ICLR 2026 Papers

We introduce jqBench, a new benchmark for evaluating language models on JSON querying and transformation tasks, where the intent can be given specified using natural language and/or examples. Whereas jqBench is mainly aimed at using the `jq` tool, it can be used to evaluate other programming languages that query and/or transform JSON. Benchmarks are automatically created from two rich sources of data: Stack Overflow discussions (751 instances with instructions and examples, called jqStack) and the Spider dataset for SQL generation from natural language (893 instances with instructions and JSON Schema, called jqSpider). We describe and analyze the automated pipeline for benchmark creation, and perform extensive baseline experiments on different models to analyze the complexity and failure modes. Using implicit feedback, the best model (Claude Opus 4.1) scores 77% on the jqStack benchmarks and 81\% on the jqSpider benchmarks. Additionally, we show (1) that access to the documentation surprisingly does not help, (2) `jq` performs comparable to Python, and (3) that automatic feedback (and therefore examples) is crucial. Besides the final benchmarks, we release the intermediate artifacts from each generation step (including failed or invalid conversions) as well as an LLM-friendly version of the documentation, to facilitate further research on JSON querying and transformation.

Large Language Models (LLMs) have shown remarkable reasoning capabilities, while their practical applications are limited by severe factual hallucinations due to limitations in the timeliness, accuracy, and comprehensiveness of their parametric knowledge. Meanwhile, enhancing retrieval-augmented generation (RAG) with reasoning remains challenging due to ineffective task decomposition and redundant retrieval, which can introduce noise and degrade response quality. In this paper, we propose DeepRAG, a framework that models retrieval-augmented reasoning as a Markov Decision Process (MDP), enabling reasonable and adaptive retrieval. By iteratively decomposing queries, DeepRAG dynamically determines whether to retrieve external knowledge or rely on parametric reasoning at each step. Experiments show that DeepRAG improves retrieval efficiency and boosts answer accuracy by 25.41%, demonstrating its effectiveness in enhancing retrieval-augmented reasoning.

Large language models excel at generating individual functions or single files of code, yet generating complete repositories from scratch remains a fundamental challenge. This capability is key to building coherent software systems from high-level specifications and realizing the full potential of automated code generation. The process requires planning at two levels: deciding what features and modules to build (proposal stage) and defining their implementation details (implementation stage). Current approaches rely on natural language planning, which often produces unclear specifications, misaligned components, and brittle designs due to its inherent ambiguity and lack of structure. To address these limitations, we introduce the Repository Planning Graph (RPG), a structured representation that encodes capabilities, file structures, data flows, and functions in a unified graph. By replacing free-form natural language with an explicit blueprint, RPG enables consistent long-horizon planning for repository generation. Building on RPG, we develop ZeroRepo, a graph-driven framework that operates in three stages: proposal-level planning, implementation-level construction, and graph-guided code generation with test validation To evaluate, we construct RepoCraft, a benchmark of six real-world projects with 1,052 tasks. On RepoCraft, ZeroRepo produces nearly 36K Code Lines and 445K Code Tokens, on average 3.9× larger than the strongest baseline (Claude Code), and 68× larger than others. It also achieves 81.5% coverage and 69.7% test accuracy, improving over Claude Code by 27.3 and 35.8 points. Further analysis shows that RPG models complex dependencies, enables more sophisticated planning through near-linear scaling, and improves agent understanding of repositories, thus accelerating localization. Our data and code are available at https://github.com/microsoft/RPG-ZeroRepo.

Code language models (LMs) represent valuable intellectual property (IP) as their training involves immense investments, including large-scale code corpora, proprietary annotations, extensive computational resources, and specialized designs. Hence the threat of model IP infringements such as unauthorized redistribution or model theft has become increasingly concerning. While neural network watermarking has been widely studied as a measure to support model ownership verification, watermarking code LMs is particularly challenging due to the seemingly conflicting requirements of code generation: adhering to strict syntactic rules and semantic consistency while allowing flexible changes to embed watermarks, keeping high fidelity of the generated content while being robust to extraction attacks, etc. To resolve the issues, we propose CodeGenGuard, a watermarking framework for code LMs. CodeGenGuard leverages semantic-preserving transformations (SPTs) to encode the watermark and incorporates a dead-code-based data augmentation pipeline to diversify SPT patterns. To improve robustness, we incorporate an efficient dual-LoRA shadow training scheme and an optimizable trigger prompt that learns to extract watermark from both the watermarked and the shadow models. As most SPTs take place in specific contexts, we implant auxiliary prompts during verification to encourage the generation of the context, further enhancing the detection rate. Evaluation results on representative code generation models demonstrate that CodeGenGuard achieves superior watermarking performance to the state-of-the-art.

