACL 2025

Welcome to 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics

Large language models (LLMs) have shown great performance on complex reasoning tasks but often require generating long intermediate thoughts before reaching a final answer. During generation, LLMs rely on a key-value (KV) cache for autoregressive decoding. However, the memory footprint of the KV cache grows with output length. Prior work on KV cache optimization mostly focus on compressing the long input context, while retaining the full KV cache for decoding. For tasks requiring long output generation, this leads to increased computational and memory costs. In this paper, we introduce ZoomR, a novel approach that enables LLMs to adaptively compress verbose reasoning thoughts into summaries and uses a dynamic KV cache selection policy that leverages these summaries while also strategically "zooming in" on fine-grained details. By using summary keys as a coarse-grained index during decoding, ZoomR uses the query to retrieve details for only the most important thoughts. This hierarchical strategy significantly reduces memory usage by avoiding full-cache attention at each step. Experiments across math and reasoning tasks show that our approach achieves competitive performance compared to baselines, while reducing inference memory requirements by more than $4 \times$. These results demonstrate that a multi-granularity KV selection enables more memory efficient decoding, especially for long output generation.

technical paper

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. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<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>

To access this event page, you need to log in with the **email address you registered with**. 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.

**Online Registration Form**: https://acl.swoogo.com/acl2026

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.

As Large Language Models (LLMs) can now process extremely long contexts, efficient inference over these extended inputs has become increasingly important, especially for emerging applications like LLM agents that highly depend on this capability. Speculative decoding (SD) offers a promising lossless acceleration technique compared to lossy alternatives such as quantization and model cascades. However, most state-of-the-art SD methods are trained on short texts (typically fewer than 4k tokens), making them unsuitable for long-context scenarios. Specifically, adapting these methods to long contexts presents three key challenges: (1) the excessive memory demands posed by draft models due to large Key-Value (KV) cache; (2) performance degradation resulting from the mismatch between short-context training and long-context inference; and (3) inefficiencies in tree attention mechanisms when managing long token sequences. This work introduces LongSpec, a framework that addresses these challenges through three core innovations: a memory-efficient draft model with a constant-sized KV cache; novel position indices that mitigate the training–inference mismatch; and an attention aggregation strategy that combines fast prefix computation with standard tree attention to enable efficient decoding. Experimental results confirm the effectiveness of LongSpec, achieving up to a 3.26x speedup over strong Flash Attention baselines across five long-context understanding datasets, as well as a 2.34x reduction in wall-clock time on four math reasoning tasks with the QwQ model, demonstrating significant latency improvements for long-context applications.

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification

Large language models (LLMs) exhibit strong general reasoning, yet the community lacks controllable, scalable, and verifiable tools to analyze and improve these abilities. We present SATQuest, a verifier that generates diverse SAT-based reasoning tasks directly from Conjunctive Normal Form (CNF) instances and checks answers objectively with PySAT. SATQuest factorizes evaluation along three orthogonal dimensions—instance, problem type, and question format—enabling fine-grained, multi-dimensional analysis and reinforcement fine-tuning. Randomized CNF generation mitigates memorization and supports reproducible experiments. Using SATQuest, we benchmark a range of open- and closed-weight LLMs and uncover persistent gaps in logical reasoning, particularly on higher-complexity tasks and in transfer beyond familiar mathematical notation to machine or narrative formats. We further show that reinforcement fine-tuning with SATQuest rewards substantially boosts targeted performance and generalizes to larger instances, while cross-format robustness remains challenging. Collectively, SATQuest provides verifier-backed infrastructure for controlled, scalable, and reproducible empirical research on LLM logical reasoning and its training.

SATQuest: A Verifier for Logical Reasoning Evaluation and Reinforcement Fine-Tuning of LLMs

Large vision–language models (VLMs) commonly process images at native or high resolution to remain effective across tasks. This inflates visual tokens to 97-99% of total tokens, resulting in high compute and latency, even when low-resolution images would suffice. We introduce \emph{CARES}—a \textbf{C}ontext-\textbf{A}ware \textbf{R}esolution \textbf{S}elector, a lightweight preprocessing module that, given an image–query pair, predicts the \emph{minimal} sufficient input resolution. CARES uses a compact VLM (350M) to extract features and predict when a target pretrained VLM's response converges to its peak ability to answer correctly. Though trained as a discrete classifier over a set of optional resolutions, CARES interpolates continuous resolutions at inference for fine-grained control. Across five multimodal benchmarks spanning documents and natural images, as well as diverse target VLMs, CARES preserves task performance while reducing compute by up to 80%.

