date
timestamp[ns]date 2023-05-05 00:00:00
2025-03-27 00:00:00
| arxiv_id
stringlengths 10
10
| title
stringlengths 8
177
| authors
sequencelengths 1
942
| github
stringlengths 0
116
| abstract
stringlengths 165
1.92k
|
|---|---|---|---|---|---|
2025-02-05T00:00:00 | 2501.19066 | Concept Steerers: Leveraging K-Sparse Autoencoders for Controllable Generations | [
"Dahye Kim",
"Deepti Ghadiyaram"
] | Despite the remarkable progress in text-to-image generative models, they are prone to adversarial attacks and inadvertently generate unsafe, unethical content. Existing approaches often rely on fine-tuning models to remove specific concepts, which is computationally expensive, lack scalability, and/or compromise generation quality. In this work, we propose a novel framework leveraging k-sparse autoencoders (k-SAEs) to enable efficient and interpretable concept manipulation in diffusion models. Specifically, we first identify interpretable monosemantic concepts in the latent space of text embeddings and leverage them to precisely steer the generation away or towards a given concept (e.g., nudity) or to introduce a new concept (e.g., photographic style). Through extensive experiments, we demonstrate that our approach is very simple, requires no retraining of the base model nor LoRA adapters, does not compromise the generation quality, and is robust to adversarial prompt manipulations. Our method yields an improvement of 20.01% in unsafe concept removal, is effective in style manipulation, and is sim5x faster than current state-of-the-art. |
|
2025-02-05T00:00:00 | 2502.00674 | Rethinking Mixture-of-Agents: Is Mixing Different Large Language Models Beneficial? | [
"Wenzhe Li",
"Yong Lin",
"Mengzhou Xia",
"Chi Jin"
] | Ensembling outputs from diverse sources is a straightforward yet effective approach to boost performance. Mixture-of-Agents (MoA) is one such popular ensemble method that aggregates outputs from multiple different Large Language Models (LLMs). This paper raises the question in the context of language models: is mixing different LLMs truly beneficial? We propose Self-MoA -- an ensemble method that aggregates outputs from only the single top-performing LLM. Our extensive experiments reveal that, surprisingly, Self-MoA outperforms standard MoA that mixes different LLMs in a large number of scenarios: Self-MoA achieves 6.6% improvement over MoA on the AlpacaEval 2.0 benchmark, and an average of 3.8% improvement across various benchmarks, including MMLU, CRUX, and MATH. Applying Self-MoA to one of the top-ranking models in AlpacaEval 2.0 directly achieves the new state-of-the-art performance on the leaderboard. To understand the effectiveness of Self-MoA, we systematically investigate the trade-off between diversity and quality of outputs under various MoA settings. We confirm that the MoA performance is rather sensitive to the quality, and mixing different LLMs often lowers the average quality of the models. To complement the study, we identify the scenarios where mixing different LLMs could be helpful. This paper further introduces a sequential version of Self-MoA, that is capable of aggregating a large number of LLM outputs on-the-fly over multiple rounds, and is as effective as aggregating all outputs at once. |
|
2025-02-05T00:00:00 | 2502.01839 | Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verification | [
"Eric Zhao",
"Pranjal Awasthi",
"Sreenivas Gollapudi"
] | Sampling-based search, a simple paradigm for utilizing test-time compute, involves generating multiple candidate responses and selecting the best one -- typically by verifying each response for correctness. In this paper, we study the scaling trends governing sampling-based search. Among our findings is that simply scaling up a minimalist implementation that uses only random sampling and direct self-verification results in sustained performance improvements that, for example, elevate the Gemini v1.5 Pro model's reasoning capabilities past that of o1-Preview on popular benchmarks. We partially attribute the scalability of sampling-based search to a phenomenon of implicit scaling, where sampling a larger pool of responses in turn improves verification accuracy. We further identify two useful principles for improving self-verification capabilities with test-time compute: (1) comparing across responses provides helpful signals about the locations of errors and hallucinations, and (2) different model output styles are useful for different contexts -- chains of thought are useful for reasoning but harder to verify. We also find that, though accurate verification can be elicited, frontier models demonstrate remarkably weak out-of-box verification capabilities and introduce a benchmark to measure progress on these deficiencies. |
|
2025-02-05T00:00:00 | 2502.02589 | COCONut-PanCap: Joint Panoptic Segmentation and Grounded Captions for Fine-Grained Understanding and Generation | [
"Xueqing Deng",
"Qihang Yu",
"Ali Athar",
"Chenglin Yang",
"Linjie Yang",
"Xiaojie Jin",
"Xiaohui Shen",
"Liang-Chieh Chen"
] | This paper introduces the COCONut-PanCap dataset, created to enhance panoptic segmentation and grounded image captioning. Building upon the COCO dataset with advanced COCONut panoptic masks, this dataset aims to overcome limitations in existing image-text datasets that often lack detailed, scene-comprehensive descriptions. The COCONut-PanCap dataset incorporates fine-grained, region-level captions grounded in panoptic segmentation masks, ensuring consistency and improving the detail of generated captions. Through human-edited, densely annotated descriptions, COCONut-PanCap supports improved training of vision-language models (VLMs) for image understanding and generative models for text-to-image tasks. Experimental results demonstrate that COCONut-PanCap significantly boosts performance across understanding and generation tasks, offering complementary benefits to large-scale datasets. This dataset sets a new benchmark for evaluating models on joint panoptic segmentation and grounded captioning tasks, addressing the need for high-quality, detailed image-text annotations in multi-modal learning. |
|
2025-02-05T00:00:00 | 2501.19389 | Federated Sketching LoRA: On-Device Collaborative Fine-Tuning of Large Language Models | [
"Wenzhi Fang",
"Dong-Jun Han",
"Liangqi Yuan",
"Seyyedali Hosseinalipour",
"Christopher G. Brinton"
] | Fine-tuning large language models (LLMs) on devices is attracting increasing interest. Recent works have fused low-rank adaptation (LoRA) techniques with federated fine-tuning to mitigate challenges associated with device model sizes and data scarcity. Still, the heterogeneity of computational resources remains a critical bottleneck: while higher-rank modules generally enhance performance, varying device capabilities constrain LoRA's feasible rank range. Existing approaches attempting to resolve this issue either lack analytical justification or impose additional computational overhead, leaving a wide gap for an efficient and theoretically-grounded solution. To address these challenges, we propose federated sketching LoRA (FSLoRA), which leverages a sketching mechanism to enable devices to selectively update submatrices of global LoRA modules maintained by the server. By adjusting the sketching ratios, which determine the ranks of the submatrices on the devices, FSLoRA flexibly adapts to device-specific communication and computational constraints. We provide a rigorous convergence analysis of FSLoRA that characterizes how the sketching ratios affect the convergence rate. Through comprehensive experiments on multiple datasets and LLM models, we demonstrate FSLoRA's superior performance compared to various baselines. |
|
2025-02-05T00:00:00 | 2502.00840 | Activation Approximations Can Incur Safety Vulnerabilities Even in Aligned LLMs: Comprehensive Analysis and Defense | [
"Jiawen Zhang",
"Kejia Chen",
"Lipeng He",
"Jian Lou",
"Dan Li",
"Zunlei Feng",
"Mingli Song",
"Jian Liu",
"Kui Ren",
"Xiaohu Yang"
] | Large Language Models (LLMs) have showcased remarkable capabilities across various domains. Accompanying the evolving capabilities and expanding deployment scenarios of LLMs, their deployment challenges escalate due to their sheer scale and the advanced yet complex activation designs prevalent in notable model series, such as Llama, Gemma, and Mistral. These challenges have become particularly pronounced in resource-constrained deployment scenarios, where mitigating inference efficiency bottlenecks is imperative. Among various recent efforts, activation approximation has emerged as a promising avenue for pursuing inference efficiency, sometimes considered indispensable in applications such as private inference. Despite achieving substantial speedups with minimal impact on utility, even appearing sound and practical for real-world deployment, the safety implications of activation approximations remain unclear. In this work, we fill this critical gap in LLM safety by conducting the first systematic safety evaluation of activation approximations. Our safety vetting spans seven sota techniques across three popular categories, revealing consistent safety degradation across ten safety-aligned LLMs. |
|
2025-02-05T00:00:00 | 2501.19054 | Text-to-CAD Generation Through Infusing Visual Feedback in Large Language Models | [
"Ruiyu Wang",
"Yu Yuan",
"Shizhao Sun",
"Jiang Bian"