Neural vocoders have recently advanced waveform generation, yielding natural and expressive audio. Among these approaches, iSTFT-based vocoders have recently gained attention. They predict a complex-valued spectrogram and then synthesize the waveform via iSTFT, thereby avoiding learned upsampling stages that can increase computational cost. However, current approaches use real-valued networks that process the real and imaginary parts independently. This separation limits their ability to capture the inherent structure of complex spectrograms. We present ComVo, a Complex-valued neural Vocoder whose generator and discriminator use native complex arithmetic. This enables an adversarial training framework that provides structured feedback in complex-valued representations. To guide phase transformations in a structured manner, we introduce phase quantization, which discretizes phase values and regularizes the training process. Finally, we propose a block-matrix computation scheme to improve training efficiency by reducing redundant operations. Experiments demonstrate that ComVo achieves higher synthesis quality than comparable real-valued baselines, and that its block-matrix scheme reduces training time by 25%. Audio samples and code are available at https://hs-oh-prml.github.io/ComVo/.

Sketching represents humanity's most intuitive form of visual expression -- a universal language that transcends barriers. Although recent diffusion models integrate sketches with text, they often regard the complete sketch merely as a static visual constraint, neglecting the human preference information inherently conveyed during the dynamic sketching process.This oversight leads to images that, despite technical adherence to sketches, fail to align with human aesthetic expectations. Our framework, SketchEvo, harnesses the dynamic evolution of sketches by capturing the progression from initial strokes to completed drawing. Current preference alignment techniques struggle with sketch-guided generation because the dual constraints of text and sketch create insufficiently different latent samples when using noise perturbations alone. SketchEvo addresses this through two complementary innovations: first, by leveraging sketches at different completion stages to create meaningfully divergent samples for effective aesthetic learning during training; second, through a sequence-guided rollback mechanism that applies these learned preferences during inference by balancing textual semantics with structural guidance. Extensive experiments demonstrate that these complementary approaches enable SketchEvo to deliver improved aesthetic quality while maintaining sketch fidelity, successfully generalizing to incomplete and abstract sketches throughout the drawing process.

Human–object–scene interactions (HOSI) generation has broad applications in embodied AI, simulation, and animation. Unlike human–object interaction (HOI) and human–scene interaction (HSI), HOSI generation requires reasoning over dynamic object–scene changes, yet suffers from limited annotated data. To address these issues, we propose a coarse‑to‑fine instruction‑conditioned interaction generation framework that is explicitly aligned with the iterative denoising process of a consistency model. In particular, we adopt a dynamic perception strategy that leverages trajectories from the preceding refinement to update scene context and condition subsequent refinement at each denoising step of consistency model, yielding consistent interactions. To further reduce physical artifacts, we introduce a bump‑aware guidance that mitigates collisions and penetrations during sampling without requiring fine‑grained scene geometry, enabling real‑time generation. To overcome data scarcity, we design a hybrid training startegy that synthesizes pseudo‑HOSI samples by injecting voxelized scene occupancy into HOI datasets and jointly trains with high‑fidelity HSI data, allowing interaction learning while preserving realistic scene awareness. Extensive experiments demonstrate that our method achieves state‑of‑the‑art performance in both HOSI and HOI generation, and strong generalization to unseen scenes. Code and datasets will be released upon acceptance.

Retrieval-Augmented Generation (RAG) mitigates hallucinations in Large Language Models (LLMs) by integrating external knowledge. However, conflicts between parametric knowledge and retrieved context pose challenges, particularly when retrieved information is unreliable or the model's internal knowledge is outdated. In such cases, LLMs struggle to determine whether to rely more on their own parameters or the conflicted context. To address this, we propose CK-PLUG, a plug-and-play method for controlling LLMs' reliance on parametric and contextual knowledge. We introduce a novel knowledge consistency metric, Confidence Gain, which detects knowledge conflicts by measuring entropy shifts in token probability distributions after context insertion. CK-PLUG then enables fine-grained control over knowledge preference by adjusting the probability distribution of tokens with negative confidence gain through a single tuning parameter. Experiments demonstrate CK-PLUG's ability to significantly regulate knowledge reliance in counterfactual RAG scenarios while maintaining generation fluency and knowledge accuracy. For instance, on LLaMA3-8B, memory recall (MR) of RAG response can be adjusted within a broad range (9.9%-71.9%), compared to the baseline of 42.1%. Moreover, CK-PLUG supports adaptive control based on the model's confidence in both internal and external knowledge, achieving consistent performance improvements across various general RAG tasks. Our code is available at: https://anonymous.4open.science/r/CK-PLUG-Ano-8E62