CARES: Context-Aware Resolution Selector for VLMs

Parameter-Efficient Fine-Tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA), have significantly reduced the number of trainable parameters needed in fine-tuning large language models (LLMs). The developments of LoRA-style adapters have considered two main directions: (1) enhancing model expressivity with high-rank adapters, and (2) aiming for further parameter reduction, as exemplified by vector-based methods. However, these approaches come with a trade-off, as achieving the expressivity of high-rank weight updates typically comes at the cost of sacrificing the extreme parameter efficiency offered by vector-based techniques. To address this issue, we propose a vector-based random Tensor network for high-Rank Adaptation (TeRA), a novel PEFT method that achieves high-rank weight updates while retaining the parameter efficiency of vector-based PEFT adapters. This is achieved by parametrizing the tensorized weight update matrix as a Tucker-like tensor network (TN), whereby large randomly initialized factors are frozen and shared across layers, while only small layer-specific scaling vectors, corresponding to diagonal entries of factor matrices, are trained. Comprehensive experiments demonstrate that TeRA matches or even outperforms existing high-rank adapters, while requiring as few trainable parameters as vector-based methods. Theoretical analysis and ablation studies validate the effectiveness of the proposed TeRA method. The code is available at https://github.com/guyuxuan9/TeRA.

TeRA: Vector-based Random Tensor Network for High-Rank Adaptation of Large Language Models

Large language models (LLMs) have demonstrated remarkable progress in reasoning, often through supervised fine-tuning (SFT). However, SFT is resource-intensive, relying on large curated datasets, rejection-sampled demonstrations, and uniform optimization across all tokens—even though only a fraction carry meaningful learning value. In this work, we explore a counterintuitive idea: can smaller language models (SLMs) teach larger language models (LLMs) by revealing high-value reasoning moments that reflect the latter's unique strength? We propose LightReasoner, a novel framework that leverages the behavioral divergence between a stronger expert model (LLM) and a weaker amateur model (SLM). LightReasoner operates in two stages: (1) a sampling stage that pinpoints critical reasoning moments and constructs supervision examples capturing the expert's advantage through expert–amateur contrast, and (2) a fine-tuning stage that aligns the expert model with these distilled examples, amplifying its reasoning strengths. Across seven benchmarks, LightReasoner improves accuracy by up to 28.1%, while reducing time consumption by 90%, sampled problems by 80%, and tuned token usage by 99%, all without relying on ground-truth labels. By turning weaker SLMs into effective teaching signals, LightReasoner offers a scalable and resource-efficient approach for advancing LLM reasoning.

LightReasoner: Can Small Language Models Teach Large Language Models Reasoning?

While Large Language Models (LLMs) excel at code generation by learning from vast code corpora, a fundamental semantic gap remains between their training on textual patterns and the goal of functional correctness, which is governed by formal execution semantics. Reinforcement Learning with Verifiable Rewards (RLVR) approaches attempt to bridge this gap using outcome rewards from executing test cases. However, solely relying on binary pass/fail signals is inefficient for establishing a well-aligned connection between the textual representation of code and its execution semantics, especially for subtle logical errors within the code. In this paper, we propose CODERL+, a novel approach that integrates execution semantics alignment into the RLVR training pipeline for code generation. CODERL+ enables the model to infer variable-level execution trajectory, providing a direct learning signal of execution semantics. CODERL+ can construct execution semantics alignment directly using existing on-policy rollouts and integrates seamlessly with various RL algorithms. Extensive experiments demonstrate that CODERL+ outperforms post-training baselines (including RLVR and Distillation), achieving a 4.6% average relative improvement in pass@1. CODERL+ generalizes effectively to other coding tasks, yielding 15.5% and 4.4% higher accuracy on code-reasoning and test-output-generation benchmarks, respectively. CODERL+ shows strong applicability across diverse RL algorithms and LLMs. Furthermore, probe analyses provide compelling evidence that CODERL+ strengthens the alignment between code’s textual representations and its underlying execution semantics.