] | Creating Computer-Aided Design (CAD) models requires significant expertise and effort. Text-to-CAD, which converts textual descriptions into CAD parametric sequences, is crucial in streamlining this process. Recent studies have utilized ground-truth parametric sequences, known as sequential signals, as supervision to achieve this goal. However, CAD models are inherently multimodal, comprising parametric sequences and corresponding rendered visual objects. Besides,the rendering process from parametric sequences to visual objects is many-to-one. Therefore, both sequential and visual signals are critical for effective training. In this work, we introduce CADFusion, a framework that uses Large Language Models (LLMs) as the backbone and alternates between two training stages: the sequential learning (SL) stage and the visual feedback (VF) stage. In the SL stage, we train LLMs using ground-truth parametric sequences, enabling the generation of logically coherent parametric sequences. In the VF stage, we reward parametric sequences that render into visually preferred objects and penalize those that do not, allowing LLMs to learn how rendered visual objects are perceived and evaluated. These two stages alternate throughout the training, ensuring balanced learning and preserving benefits of both signals. Experiments demonstrate that CADFusion significantly improves performance, both qualitatively and quantitatively. |
|
2025-02-06T00:00:00 | 2502.01506 | TwinMarket: A Scalable Behavioral and Social Simulation for Financial Markets | [
"Yuzhe Yang",
"Yifei Zhang",
"Minghao Wu",
"Kaidi Zhang",
"Yunmiao Zhang",
"Honghai Yu",
"Yan Hu",
"Benyou Wang"
] | The study of social emergence has long been a central focus in social science. Traditional modeling approaches, such as rule-based Agent-Based Models (ABMs), struggle to capture the diversity and complexity of human behavior, particularly the irrational factors emphasized in behavioral economics. Recently, large language model (LLM) agents have gained traction as simulation tools for modeling human behavior in social science and role-playing applications. Studies suggest that LLMs can account for cognitive biases, emotional fluctuations, and other non-rational influences, enabling more realistic simulations of socio-economic dynamics. In this work, we introduce TwinMarket, a novel multi-agent framework that leverages LLMs to simulate socio-economic systems. Specifically, we examine how individual behaviors, through interactions and feedback mechanisms, give rise to collective dynamics and emergent phenomena. Through experiments in a simulated stock market environment, we demonstrate how individual actions can trigger group behaviors, leading to emergent outcomes such as financial bubbles and recessions. Our approach provides valuable insights into the complex interplay between individual decision-making and collective socio-economic patterns. |
|
2025-02-06T00:00:00 | 2502.02339 | Boosting Multimodal Reasoning with MCTS-Automated Structured Thinking | [
"Jinyang Wu",
"Mingkuan Feng",
"Shuai Zhang",
"Ruihan Jin",
"Feihu Che",
"Zengqi Wen",
"Jianhua Tao"
] | Multimodal large language models (MLLMs) exhibit impressive capabilities but still face challenges in complex visual reasoning. While recent efforts attempt to enhance MLLMs' reasoning by incorporating OpenAI o1-like structured thinking through explicit search structures or teacher-guided distillation, they often struggle to balance performance and efficiency. A critical limitation is their heavy reliance on extensive data and search spaces, resulting in low-efficiency implicit insight extraction and data utilization. To address this, we propose AStar, an Automated Structured thinking paradigm for multimodal reasoning via Monte Carlo Tree Search (MCTS). AStar automatically derives high-level cognitive reasoning patterns from limited data using MCTS-powered hierarchical structures. Building on these explicit patterns, we design a unified reasoning framework that seamlessly integrates models' internal reasoning capabilities and external reasoning guidelines, enabling efficient inference with minimal tree iterations. This novel paradigm strikes a compelling balance between performance and efficiency. Extensive experiments demonstrate AStar's effectiveness, achieving superior accuracy (54.0%) on the MathVerse benchmark with a 7B backbone, surpassing GPT-4o (50.2%) while maintaining substantial data and computational efficiency. |
|
2025-02-06T00:00:00 | 2502.03373 | Demystifying Long Chain-of-Thought Reasoning in LLMs | [
"Edward Yeo",
"Yuxuan Tong",
"Morry Niu",
"Graham Neubig",
"Xiang Yue"
] | https://github.com/eddycmu/demystify-long-cot | Scaling inference compute enhances reasoning in large language models (LLMs), with long chains-of-thought (CoTs) enabling strategies like backtracking and error correction. Reinforcement learning (RL) has emerged as a crucial method for developing these capabilities, yet the conditions under which long CoTs emerge remain unclear, and RL training requires careful design choices. In this study, we systematically investigate the mechanics of long CoT reasoning, identifying the key factors that enable models to generate long CoT trajectories. Through extensive supervised fine-tuning (SFT) and RL experiments, we present four main findings: (1) While SFT is not strictly necessary, it simplifies training and improves efficiency; (2) Reasoning capabilities tend to emerge with increased training compute, but their development is not guaranteed, making reward shaping crucial for stabilizing CoT length growth; (3) Scaling verifiable reward signals is critical for RL. We find that leveraging noisy, web-extracted solutions with filtering mechanisms shows strong potential, particularly for out-of-distribution (OOD) tasks such as STEM reasoning; and (4) Core abilities like error correction are inherently present in base models, but incentivizing these skills effectively for complex tasks via RL demands significant compute, and measuring their emergence requires a nuanced approach. These insights provide practical guidance for optimizing training strategies to enhance long CoT reasoning in LLMs. Our code is available at: https://github.com/eddycmu/demystify-long-cot. |
2025-02-06T00:00:00 | 2502.01618 | A Probabilistic Inference Approach to Inference-Time Scaling of LLMs using Particle-Based Monte Carlo Methods | [
"Isha Puri",
"Shivchander Sudalairaj",
"Guangxuan Xu",
"Kai Xu",
"Akash Srivastava"
] | Large language models (LLMs) have achieved significant performance gains via scaling up model sizes and/or data. However, recent evidence suggests diminishing returns from such approaches, motivating scaling the computation spent at inference time. Existing inference-time scaling methods, usually with reward models, cast the task as a search problem, which tends to be vulnerable to reward hacking as a consequence of approximation errors in reward models. In this paper, we instead cast inference-time scaling as a probabilistic inference task and leverage sampling-based techniques to explore the typical set of the state distribution of a state-space model with an approximate likelihood, rather than optimize for its mode directly. We propose a novel inference-time scaling approach by adapting particle-based Monte Carlo methods to this task. Our empirical evaluation demonstrates that our methods have a 4-16x better scaling rate over our deterministic search counterparts on various challenging mathematical reasoning tasks. Using our approach, we show that Qwen2.5-Math-1.5B-Instruct can surpass GPT-4o accuracy in only 4 rollouts, while Qwen2.5-Math-7B-Instruct scales to o1 level accuracy in only 32 rollouts. Our work not only presents an effective method to inference-time scaling, but also connects the rich literature in probabilistic inference with inference-time scaling of LLMs to develop more robust algorithms in future work. Code and further information is available at https://probabilistic-inference-scaling.github.io. |
|
2025-02-06T00:00:00 | 2502.03387 | LIMO: Less is More for Reasoning | [
"Yixin Ye",
"Zhen Huang",
"Yang Xiao",
"Ethan Chern",
"Shijie Xia",
"Pengfei Liu"
] | https://github.com/GAIR-NLP/LIMO | We present a fundamental discovery that challenges our understanding of how complex reasoning emerges in large language models. While conventional wisdom suggests that sophisticated reasoning tasks demand extensive training data (>100,000 examples), we demonstrate that complex mathematical reasoning abilities can be effectively elicited with surprisingly few examples. Through comprehensive experiments, our proposed model LIMO demonstrates unprecedented performance in mathematical reasoning. With merely 817 curated training samples, LIMO achieves 57.1% accuracy on AIME and 94.8% on MATH, improving from previous SFT-based models' 6.5% and 59.2% respectively, while only using 1% of the training data required by previous approaches. LIMO demonstrates exceptional out-of-distribution generalization, achieving 40.5% absolute improvement across 10 diverse benchmarks, outperforming models trained on 100x more data, challenging the notion that SFT leads to memorization rather than generalization. Based on these results, we propose the Less-Is-More Reasoning Hypothesis (LIMO Hypothesis): In foundation models where domain knowledge has been comprehensively encoded during pre-training, sophisticated reasoning capabilities can emerge through minimal but precisely orchestrated demonstrations of cognitive processes. This hypothesis posits that the elicitation threshold for complex reasoning is determined by two key factors: (1) the completeness of the model's encoded knowledge foundation during pre-training, and (2) the effectiveness of post-training examples as "cognitive templates" that show the model how to utilize its knowledge base to solve complex reasoning tasks. To facilitate reproducibility and future research in data-efficient reasoning, we release LIMO as a comprehensive open-source suite at https://github.com/GAIR-NLP/LIMO. |