Cover songs constitute a vital aspect of musical culture, preserving the core melody of an original composition while reinterpreting it to infuse novel emotional depth and thematic emphasis. Although prior research has explored the reinterpretation of instrumental music through melody-conditioned text-to-music models, the task of cover song generation remains largely unaddressed. In this work, we reformulate our cover song generation as a conditional generation, which simultaneously generates new vocals and accompaniment conditioned on the original vocal melody and text prompts. To this end, we present SongEcho, which leverages Instance-Adaptive Element-wise Linear Modulation (IA-EiLM), a framework that incorporates controllable generation by improving both conditioning injection mechanism and conditional representation. To enhance the conditioning injection mechanism, we extend Feature-wise Linear Modulation (FiLM) to an Element-wise Linear Modulation (EiLM), to facilitate precise temporal alignment in melody control. For conditional representations, we propose Instance-Adaptive Condition Refinement (IACR), which refines conditioning features by interacting with the hidden states of the generative model, yielding instance-adaptive conditioning. Additionally, to address the scarcity of large-scale, open-source full-song datasets, we construct Suno70k, a high-quality AI song dataset enriched with comprehensive annotations. Experimental results across multiple datasets demonstrate that our approach generates superior cover songs compared to existing methods, while requiring fewer than 30% of the trainable parameters. The code, dataset, and demos are available at [https://github.com/lsfhuihuiff/SongEcho_ICLR2026](https://github.com/lsfhuihuiff/SongEcho_ICLR2026).

Repeated Sampling (RS) is a simple inference-time algorithm that has been shown to improve model performance on complex tasks. Although it is an effective way of scaling inference time, it often struggles to generate diverse solution candidates, frequently relying on the same underlying approach to solve the problem and thus producing redundant samples. To address this limitation, we propose a new inference algorithm, GuidedSampling, which decouples the exploration and generation phases during inference, increasing diversity of generated candidate solutions. The exploration phase identifies multiple concepts that can be utilized to solve the problem, while the generation phase applies a specific concept to provide final solution candidates. We first define the theoretical bounds of GuidedSampling and then empirically demonstrate that it improves the performance of base model at pass@50 by on an average $\sim21.6$% across various benchmarks compared to RS. Furthermore, models trained on trajectories of GuidedSampling exhibit substantial performance improvements at pass@5 by on an average $\sim9.7$%, compared to models trained on traditional RS. Additionally, models trained with GuidedSampling increases the average number of concepts per instance ($1.67 \to 3.03$), yielding a diverse set of candidates than traditional RS.

Humor is a commonly used and high-level human language in daily life. However, humor generation is a challenging task for large language models (LLMs) in multi-modal contexts, but with many useful applications of funny caption generation for images, requiring visual understanding, humor reasoning, creative imagination, and so on. Existing LLM-based approaches rely on reasoning chains or self-improvement, which suffer from limited creativity and interpretability. To address these bottlenecks, we develop a novel LLM-based humor generation mechanism based on a fundamental humor theory, GTVH. To produce funny and script-opposite captions, we introduce a humor-theory-driven multi-role LLM collaboration framework augmented with humor retrieval (HOMER). The framework consists of three LLM-based roles: (1) conflicting-script extractor that grounds humor in key script oppositions, forming the basis of caption generation; (2) retrieval-augmented hierarchical imaginator that identifies key humor targets and expands the creative space through diverse associations structured as imagination trees; and (3) caption generator that produces funny and diverse captions conditioned on the obtained knowledge. Extensive experiments on two New Yorker Cartoon benchmarking datasets show that HOMER outperforms state-of-the-art baselines and powerful LLM reasoning strategies on multi-modal humor captioning.

Large Language Models (LLMs) show remarkable code generation capabilities but often produce imperfect code with subtle bugs. A critical bottleneck for improving code quality is the scarcity of high-quality test cases. Existing approaches, primarily based on Supervised Fine-Tuning (SFT) over static datasets, are limited in their ability to discover novel bugs and struggle with the fundamental trade-off between generating error-triggering inputs and maintaining correct expected outputs. To address these limitations, we reframe test case generation as an iterative, adversarial process. We introduce ATGEN (Adversarial Test Generator), a novel framework that trains a test case generator via Reinforcement Learning (RL) in an adversarial loop with an evolving code generator. Instead of learning from a fixed set of bugs, our test generator is dynamically trained to create "attacking" I/O pairs for buggy code that is itself being iteratively generated. This process is guided by a reward function that explicitly balances the dual objectives of maximizing the bug detection rate and maintaining high output accuracy. Extensive experiments show that ATGEN dramatically outperforms the state-of-the-art SFT-based approach, UTGen, improving IO Accuracy by nearly 40 absolute points (71.56% vs. 31.83%) and more than doubling the Attack Rate (34.02% vs. 16.24%). The adversarial curriculum is particularly effective for hard-to-detect bugs, achieving an attack rate more than double that of the strongest baseline. Furthermore, tests generated by ATGEN serve as a more effective filter in Best-of-N code generation, significantly closing the gap to the human expert upper bound. Our work establishes a new and more effective paradigm for automated test generation and debugging for LLMs.