CODERL+: Improving Code Generation via Reinforcement with Execution Semantics Alignment

While textual frequency has been validated as relevant to human cognition in reading speed, its relatedness to Large Language Models (LLMs) is seldom studied. We propose a novel research direction in terms of textual data frequency, which is an understudied topic. to the best of our knowledge. Our framework is composed of three units. First, this paper proposes \textbf{T}extual \textbf{F}requency \textbf{L}aw (TFL), which indicates that frequent textual data should be preferred for LLMs for both prompting and fine-tuning. Since many LLMs are closed-source in their training data, we propose using online resources to estimate the sentence-level frequency. We can then use an input paraphraser to paraphrase the input into a more frequent textual expression. Next, we propose \textbf{T}extual \textbf{F}requency \textbf{D}istillation (TFD) by querying LLMs to conduct story completion by further extending the sentences in the datasets, and the resulting corpora are used to adjust the initial estimation. Finally, we propose \textbf{C}urriculum \textbf{T}extual \textbf{F}requency \textbf{T}raining (CTFT) that fine-tunes LLMs in an increasing order of sentence-level frequency. Experiments are conducted on our curated dataset \textbf{T}extual \textbf{F}requency \textbf{P}aired \textbf{D}ataset (TFPD) on math reasoning and machine translation. Results indicate the effectiveness of our framework.

Adam's Law: Textual Frequency Law on Large Language Models

Medical calculators are fundamental to quantitative, evidence-based clinical practice. However, their real-world use is an adaptive, multi-stage process, requiring proactive EHR data acquisition, scenario-dependent calculator selection, and multi-step computation, whereas current benchmarks focus only on static single-step calculations with explicit instructions. To address these limitations, we introduce MedMCP-Calc, the first benchmark for evaluating LLMs in realistic medical calculator scenarios through Model Context Protocol (MCP) integration. MedMCP-Calc comprises 118 scenario tasks across 4 clinical domains, featuring fuzzy task descriptions mimicking natural queries, structured EHR database interaction, external reference retrieval, and process-level evaluation. Our evaluation of 23 leading models reveals critical limitations: even top performers like GPT-5 exhibit substantial gaps, including difficulty selecting appropriate calculators for end-to-end workflows given fuzzy queries, poor performance in iterative SQL-based database interactions, and marked reluctance to leverage external tools for numerical computation. Performance also varies considerably across clinical domains. Building on these findings, we develop CalcMate, a fine-tuned model incorporating scenario planning and tool augmentation, achieving state-of-the-art performance among open-source models.

MedMCP-Calc: Benchmarking LLMs for Realistic Medical Calculator Scenarios via MCP Integration

Large Vision–Language Models (LVLMs) achieve strong performance across multimodal tasks by integrating visual perception with language understanding. However, how vision information contributes to the model’s decoding process remains under-explored, as reflected in frequent hallucinations. Through a series of analyses, we found that (i) vision tokens provide meaningful visual information even when hallucinations occur, and (ii) their semantics are encoded in the textual space and become explicit under appropriate vocabulary constraints. Building on these observations, we propose ReVisiT, a simple training-free decoding method that guides text generation in LVLMs by Referencing Vision Tokens. Our approach leverages the semantic information embedded within vision tokens by projecting them into the text token distribution. Specifically, ReVisiT dynamically selects the most relevant vision token at each decoding step via context-aware constrained divergence minimization. Then, ReVisiT uses its constrained projection to refine the output distribution to better incorporate visual semantics. Across five benchmarks on recent LVLMs, ReVisiT consistently enhances visual grounding with minimal computational overhead, and achieves competitive or superior results to state-of-the-art decoding baselines while reducing computational cost by up to $2\times$.

Revisit What You See: Revealing Visual Semantics in Vision Tokens to Guide LVLM Decoding

For generative AI agents to partner effectively with human users, the ability to accurately predict human intent is critical. But this ability to collaborate remains limited by a critical deficit: an inability to model common ground. We present a referential communication experiment with a factorial design involving director-matcher pairs (human-human, human-AI, AI-human, and AI-AI) that interact with multiple turns in repeated rounds to match pictures of objects not associated with any obvious lexicalized labels. We show that LVLMs cannot 
interactively generate and resolve referring expressions in a way that enables smooth communication, a crucial skill that underlies human language use. We release our corpus of 356 dialogues (89 pairs over 4 rounds each) along with the online pipeline for data collection and the tools for analyzing accuracy, efficiency, and lexical overlap.

Downloads

Next from ACL 2026

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification

Similar lecture

MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ACL 2026

LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification

Similar lecture

MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference

Downloads