2025-02-06T00:00:00 | 2502.02737 | SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model | [
"Loubna Ben Allal",
"Anton Lozhkov",
"Elie Bakouch",
"Gabriel Martín Blázquez",
"Guilherme Penedo",
"Lewis Tunstall",
"Andrés Marafioti",
"Hynek Kydlíček",
"Agustín Piqueres Lajarín",
"Vaibhav Srivastav",
"Joshua Lochner",
"Caleb Fahlgren",
"Xuan-Son Nguyen",
"Clémentine Fourrier",
"Ben Burtenshaw",
"Hugo Larcher",
"Haojun Zhao",
"Cyril Zakka",
"Mathieu Morlon",
"Colin Raffel",
"Leandro von Werra",
"Thomas Wolf"
] | While large language models have facilitated breakthroughs in many applications of artificial intelligence, their inherent largeness makes them computationally expensive and challenging to deploy in resource-constrained settings. In this paper, we document the development of SmolLM2, a state-of-the-art "small" (1.7 billion parameter) language model (LM). To attain strong performance, we overtrain SmolLM2 on ~11 trillion tokens of data using a multi-stage training process that mixes web text with specialized math, code, and instruction-following data. We additionally introduce new specialized datasets (FineMath, Stack-Edu, and SmolTalk) at stages where we found existing datasets to be problematically small or low-quality. To inform our design decisions, we perform both small-scale ablations as well as a manual refinement process that updates the dataset mixing rates at each stage based on the performance at the previous stage. Ultimately, we demonstrate that SmolLM2 outperforms other recent small LMs including Qwen2.5-1.5B and Llama3.2-1B. To facilitate future research on LM development as well as applications of small LMs, we release both SmolLM2 as well as all of the datasets we prepared in the course of this project. |
|
2025-02-06T00:00:00 | 2502.03275 | Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning | [
"DiJia Su",
"Hanlin Zhu",
"Yingchen Xu",
"Jiantao Jiao",
"Yuandong Tian",
"Qinqing Zheng"
] | Large Language Models (LLMs) excel at reasoning and planning when trained on chainof-thought (CoT) data, where the step-by-step thought process is explicitly outlined by text tokens. However, this results in lengthy inputs where many words support textual coherence rather than core reasoning information, and processing these inputs consumes substantial computation resources. In this work, we propose a hybrid representation of the reasoning process, where we partially abstract away the initial reasoning steps using latent discrete tokens generated by VQ-VAE, significantly reducing the length of reasoning traces. We explore the use of latent trace abstractions in two scenarios: 1) training the model from scratch for the Keys-Finding Maze problem, 2) fine-tuning LLMs on this hybrid data with an extended vocabulary including unseen latent tokens, for both logical and mathematical reasoning problems. To facilitate effective learning, we introduce a simple training procedure that randomly mixes latent and text tokens, which enables fast adaptation to new latent tokens. Our approach consistently outperforms the baselines methods in various benchmarks. |
|
2025-02-06T00:00:00 | 2502.01154 | Jailbreaking with Universal Multi-Prompts | [
"Yu-Ling Hsu",
"Hsuan Su",
"Shang-Tse Chen"
] | Large language models (LLMs) have seen rapid development in recent years, revolutionizing various applications and significantly enhancing convenience and productivity. However, alongside their impressive capabilities, ethical concerns and new types of attacks, such as jailbreaking, have emerged. While most prompting techniques focus on optimizing adversarial inputs for individual cases, resulting in higher computational costs when dealing with large datasets. Less research has addressed the more general setting of training a universal attacker that can transfer to unseen tasks. In this paper, we introduce JUMP, a prompt-based method designed to jailbreak LLMs using universal multi-prompts. We also adapt our approach for defense, which we term DUMP. Experimental results demonstrate that our method for optimizing universal multi-prompts outperforms existing techniques. |
|
2025-02-06T00:00:00 | 2502.01105 | LayerTracer: Cognitive-Aligned Layered SVG Synthesis via Diffusion Transformer | [
"Yiren Song",
"Danze Chen",
"Mike Zheng Shou"
] | Generating cognitive-aligned layered SVGs remains challenging due to existing methods' tendencies toward either oversimplified single-layer outputs or optimization-induced shape redundancies. We propose LayerTracer, a diffusion transformer based framework that bridges this gap by learning designers' layered SVG creation processes from a novel dataset of sequential design operations. Our approach operates in two phases: First, a text-conditioned DiT generates multi-phase rasterized construction blueprints that simulate human design workflows. Second, layer-wise vectorization with path deduplication produces clean, editable SVGs. For image vectorization, we introduce a conditional diffusion mechanism that encodes reference images into latent tokens, guiding hierarchical reconstruction while preserving structural integrity. Extensive experiments demonstrate LayerTracer's superior performance against optimization-based and neural baselines in both generation quality and editability, effectively aligning AI-generated vectors with professional design cognition. |
|
2025-02-06T00:00:00 | 2502.02671 | On Teacher Hacking in Language Model Distillation | [
"Daniil Tiapkin",
"Daniele Calandriello",
"Johan Ferret",
"Sarah Perrin",
"Nino Vieillard",
"Alexandre Ramé",
"Mathieu Blondel"
] | Post-training of language models (LMs) increasingly relies on the following two stages: (i) knowledge distillation, where the LM is trained to imitate a larger teacher LM, and (ii) reinforcement learning from human feedback (RLHF), where the LM is aligned by optimizing a reward model. In the second RLHF stage, a well-known challenge is reward hacking, where the LM over-optimizes the reward model. Such phenomenon is in line with Goodhart's law and can lead to degraded performance on the true objective. In this paper, we investigate whether a similar phenomenon, that we call teacher hacking, can occur during knowledge distillation. This could arise because the teacher LM is itself an imperfect approximation of the true distribution. To study this, we propose a controlled experimental setup involving: (i) an oracle LM representing the ground-truth distribution, (ii) a teacher LM distilled from the oracle, and (iii) a student LM distilled from the teacher. Our experiments reveal the following insights. When using a fixed offline dataset for distillation, teacher hacking occurs; moreover, we can detect it by observing when the optimization process deviates from polynomial convergence laws. In contrast, employing online data generation techniques effectively mitigates teacher hacking. More precisely, we identify data diversity as the key factor in preventing hacking. Overall, our findings provide a deeper understanding of the benefits and limitations of distillation for building robust and efficient LMs. |
|
2025-02-06T00:00:00 | 2502.02928 | Large Language Model Guided Self-Debugging Code Generation | [
"Muntasir Adnan",
"Zhiwei Xu",
"Carlos C. N. Kuhn"
] | Automated code generation is gaining significant importance in intelligent computer programming and system deployment. However, current approaches often face challenges in computational efficiency and lack robust mechanisms for code parsing and error correction. In this work, we propose a novel framework, PyCapsule, with a simple yet effective two-agent pipeline and efficient self-debugging modules for Python code generation. PyCapsule features sophisticated prompt inference, iterative error handling, and case testing, ensuring high generation stability, safety, and correctness. Empirically, PyCapsule achieves up to 5.7% improvement of success rate on HumanEval, 10.3% on HumanEval-ET, and 24.4% on BigCodeBench compared to the state-of-art methods. We also observe a decrease in normalized success rate given more self-debugging attempts, potentially affected by limited and noisy error feedback in retention. PyCapsule demonstrates broader impacts on advancing lightweight and efficient code generation for artificial intelligence systems. |
|
2025-02-06T00:00:00 | 2502.00306 | Riddle Me This! Stealthy Membership Inference for Retrieval-Augmented Generation | [
"Ali Naseh",
"Yuefeng Peng",
"Anshuman Suri",
"Harsh Chaudhari",
"Alina Oprea",
"Amir Houmansadr"
] | Retrieval-Augmented Generation (RAG) enables Large Language Models (LLMs) to generate grounded responses by leveraging external knowledge databases without altering model parameters. Although the absence of weight tuning prevents leakage via model parameters, it introduces the risk of inference adversaries exploiting retrieved documents in the model's context. Existing methods for membership inference and data extraction often rely on jailbreaking or carefully crafted unnatural queries, which can be easily detected or thwarted with query rewriting techniques common in RAG systems. In this work, we present Interrogation Attack (IA), a membership inference technique targeting documents in the RAG datastore. By crafting natural-text queries that are answerable only with the target document's presence, our approach demonstrates successful inference with just 30 queries while remaining stealthy; straightforward detectors identify adversarial prompts from existing methods up to ~76x more frequently than those generated by our attack. We observe a 2x improvement in TPR@1%FPR over prior inference attacks across diverse RAG configurations, all while costing less than $0.02 per document inference. |
|