Diffusion Transformers (DiTs) have emerged as the dominant architecture for visual generation, powering state-of-the-art image and video models. By representing images as patch tokens with positional encodings (PEs), DiTs combine Transformer scalability with spatial and temporal inductive biases. In this work, we revisit how DiTs organize visual content and discover that patch tokens exhibit a surprising degree of independence: even when PEs are perturbed, DiTs still produce globally coherent outputs, indicating that spatial coherence is primarily governed by PEs. Motivated by this finding, we introduce the Positional Encoding Field (PE-Field), which extends positional encodings from the 2D plane to a structured 3D field. PE-Field incorporates depth-aware encodings for volumetric reasoning and hierarchical encodings for fine-grained sub-patch control, enabling DiTs to model geometry directly in 3D space. Our PE-Field–augmented DiT achieves state-of-the-art performance on single-image novel view synthesis and generalizes to controllable spatial image editing.

Probabilistic programming offers a powerful framework for modeling uncertainty, yet statistical model discovery in this domain entails navigating an immense search space under strict domain‐specific constraints. When small language models are tasked with generating probabilistic programs, they frequently produce outputs that suffer from both syntactic, and semantic errors, such as flawed inference constructs. Motivated by probabilistic programmers’ domain expertise and debugging strategies, we introduce RefineStat, a language model–driven framework that enforces semantic constraints ensuring synthesized programs contain valid distributions, well‐formed parameters, and then applies diagnostic‐aware refinement by resampling prior or likelihood components whenever reliability checks fail. We evaluate RefineStat on multiple probabilistic-programming code-generation tasks using smaller language models (SLMs) and find that it produces programs that are both syntactically sound and statistically reliable, often matching or surpassing those from closed-source large language models (e.g., OpenAI o3).

Visual autoregressive (VAR) models generate images through next-scale prediction, naturally achieving coarse-to-fine, fast, high-fidelity synthesis mirroring human perception. In practice, this hierarchy can drift at inference time, as limited capacity and accumulated error cause the model to deviate from its coarse-to-fine nature. We revisit this limitation from an information-theoretic perspective and deduce that ensuring each scale to contribute high-frequency content not explained by earlier scales mitigates the train–inference discrepancy. With this insight, we propose Scaled Spatial Guidance (SSG), a training-free, inference-time guidance that steers generation toward the intended hierarchy while maintaining global coherence. SSG emphasizes target high-frequency signals, defined as the semantic residual, isolated from a coarser prior. To obtain this prior, we leverage a principled frequency-domain procedure, Discrete Spatial Enhancement (DSE), devised to sharpen and better isolate the semantic residual through frequency-aware construction. SSG applies broadly across VAR models leveraging discrete visual tokens, regardless of tokenization design or conditioning modality. Experiments demonstrate SSG yields consistent gains in fidelity and diversity while preserving low latency, revealing untapped efficiency in coarse-to-fine image generation.

We propose Sci2Pol-Bench and Sci2Pol-Corpus, the first benchmark and training dataset for evaluating and fine-tuning large language models (LLMs) on policy brief generation from a scientific paper. We build Sci2Pol-Bench on a five-stage taxonomy to mirror the human writing process: (i) Autocompletion, (ii) Understanding, (iii) Summarization, (iv) Generation, and (v) Verification. It features 18 tasks in multiple-choice and open-ended formats. Specifically, for the Generation stage, we show that BERTScore and ROUGE scores fail to capture the quality of brief writing, and introduce a new LLM-based evaluation metric aligned with expert judgement. Using this benchmark, we evaluate 13 leading open-source and commercial LLMs to uncover key limitations. To improve LLM performance on brief writing, we curate the Sci2Pol-Corpus for fine-tuning. We start by linking each cited scientific paper to its corresponding policy document, drawn from 5.6 million policy records. This produces 140,000 candidate pairs. We then employ an LLM-as-a-judge to filter high-quality examples, followed by in-context polishing using three expert-written samples as references. This process yields a final set of 639 new pairs. Finally, we fine-tune three models on Sci2Pol-Corpus: LLaMA-3.1-8B, Gemma-12B, and Gemma-27B. Fine-tuning leads to consistent performance improvements across Sci2Pol-Bench. Notably, after fine-tuning, Gemma-27B surpasses the much larger GPT-4o and DeepSeek-V3 (671B). These demonstrate the effectiveness of our corpus in bridging the gap between science and policy.