2025-02-06T00:00:00 | 2502.00226 | HackerRank-ASTRA: Evaluating Correctness & Consistency of Large Language Models on cross-domain multi-file project problems | [
"Jun Xing",
"Mayur Bhatia",
"Sahil Phulwani",
"Darshan Suresh",
"Rafik Matta"
] | Evaluating the real-world applicability of large language models (LLMs) provides valuable insights for their development and use in software development tasks. Existing benchmarks often focus on standalone coding problems or specific libraries, overlooking multi-file, project-based scenarios and lacking a rigorous evaluation of consistency. The HackerRank-ASTRA Benchmark introduces project-based coding problems that mirror real-world scenarios. It evaluates model consistency through 32 runs (k = 32) and median standard deviation while incorporating taxonomy-level analysis to assess sub-skill capabilities. Initial evaluations on 65 problems show that the top three models -- o1, o1-preview, and Claude-3.5-Sonnet-1022 -- achieved comparable average scores of 75%, with no statistically significant differences in performance. Notably, Claude-3.5-Sonnet-1022 demonstrated the highest consistency across problems, with low variability (SD = 0.0497), which was statistically significant compared to other models, highlighting its reliability for real-world software development tasks. |
|
2025-02-06T00:00:00 | 2502.02421 | Activation-Informed Merging of Large Language Models | [
"Amin Heyrani Nobari",
"Kaveh Alimohammadi",
"Ali ArjomandBigdeli",
"Akash Srivastava",
"Faez Ahmed",
"Navid Azizan"
] | Model merging, a method that combines the parameters and embeddings of multiple fine-tuned large language models (LLMs), offers a promising approach to enhance model performance across various tasks while maintaining computational efficiency. This paper introduces Activation-Informed Merging (AIM), a technique that integrates the information from the activation space of LLMs into the merging process to improve performance and robustness. AIM is designed as a flexible, complementary solution that is applicable to any existing merging method. It aims to preserve critical weights from the base model, drawing on principles from continual learning~(CL) and model compression. Utilizing a task-agnostic calibration set, AIM selectively prioritizes essential weights during merging. We empirically demonstrate that AIM significantly enhances the performance of merged models across multiple benchmarks. Our findings suggest that considering the activation-space information can provide substantial advancements in the model merging strategies for LLMs with up to 40\% increase in benchmark performance. |
|
2025-02-07T00:00:00 | 2502.04306 | ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference Optimization | [
"Yinjie Wang",
"Ling Yang",
"Guohao Li",
"Mengdi Wang",
"Bryon Aragam"
] | https://github.com/Gen-Verse/ScoreFlow | Recent research has leveraged large language model multi-agent systems for complex problem-solving while trying to reduce the manual effort required to build them, driving the development of automated agent workflow optimization methods. However, existing methods remain inflexible due to representational limitations, a lack of adaptability, and poor scalability when relying on discrete optimization techniques. We address these challenges with ScoreFlow, a simple yet high-performance framework that leverages efficient gradient-based optimization in a continuous space. ScoreFlow incorporates Score-DPO, a novel variant of the direct preference optimization method that accounts for quantitative feedback. Across six benchmarks spanning question answering, coding, and mathematical reasoning, ScoreFlow achieves an 8.2% improvement over existing baselines. Moreover, it empowers smaller models to outperform larger ones with lower inference costs. Project: https://github.com/Gen-Verse/ScoreFlow |
2025-02-07T00:00:00 | 2502.04153 | UltraIF: Advancing Instruction Following from the Wild | [
"Kaikai An",
"Li Sheng",
"Ganqu Cui",
"Shuzheng Si",
"Ning Ding",
"Yu Cheng",
"Baobao Chang"
] | https://github.com/kkk-an/UltraIF | Instruction-following made modern large language models (LLMs) helpful assistants. However, the key to taming LLMs on complex instructions remains mysterious, for that there are huge gaps between models trained by open-source community and those trained by leading companies. To bridge the gap, we propose a simple and scalable approach UltraIF for building LLMs that can follow complex instructions with open-source data. UltraIF first decomposes real-world user prompts into simpler queries, constraints, and corresponding evaluation questions for the constraints. Then, we train an UltraComposer to compose constraint-associated prompts with evaluation questions. This prompt composer allows us to synthesize complicated instructions as well as filter responses with evaluation questions. In our experiment, for the first time, we successfully align LLaMA-3.1-8B-Base to catch up with its instruct version on 5 instruction-following benchmarks without any benchmark information, using only 8B model as response generator and evaluator. The aligned model also achieved competitive scores on other benchmarks. Moreover, we also show that UltraIF could further improve LLaMA-3.1-8B-Instruct through self-alignment, motivating broader use cases for the method. Our code will be available at https://github.com/kkk-an/UltraIF. |
2025-02-07T00:00:00 | 2502.04295 | Beyond Prompt Content: Enhancing LLM Performance via Content-Format Integrated Prompt Optimization | [
"Yuanye Liu",
"Jiahang Xu",
"Li Lyna Zhang",
"Qi Chen",
"Xuan Feng",
"Yang Chen",
"Zhongxin Guo",
"Yuqing Yang",
"Cheng Peng"
] | https://github.com/HenryLau7/CFPO | Large Language Models (LLMs) have shown significant capability across various tasks, with their real-world effectiveness often driven by prompt design. While recent research has focused on optimizing prompt content, the role of prompt formatting, a critical but often overlooked dimension, has received limited systematic investigation. In this paper, we introduce Content-Format Integrated Prompt Optimization (CFPO), an innovative methodology that jointly optimizes both prompt content and formatting through an iterative refinement process. CFPO leverages natural language mutations to explore content variations and employs a dynamic format exploration strategy that systematically evaluates diverse format options. Our extensive evaluations across multiple tasks and open-source LLMs demonstrate that CFPO demonstrates measurable performance improvements compared to content-only optimization methods. This highlights the importance of integrated content-format optimization and offers a practical, model-agnostic approach to enhancing LLM performance. Code will be available at https://github.com/HenryLau7/CFPO. |
2025-02-07T00:00:00 | 2502.04296 | Learning Real-World Action-Video Dynamics with Heterogeneous Masked Autoregression | [
"Lirui Wang",
"Kevin Zhao",
"Chaoqi Liu",
"Xinlei Chen"
] | We propose Heterogeneous Masked Autoregression (HMA) for modeling action-video dynamics to generate high-quality data and evaluation in scaling robot learning. Building interactive video world models and policies for robotics is difficult due to the challenge of handling diverse settings while maintaining computational efficiency to run in real time. HMA uses heterogeneous pre-training from observations and action sequences across different robotic embodiments, domains, and tasks. HMA uses masked autoregression to generate quantized or soft tokens for video predictions. \ourshort achieves better visual fidelity and controllability than the previous robotic video generation models with 15 times faster speed in the real world. After post-training, this model can be used as a video simulator from low-level action inputs for evaluating policies and generating synthetic data. See this link https://liruiw.github.io/hma for more information. |
|
2025-02-07T00:00:00 | 2502.03860 | BOLT: Bootstrap Long Chain-of-Thought in Language Models without Distillation | [
"Bo Pang",
"Hanze Dong",
"Jiacheng Xu",
"Silvio Savarese",
"Yingbo Zhou",
"Caiming Xiong"
] | Large language models (LLMs), such as o1 from OpenAI, have demonstrated remarkable reasoning capabilities. o1 generates a long chain-of-thought (LongCoT) before answering a question. LongCoT allows LLMs to analyze problems, devise plans, reflect, and backtrack effectively. These actions empower LLM to solve complex problems. After the release of o1, many teams have attempted to replicate its LongCoT and reasoning capabilities. In terms of methods, they primarily rely on knowledge distillation with data from existing models with LongCoT capacities (e.g., OpenAI-o1, Qwen-QwQ, DeepSeek-R1-Preview), leaving significant uncertainties on systematically developing such reasoning abilities. In terms of data domains, these works focus narrowly on math while a few others include coding, limiting their generalizability. This paper introduces a novel approach to enable LLM's LongCoT capacity without distillation from o1-like models or expensive human annotations, where we bootstrap LongCoT (BOLT) from a standard instruct model. BOLT involves three stages: 1) LongCoT data bootstrapping with in-context learning on a standard instruct model; 2) LongCoT supervised finetuning; 3) online training to further refine LongCoT capacities. In BOLT, only a few in-context examples need to be constructed during the bootstrapping stage; in our experiments, we created 10 examples, demonstrating the feasibility of this approach. We use Llama-3.1-70B-Instruct to bootstrap LongCoT and apply our method to various model scales (7B, 8B, 70B). We achieve impressive performance on a variety of benchmarks, Arena-Hard, MT-Bench, WildBench, ZebraLogic, MATH500, which evaluate diverse task-solving and reasoning capabilities. |
|
2025-02-07T00:00:00 | 2502.02358 | MotionLab: Unified Human Motion Generation and Editing via the Motion-Condition-Motion Paradigm | [
"Ziyan Guo",
"Zeyu Hu",
"Na Zhao",
"De Wen Soh"
] | Human motion generation and editing are key components of computer graphics and vision. However, current approaches in this field tend to offer isolated solutions tailored to specific tasks, which can be inefficient and impractical for real-world applications. While some efforts have aimed to unify motion-related tasks, these methods simply use different modalities as conditions to guide motion generation. Consequently, they lack editing capabilities, fine-grained control, and fail to facilitate knowledge sharing across tasks. To address these limitations and provide a versatile, unified framework capable of handling both human motion generation and editing, we introduce a novel paradigm: Motion-Condition-Motion, which enables the unified formulation of diverse tasks with three concepts: source motion, condition, and target motion. Based on this paradigm, we propose a unified framework, MotionLab, which incorporates rectified flows to learn the mapping from source motion to target motion, guided by the specified conditions. In MotionLab, we introduce the 1) MotionFlow Transformer to enhance conditional generation and editing without task-specific modules; 2) Aligned Rotational Position Encoding} to guarantee the time synchronization between source motion and target motion; 3) Task Specified Instruction Modulation; and 4) Motion Curriculum Learning for effective multi-task learning and knowledge sharing across tasks. Notably, our MotionLab demonstrates promising generalization capabilities and inference efficiency across multiple benchmarks for human motion. Our code and additional video results are available at: https://diouo.github.io/motionlab.github.io/. |
|
2025-02-07T00:00:00 | 2502.03544 | Gold-medalist Performance in Solving Olympiad Geometry with AlphaGeometry2 | [
"Yuri Chervonyi",
"Trieu H. Trinh",
"Miroslav Olšák",
"Xiaomeng Yang",
"Hoang Nguyen",
"Marcelo Menegali",
"Junehyuk Jung",
"Vikas Verma",
"Quoc V. Le",
"Thang Luong"
] | We present AlphaGeometry2, a significantly improved version of AlphaGeometry introduced in Trinh et al. (2024), which has now surpassed an average gold medalist in solving Olympiad geometry problems. To achieve this, we first extend the original AlphaGeometry language to tackle harder problems involving movements of objects, and problems containing linear equations of angles, ratios, and distances. This, together with other additions, has markedly improved the coverage rate of the AlphaGeometry language on International Math Olympiads (IMO) 2000-2024 geometry problems from 66% to 88%. The search process of AlphaGeometry2 has also been greatly improved through the use of Gemini architecture for better language modeling, and a novel knowledge-sharing mechanism that combines multiple search trees. Together with further enhancements to the symbolic engine and synthetic data generation, we have significantly boosted the overall solving rate of AlphaGeometry2 to 84% for all geometry problems over the last 25 years, compared to 54% previously. AlphaGeometry2 was also part of the system that achieved silver-medal standard at IMO 2024 https://dpmd.ai/imo-silver. Last but not least, we report progress towards using AlphaGeometry2 as a part of a fully automated system that reliably solves geometry problems directly from natural language input. |
|
2025-02-07T00:00:00 | 2502.04128 | Llasa: Scaling Train-Time and Inference-Time Compute for Llama-based Speech Synthesis | [
"Zhen Ye",
"Xinfa Zhu",
"Chi-Min Chan",
"Xinsheng Wang",
"Xu Tan",
"Jiahe Lei",
"Yi Peng",
"Haohe Liu",
"Yizhu Jin",
"Zheqi DAI",
"Hongzhan Lin",
"Jianyi Chen",
"Xingjian Du",
"Liumeng Xue",
"Yunlin Chen",
"Zhifei Li",
"Lei Xie",
"Qiuqiang Kong",
"Yike Guo",
"Wei Xue"
] | Recent advances in text-based large language models (LLMs), particularly in the GPT series and the o1 model, have demonstrated the effectiveness of scaling both training-time and inference-time compute. However, current state-of-the-art TTS systems leveraging LLMs are often multi-stage, requiring separate models (e.g., diffusion models after LLM), complicating the decision of whether to scale a particular model during training or testing. This work makes the following contributions: First, we explore the scaling of train-time and inference-time compute for speech synthesis. Second, we propose a simple framework Llasa for speech synthesis that employs a single-layer vector quantizer (VQ) codec and a single Transformer architecture to fully align with standard LLMs such as Llama. Our experiments reveal that scaling train-time compute for Llasa consistently improves the naturalness of synthesized speech and enables the generation of more complex and accurate prosody patterns. Furthermore, from the perspective of scaling inference-time compute, we employ speech understanding models as verifiers during the search, finding that scaling inference-time compute shifts the sampling modes toward the preferences of specific verifiers, thereby improving emotional expressiveness, timbre consistency, and content accuracy. In addition, we released the checkpoint and training code for our TTS model (1B, 3B, 8B) and codec model publicly available. |
|
2025-02-07T00:00:00 | 2502.04270 | PILAF: Optimal Human Preference Sampling for Reward Modeling | [
"Yunzhen Feng",
"Ariel Kwiatkowski",
"Kunhao Zheng",
"Julia Kempe",
"Yaqi Duan"
] | As large language models increasingly drive real-world applications, aligning them with human values becomes paramount. Reinforcement Learning from Human Feedback (RLHF) has emerged as a key technique, translating preference data into reward models when oracle human values remain inaccessible. In practice, RLHF mostly relies on approximate reward models, which may not consistently guide the policy toward maximizing the underlying human values. We propose Policy-Interpolated Learning for Aligned Feedback (PILAF), a novel response sampling strategy for preference labeling that explicitly aligns preference learning with maximizing the underlying oracle reward. PILAF is theoretically grounded, demonstrating optimality from both an optimization and a statistical perspective. The method is straightforward to implement and demonstrates strong performance in iterative and online RLHF settings where feedback curation is critical. |
|
2025-02-07T00:00:00 | 2502.03621 | DynVFX: Augmenting Real Videos with Dynamic Content | [
"Danah Yatim",
"Rafail Fridman",
"Omer Bar-Tal",
"Tali Dekel"
] | We present a method for augmenting real-world videos with newly generated dynamic content. Given an input video and a simple user-provided text instruction describing the desired content, our method synthesizes dynamic objects or complex scene effects that naturally interact with the existing scene over time. The position, appearance, and motion of the new content are seamlessly integrated into the original footage while accounting for camera motion, occlusions, and interactions with other dynamic objects in the scene, resulting in a cohesive and realistic output video. We achieve this via a zero-shot, training-free framework that harnesses a pre-trained text-to-video diffusion transformer to synthesize the new content and a pre-trained Vision Language Model to envision the augmented scene in detail. Specifically, we introduce a novel inference-based method that manipulates features within the attention mechanism, enabling accurate localization and seamless integration of the new content while preserving the integrity of the original scene. Our method is fully automated, requiring only a simple user instruction. We demonstrate its effectiveness on a wide range of edits applied to real-world videos, encompassing diverse objects and scenarios involving both camera and object motion. |
|
2025-02-07T00:00:00 | 2502.03639 | Towards Physical Understanding in Video Generation: A 3D Point Regularization Approach | [
"Yunuo Chen",
"Junli Cao",
"Anil Kag",
"Vidit Goel",
"Sergei Korolev",
"Chenfanfu Jiang",
"Sergey Tulyakov",
"Jian Ren"
] | We present a novel video generation framework that integrates 3-dimensional geometry and dynamic awareness. To achieve this, we augment 2D videos with 3D point trajectories and align them in pixel space. The resulting 3D-aware video dataset, PointVid, is then used to fine-tune a latent diffusion model, enabling it to track 2D objects with 3D Cartesian coordinates. Building on this, we regularize the shape and motion of objects in the video to eliminate undesired artifacts, \eg, nonphysical deformation. Consequently, we enhance the quality of generated RGB videos and alleviate common issues like object morphing, which are prevalent in current video models due to a lack of shape awareness. With our 3D augmentation and regularization, our model is capable of handling contact-rich scenarios such as task-oriented videos. These videos involve complex interactions of solids, where 3D information is essential for perceiving deformation and contact. Furthermore, our model improves the overall quality of video generation by promoting the 3D consistency of moving objects and reducing abrupt changes in shape and motion. |
|
2025-02-07T00:00:00 | 2502.04299 | MotionCanvas: Cinematic Shot Design with Controllable Image-to-Video Generation | [
"Jinbo Xing",
"Long Mai",
"Cusuh Ham",
"Jiahui Huang",
"Aniruddha Mahapatra",
"Chi-Wing Fu",
"Tien-Tsin Wong",
"Feng Liu"