Synthetic time series are vital for data augmentation, stress testing, and prototyping in quantitative finance. Yet in cryptocurrency markets, characterized by 24/7 trading, extreme volatility, and rapid regime shifts, existing Time Series Generation (TSG) methods and benchmarks often fall short, jeopardizing practical utility. Most prior work targets non-financial or traditional financial domains, focuses narrowly on classification and forecasting while neglecting crypto-specific complexities, and lacks critical financial evaluations, particularly for trading applications. To bridge these gaps, we introduce \textbf{CTBench}, the first \textbf{C}ryptocurrency \textbf{T}ime series generation \textbf{Bench}mark. It curates an open-source dataset of 452 tokens and evaluates models across 13 metrics spanning forecasting accuracy, rank fidelity, trading performance, risk assessment, and computational efficiency. A key innovation is a dual-task evaluation framework: the Predictive Utility measures how well synthetic data preserves temporal and cross-sectional patterns for forecasting, while the Statistical Arbitrage assesses whether reconstructed series support mean-reverting signals for trading. We systematically benchmark eight state-of-the-art models from five TSG families across four market regimes, revealing trade-offs between statistical quality and real-world profitability. Notably, CTBench provides ranking analysis and practical guidance for deploying TSG models in crypto analytics and trading applications. The source code is available at \url{https://anonymous.4open.science/r/CTBench-F5A3/}.

The growing volume of academic papers has made it increasingly difficult for researchers to efficiently extract key information. While large language models (LLMs) based agents are capable of automating question answering (QA) workflows for scientific papers, there still lacks a comprehensive and realistic benchmark to evaluate their capabilities. Moreover, training an interactive agent for this task is hindered by the shortage of high-quality interaction trajectories. In this work, we propose M4PQA, a human-annotated comprehensive paper QA dataset in the field of artificial intelligence, with 13,948 papers and 1,246 questions, that encompasses multi-task, multi-modal and instance-level evaluation. Furthermore, we propose ExTrActor, an automated framework for instruction data synthesis. With three LLM-based agents, ExTrActor can perform example generation and trajectory collection without human intervention. Evaluations of multiple open-source and proprietary models show that most models underperform on M4PQA, demonstrating its quality. Extensive experiments confirm that ExTrActor consistently improves the multi-turn tool-use capability of small models, enabling them to achieve performance comparable to larger ones.

Code evaluation and reinforcement learning rely critically on test cases. However, collecting golden test cases is hard and expensive, motivating the use of LLMs for automatic test case generation. This, in turn, raises a pivotal challenge: how can we rigorously evaluate the quality of the generated test cases? Existing benchmarks often evaluate the exclusion ratio on large, unstructured collections of wrong codes, leading to high computational costs and severe score inflation. Furthermore, they inadvertently reward generators that detect common, trivial bugs, while failing to penalize their inability to identify rare yet critical faults. In this work, we connect two fundamental questions: (1) What is the minimal set of wrong codes sufficient to represent the entire error space? and (2) What is the minimal set of test cases needed to distinguish them? We introduce a novel framework that formalizes benchmark construction as finding an optimal diagnostic basis in a binary code-test matrix, where rows represent wrong codes and columns represent test case results. The rank of this matrix plays a dual role. It specifies the minimal number of independent error patterns, which determines the size of wrong codes. It also provides a tight upper bound on the number of test cases required for complete fault coverage. Our objective is to identify a basis of size equal to the matrix rank that maximizes internal diversity, which is defined as the average pairwise Jaccard similarity of the codes' failure signatures (i.e., the matrix rows). To tackle this NP-hard problem, we propose WrongSelect, an efficient approximation algorithm combining pre-filtering and random-restart local search to select maximally diverse wrong codes. Applying this framework to millions of competitive programming submissions, we construct TC-Bench, a compact, diverse, and inflation-resistant benchmark. Extensive experiments show that even the most advanced test case generation methods achieve only ~60\% exclusion rates on TC-Bench, exposing a significant gap in their diagnostic power and highlighting substantial room for future improvement.