] | This paper presents a method that allows users to design cinematic video shots in the context of image-to-video generation. Shot design, a critical aspect of filmmaking, involves meticulously planning both camera movements and object motions in a scene. However, enabling intuitive shot design in modern image-to-video generation systems presents two main challenges: first, effectively capturing user intentions on the motion design, where both camera movements and scene-space object motions must be specified jointly; and second, representing motion information that can be effectively utilized by a video diffusion model to synthesize the image animations. To address these challenges, we introduce MotionCanvas, a method that integrates user-driven controls into image-to-video (I2V) generation models, allowing users to control both object and camera motions in a scene-aware manner. By connecting insights from classical computer graphics and contemporary video generation techniques, we demonstrate the ability to achieve 3D-aware motion control in I2V synthesis without requiring costly 3D-related training data. MotionCanvas enables users to intuitively depict scene-space motion intentions, and translates them into spatiotemporal motion-conditioning signals for video diffusion models. We demonstrate the effectiveness of our method on a wide range of real-world image content and shot-design scenarios, highlighting its potential to enhance the creative workflows in digital content creation and adapt to various image and video editing applications. |
|
2025-02-07T00:00:00 | 2502.03032 | Analyze Feature Flow to Enhance Interpretation and Steering in Language Models | [
"Daniil Laptev",
"Nikita Balagansky",
"Yaroslav Aksenov",
"Daniil Gavrilov"
] | We introduce a new approach to systematically map features discovered by sparse autoencoder across consecutive layers of large language models, extending earlier work that examined inter-layer feature links. By using a data-free cosine similarity technique, we trace how specific features persist, transform, or first appear at each stage. This method yields granular flow graphs of feature evolution, enabling fine-grained interpretability and mechanistic insights into model computations. Crucially, we demonstrate how these cross-layer feature maps facilitate direct steering of model behavior by amplifying or suppressing chosen features, achieving targeted thematic control in text generation. Together, our findings highlight the utility of a causal, cross-layer interpretability framework that not only clarifies how features develop through forward passes but also provides new means for transparent manipulation of large language models. |
|
2025-02-07T00:00:00 | 2502.04328 | Ola: Pushing the Frontiers of Omni-Modal Language Model with Progressive Modality Alignment | [
"Zuyan Liu",
"Yuhao Dong",
"Jiahui Wang",
"Ziwei Liu",
"Winston Hu",
"Jiwen Lu",
"Yongming Rao"
] | https://github.com/Ola-Omni/Ola | Recent advances in large language models, particularly following GPT-4o, have sparked increasing interest in developing omni-modal models capable of understanding more modalities. While some open-source alternatives have emerged, there is still a notable lag behind specialized single-modality models in performance. In this paper, we present Ola, an Omni-modal language model that achieves competitive performance across image, video, and audio understanding compared to specialized counterparts. The core design of Ola lies in its progressive modality alignment strategy that extends the supporting modality of the language model progressively. Our training pipeline begins with the most distinct modalities: image and text, then gradually expands the skill sets of the model using speech data that connects language and audio knowledge, and video data that connects all modalities. The progressive learning pipeline also enables us to maintain a relatively small size of the cross-modal alignment data, making developing omni-modal from existing vision-language models easy and less costly. Moreover, to unlock an advanced interactive experience like GPT-4o, we further design a sentence-wise decoding solution for streaming speech generation. Extensive experiments demonstrate that Ola surpasses existing open omni-modal LLMs across all modalities while achieving highly competitive performance compared to state-of-the-art specialized models of similar sizes. We aim to make Ola a fully open omni-modal understanding solution to advance future research in this emerging field. Model weights, code, and data are open-sourced at https://github.com/Ola-Omni/Ola. |
2025-02-07T00:00:00 | 2502.04322 | Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple Interactions | [
"Yik Siu Chan",
"Narutatsu Ri",
"Yuxin Xiao",
"Marzyeh Ghassemi"
] | Despite extensive safety alignment efforts, large language models (LLMs) remain vulnerable to jailbreak attacks that elicit harmful behavior. While existing studies predominantly focus on attack methods that require technical expertise, two critical questions remain underexplored: (1) Are jailbroken responses truly useful in enabling average users to carry out harmful actions? (2) Do safety vulnerabilities exist in more common, simple human-LLM interactions? In this paper, we demonstrate that LLM responses most effectively facilitate harmful actions when they are both actionable and informative--two attributes easily elicited in multi-step, multilingual interactions. Using this insight, we propose HarmScore, a jailbreak metric that measures how effectively an LLM response enables harmful actions, and Speak Easy, a simple multi-step, multilingual attack framework. Notably, by incorporating Speak Easy into direct request and jailbreak baselines, we see an average absolute increase of 0.319 in Attack Success Rate and 0.426 in HarmScore in both open-source and proprietary LLMs across four safety benchmarks. Our work reveals a critical yet often overlooked vulnerability: Malicious users can easily exploit common interaction patterns for harmful intentions. |
|
2025-02-07T00:00:00 | 2502.04235 | MAGA: MAssive Genre-Audience Reformulation to Pretraining Corpus Expansion | [
"Xintong Hao",
"Ke Shen",
"Chenggang Li"
] | Despite the remarkable capabilities of large language models across various tasks, their continued scaling faces a critical challenge: the scarcity of high-quality pretraining data. While model architectures continue to evolve, the natural language data struggles to scale up. To tackle this bottleneck, we propose MAssive Genre-Audience~(MAGA) reformulation method, which systematic synthesizes diverse, contextually-rich pretraining data from existing corpus. This work makes three main contributions: (1) We propose MAGA reformulation method, a lightweight and scalable approach for pretraining corpus expansion, and build a 770B tokens MAGACorpus. (2) We evaluate MAGACorpus with different data budget scaling strategies, demonstrating consistent improvements across various model sizes (134M-13B), establishing the necessity for next-generation large-scale synthetic pretraining language models. (3) Through comprehensive analysis, we investigate prompt engineering's impact on synthetic training collapse and reveal limitations in conventional collapse detection metrics using validation losses. Our work shows that MAGA can substantially expand training datasets while maintaining quality, offering a reliably pathway for scaling models beyond data limitations. |
|
2025-02-07T00:00:00 | 2502.04313 | Great Models Think Alike and this Undermines AI Oversight | [
"Shashwat Goel",
"Joschka Struber",
"Ilze Amanda Auzina",
"Karuna K Chandra",
"Ponnurangam Kumaraguru",
"Douwe Kiela",
"Ameya Prabhu",
"Matthias Bethge",
"Jonas Geiping"
] | As Language Model (LM) capabilities advance, evaluating and supervising them at scale is getting harder for humans. There is hope that other language models can automate both these tasks, which we refer to as "AI Oversight". We study how model similarity affects both aspects of AI oversight by proposing a probabilistic metric for LM similarity based on overlap in model mistakes. Using this metric, we first show that LLM-as-a-judge scores favor models similar to the judge, generalizing recent self-preference results. Then, we study training on LM annotations, and find complementary knowledge between the weak supervisor and strong student model plays a crucial role in gains from "weak-to-strong generalization". As model capabilities increase, it becomes harder to find their mistakes, and we might defer more to AI oversight. However, we observe a concerning trend -- model mistakes are becoming more similar with increasing capabilities, pointing to risks from correlated failures. Our work underscores the importance of reporting and correcting for model similarity, especially in the emerging paradigm of AI oversight. |
|
2025-02-07T00:00:00 | 2502.00989 | ChartCitor: Multi-Agent Framework for Fine-Grained Chart Visual Attribution | [
"Kanika Goswami",
"Puneet Mathur",
"Ryan Rossi",
"Franck Dernoncourt"
] | Large Language Models (LLMs) can perform chart question-answering tasks but often generate unverified hallucinated responses. Existing answer attribution methods struggle to ground responses in source charts due to limited visual-semantic context, complex visual-text alignment requirements, and difficulties in bounding box prediction across complex layouts. We present ChartCitor, a multi-agent framework that provides fine-grained bounding box citations by identifying supporting evidence within chart images. The system orchestrates LLM agents to perform chart-to-table extraction, answer reformulation, table augmentation, evidence retrieval through pre-filtering and re-ranking, and table-to-chart mapping. ChartCitor outperforms existing baselines across different chart types. Qualitative user studies show that ChartCitor helps increase user trust in Generative AI by providing enhanced explainability for LLM-assisted chart QA and enables professionals to be more productive. |
|
2025-02-07T00:00:00 | 2501.19085 | Enhancing Code Generation for Low-Resource Languages: No Silver Bullet | [
"Alessandro Giagnorio",
"Alberto Martin-Lopez",
"Gabriele Bavota"
] | The advent of Large Language Models (LLMs) has significantly advanced the field of automated code generation. LLMs rely on large and diverse datasets to learn syntax, semantics, and usage patterns of programming languages. For low-resource languages (i.e., niche programming languages characterized by the scarcity of training data), the limited availability of such data hampers the models' ability to generalize effectively, resulting in poorer code generation performance as compared to high-resource languages. For this reason, there is a quest for techniques able to close this performance gap. We present an empirical study investigating the effectiveness of several approaches for boosting LLMs' performance on low-resource languages, namely: (i) a classic fine-tuning, which is however capped in size by the scarcity of training data; (ii) three variants of in-context learning, with prompts crafted to provide the LLM with additional information about the low-resource language (e.g., few-shot examples showcasing features of the targeted language); and (iii) a pre-training objective teaching the model how to translate between high- and low-resource languages. The context of our study are two low-resource languages (R and Racket) and six LLMs having different architectures and sizes. Our findings reveal that a fine-tuning is usually the best choice for smaller LLMs, possibly due to the fact that even a small dataset is sufficient to train their limited number of parameters. With the increase in size of the models, in-context learning becomes more and more effective, representing a safe and cheap bet (i.e., it always helps, but with different magnitudes). Differently, very large LLMs may deteriorate their performance on low-resource languages when fine-tuning is performed, possibly due to the lack of enough data needed to effectively update their weights. |
|
2025-02-07T00:00:00 | 2502.00988 | PlotGen: Multi-Agent LLM-based Scientific Data Visualization via Multimodal Feedback | [
"Kanika Goswami",
"Puneet Mathur",
"Ryan Rossi",
"Franck Dernoncourt"
] | Scientific data visualization is pivotal for transforming raw data into comprehensible visual representations, enabling pattern recognition, forecasting, and the presentation of data-driven insights. However, novice users often face difficulties due to the complexity of selecting appropriate tools and mastering visualization techniques. Large Language Models (LLMs) have recently demonstrated potential in assisting code generation, though they struggle with accuracy and require iterative debugging. In this paper, we propose PlotGen, a novel multi-agent framework aimed at automating the creation of precise scientific visualizations. PlotGen orchestrates multiple LLM-based agents, including a Query Planning Agent that breaks down complex user requests into executable steps, a Code Generation Agent that converts pseudocode into executable Python code, and three retrieval feedback agents - a Numeric Feedback Agent, a Lexical Feedback Agent, and a Visual Feedback Agent - that leverage multimodal LLMs to iteratively refine the data accuracy, textual labels, and visual correctness of generated plots via self-reflection. Extensive experiments show that PlotGen outperforms strong baselines, achieving a 4-6 percent improvement on the MatPlotBench dataset, leading to enhanced user trust in LLM-generated visualizations and improved novice productivity due to a reduction in debugging time needed for plot errors. |
|
2025-02-07T00:00:00 | 2502.00473 | Weak-to-Strong Diffusion with Reflection | [
"Lichen Bai",
"Masashi Sugiyama",
"Zeke Xie"
] | The goal of diffusion generative models is to align the learned distribution with the real data distribution through gradient score matching. However, inherent limitations in training data quality, modeling strategies, and architectural design lead to inevitable gap between generated outputs and real data. To reduce this gap, we propose Weak-to-Strong Diffusion (W2SD), a novel framework that utilizes the estimated difference between existing weak and strong models (i.e., weak-to-strong difference) to approximate the gap between an ideal model and a strong model. By employing a reflective operation that alternates between denoising and inversion with weak-to-strong difference, we theoretically understand that W2SD steers latent variables along sampling trajectories toward regions of the real data distribution. W2SD is highly flexible and broadly applicable, enabling diverse improvements through the strategic selection of weak-to-strong model pairs (e.g., DreamShaper vs. SD1.5, good experts vs. bad experts in MoE). Extensive experiments demonstrate that W2SD significantly improves human preference, aesthetic quality, and prompt adherence, achieving SOTA performance across various modalities (e.g., image, video), architectures (e.g., UNet-based, DiT-based, MoE), and benchmarks. For example, Juggernaut-XL with W2SD can improve with the HPSv2 winning rate up to 90% over the original results. Moreover, the performance gains achieved by W2SD markedly outweigh its additional computational overhead, while the cumulative improvements from different weak-to-strong difference further solidify its practical utility and deployability. |
|
2025-02-07T00:00:00 | 2502.04320 | ConceptAttention: Diffusion Transformers Learn Highly Interpretable Features | [
"Alec Helbling",
"Tuna Han Salih Meral",
"Ben Hoover",
"Pinar Yanardag",
"Duen Horng Chau"
] | Do the rich representations of multi-modal diffusion transformers (DiTs) exhibit unique properties that enhance their interpretability? We introduce ConceptAttention, a novel method that leverages the expressive power of DiT attention layers to generate high-quality saliency maps that precisely locate textual concepts within images. Without requiring additional training, ConceptAttention repurposes the parameters of DiT attention layers to produce highly contextualized concept embeddings, contributing the major discovery that performing linear projections in the output space of DiT attention layers yields significantly sharper saliency maps compared to commonly used cross-attention mechanisms. Remarkably, ConceptAttention even achieves state-of-the-art performance on zero-shot image segmentation benchmarks, outperforming 11 other zero-shot interpretability methods on the ImageNet-Segmentation dataset and on a single-class subset of PascalVOC. Our work contributes the first evidence that the representations of multi-modal DiT models like Flux are highly transferable to vision tasks like segmentation, even outperforming multi-modal foundation models like CLIP. |
|
2025-02-10T00:00:00 | 2502.05173 | VideoRoPE: What Makes for Good Video Rotary Position Embedding? | [
"Xilin Wei",
"Xiaoran Liu",
"Yuhang Zang",
"Xiaoyi Dong",
"Pan Zhang",
"Yuhang Cao",
"Jian Tong",
"Haodong Duan",
"Qipeng Guo",
"Jiaqi Wang",
"Xipeng Qiu",
"Dahua Lin"
] | https://github.com/Wiselnn570/VideoRoPE{https: | While Rotary Position Embedding (RoPE) and its variants are widely adopted for their long-context capabilities, the extension of the 1D RoPE to video, with its complex spatio-temporal structure, remains an open challenge. This work first introduces a comprehensive analysis that identifies four key characteristics essential for the effective adaptation of RoPE to video, which have not been fully considered in prior work. As part of our analysis, we introduce a challenging V-NIAH-D (Visual Needle-In-A-Haystack with Distractors) task, which adds periodic distractors into V-NIAH. The V-NIAH-D task demonstrates that previous RoPE variants, lacking appropriate temporal dimension allocation, are easily misled by distractors. Based on our analysis, we introduce VideoRoPE, with a 3D structure designed to preserve spatio-temporal relationships. VideoRoPE features low-frequency temporal allocation to mitigate periodic oscillations, a diagonal layout to maintain spatial symmetry, and adjustable temporal spacing to decouple temporal and spatial indexing. VideoRoPE consistently surpasses previous RoPE variants, across diverse downstream tasks such as long video retrieval, video understanding, and video hallucination. Our code will be available at https://github.com/Wiselnn570/VideoRoPE{https://github.com/Wiselnn570/VideoRoPE}. |
2025-02-10T00:00:00 | 2502.04520 | Linear Correlation in LM's Compositional Generalization and Hallucination | [
"Letian Peng",
"Chenyang An",
"Shibo Hao",
"Chengyu Dong",
"Jingbo Shang"
] | The generalization of language models (LMs) is undergoing active debates, contrasting their potential for general intelligence with their struggles with basic knowledge composition (e.g., reverse/transition curse). This paper uncovers the phenomenon of linear correlations in LMs during knowledge composition. For explanation, there exists a linear transformation between certain related knowledge that maps the next token prediction logits from one prompt to another, e.g., "X lives in the city of" rightarrow "X lives in the country of" for every given X. This mirrors the linearity in human knowledge composition, such as Paris rightarrow France. Our findings indicate that the linear transformation is resilient to large-scale fine-tuning, generalizing updated knowledge when aligned with real-world relationships, but causing hallucinations when it deviates. Empirical results suggest that linear correlation can serve as a potential identifier of LM's generalization. Finally, we show such linear correlations can be learned with a single feedforward network and pre-trained vocabulary representations, indicating LM generalization heavily relies on the latter. |
|
2025-02-10T00:00:00 | 2502.04403 | Agency Is Frame-Dependent | [
"David Abel",
"André Barreto",
"Michael Bowling",
"Will Dabney",
"Shi Dong",
"Steven Hansen",
"Anna Harutyunyan",
"Khimya Khetarpal",
"Clare Lyle",
"Razvan Pascanu",
"Georgios Piliouras",
"Doina Precup",
"Jonathan Richens",
"Mark Rowland",
"Tom Schaul",
"Satinder Singh"
] | Agency is a system's capacity to steer outcomes toward a goal, and is a central topic of study across biology, philosophy, cognitive science, and artificial intelligence. Determining if a system exhibits agency is a notoriously difficult question: Dennett (1989), for instance, highlights the puzzle of determining which principles can decide whether a rock, a thermostat, or a robot each possess agency. We here address this puzzle from the viewpoint of reinforcement learning by arguing that agency is fundamentally frame-dependent: Any measurement of a system's agency must be made relative to a reference frame. We support this claim by presenting a philosophical argument that each of the essential properties of agency proposed by Barandiaran et al. (2009) and Moreno (2018) are themselves frame-dependent. We conclude that any basic science of agency requires frame-dependence, and discuss the implications of this claim for reinforcement learning. |
|
2025-02-10T00:00:00 | 2502.04896 | Goku: Flow Based Video Generative Foundation Models | [
"Shoufa Chen",
"Chongjian Ge",
"Yuqi Zhang",
"Yida Zhang",
"Fengda Zhu",
"Hao Yang",
"Hongxiang Hao",
"Hui Wu",
"Zhichao Lai",
"Yifei Hu",
"Ting-Che Lin",
"Shilong Zhang",
"Fu Li",
"Chuan Li",
"Xing Wang",
"Yanghua Peng",
"Peize Sun",
"Ping Luo",
"Yi Jiang",
"Zehuan Yuan",
"Bingyue Peng",
"Xiaobing Liu"
] | This paper introduces Goku, a state-of-the-art family of joint image-and-video generation models leveraging rectified flow Transformers to achieve industry-leading performance. We detail the foundational elements enabling high-quality visual generation, including the data curation pipeline, model architecture design, flow formulation, and advanced infrastructure for efficient and robust large-scale training. The Goku models demonstrate superior performance in both qualitative and quantitative evaluations, setting new benchmarks across major tasks. Specifically, Goku achieves 0.76 on GenEval and 83.65 on DPG-Bench for text-to-image generation, and 84.85 on VBench for text-to-video tasks. We believe that this work provides valuable insights and practical advancements for the research community in developing joint image-and-video generation models. |
|
2025-02-10T00:00:00 | 2502.05171 | Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach | [
"Jonas Geiping",
"Sean McLeish",
"Neel Jain",
"John Kirchenbauer",
"Siddharth Singh",
"Brian R. Bartoldson",
"Bhavya Kailkhura",
"Abhinav Bhatele",
"Tom Goldstein"
] | We study a novel language model architecture that is capable of scaling test-time computation by implicitly reasoning in latent space. Our model works by iterating a recurrent block, thereby unrolling to arbitrary depth at test-time. This stands in contrast to mainstream reasoning models that scale up compute by producing more tokens. Unlike approaches based on chain-of-thought, our approach does not require any specialized training data, can work with small context windows, and can capture types of reasoning that are not easily represented in words. We scale a proof-of-concept model to 3.5 billion parameters and 800 billion tokens. We show that the resulting model can improve its performance on reasoning benchmarks, sometimes dramatically, up to a computation load equivalent to 50 billion parameters. |
|
2025-02-10T00:00:00 | 2502.04363 | On-device Sora: Enabling Diffusion-Based Text-to-Video Generation for Mobile Devices | [
"Bosung Kim",
"Kyuhwan Lee",
"Isu Jeong",
"Jungmin Cheon",
"Yeojin Lee",
"Seulki Lee"
] | https://github.com/eai-lab/On-device-Sora | We present On-device Sora, a first pioneering solution for diffusion-based on-device text-to-video generation that operates efficiently on smartphone-grade devices. Building on Open-Sora, On-device Sora applies three novel techniques to address the challenges of diffusion-based text-to-video generation on computation- and memory-limited mobile devices. First, Linear Proportional Leap (LPL) reduces the excessive denoising steps required in video diffusion through an efficient leap-based approach. Second, Temporal Dimension Token Merging (TDTM) minimizes intensive token-processing computation in attention layers by merging consecutive tokens along the temporal dimension. Third, Concurrent Inference with Dynamic Loading (CI-DL) dynamically partitions large models into smaller blocks and loads them into memory for concurrent model inference, effectively addressing the challenges of limited device memory. We implement On-device Sora on the iPhone 15 Pro, and the experimental evaluations demonstrate that it is capable of generating high-quality videos on the device, comparable to those produced by Open-Sora running on high-end GPUs. These results show that On-device Sora enables efficient and high-quality video generation on resource-constrained mobile devices, expanding accessibility, ensuring user privacy, reducing dependence on cloud infrastructure, and lowering associated costs. We envision the proposed On-device Sora as a significant first step toward democratizing state-of-the-art generative technologies, enabling video generation capabilities on commodity mobile and embedded devices. The code implementation is publicly available at an GitHub repository: https://github.com/eai-lab/On-device-Sora. |
2025-02-10T00:00:00 | 2502.04728 | Generating Symbolic World Models via Test-time Scaling of Large Language Models | [
"Zhouliang Yu",
"Yuhuan Yuan",
"Tim Z. Xiao",
"Fuxiang Frank Xia",
"Jie Fu",
"Ge Zhang",
"Ge Lin",
"Weiyang Liu"
] | Solving complex planning problems requires Large Language Models (LLMs) to explicitly model the state transition to avoid rule violations, comply with constraints, and ensure optimality-a task hindered by the inherent ambiguity of natural language. To overcome such ambiguity, Planning Domain Definition Language (PDDL) is leveraged as a planning abstraction that enables precise and formal state descriptions. With PDDL, we can generate a symbolic world model where classic searching algorithms, such as A*, can be seamlessly applied to find optimal plans. However, directly generating PDDL domains with current LLMs remains an open challenge due to the lack of PDDL training data. To address this challenge, we propose to scale up the test-time computation of LLMs to enhance their PDDL reasoning capabilities, thereby enabling the generation of high-quality PDDL domains. Specifically, we introduce a simple yet effective algorithm, which first employs a Best-of-N sampling approach to improve the quality of the initial solution and then refines the solution in a fine-grained manner with verbalized machine learning. Our method outperforms o1-mini by a considerable margin in the generation of PDDL domain, achieving over 50% success rate on two tasks (i.e., generating PDDL domains from natural language description or PDDL problems). This is done without requiring additional training. By taking advantage of PDDL as state abstraction, our method is able to outperform current state-of-the-art methods on almost all competition-level planning tasks. |
|
2025-02-10T00:00:00 | 2502.04404 | Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Models | [
"Xiao-Wen Yang",
"Xuan-Yi Zhu",
"Wen-Da Wei",
"Ding-Chu Zhang",
"Jie-Jing Shao",
"Zhi Zhou",
"Lan-Zhe Guo",
"Yu-Feng Li"
] | The integration of slow-thinking mechanisms into large language models (LLMs) offers a promising way toward achieving Level 2 AGI Reasoners, as exemplified by systems like OpenAI's o1. However, several significant challenges remain, including inefficient overthinking and an overreliance on auxiliary reward models. We point out that these limitations stem from LLMs' inability to internalize the search process, a key component of effective reasoning. A critical step toward addressing this issue is enabling LLMs to autonomously determine when and where to backtrack, a fundamental operation in traditional search algorithms. To this end, we propose a self-backtracking mechanism that equips LLMs with the ability to backtrack during both training and inference. This mechanism not only enhances reasoning ability but also efficiency by transforming slow-thinking processes into fast-thinking through self-improvement. Empirical evaluations demonstrate that our proposal significantly enhances the reasoning capabilities of LLMs, achieving a performance gain of over 40 percent compared to the optimal-path supervised fine-tuning method. We believe this study introduces a novel and promising pathway for developing more advanced and robust Reasoners. |
|
2025-02-10T00:00:00 | 2502.04350 | CodeSteer: Symbolic-Augmented Language Models via Code/Text Guidance | [
"Yongchao Chen",
"Yilun Hao",
"Yueying Liu",
"Yang Zhang",
"Chuchu Fan"
] | https://github.com/yongchao98/CodeSteer-v1.0 | Existing methods fail to effectively steer Large Language Models (LLMs) between textual reasoning and code generation, leaving symbolic computing capabilities underutilized. We introduce CodeSteer, an effective method for guiding LLM code/text generation. We construct a comprehensive benchmark SymBench comprising 37 symbolic tasks with adjustable complexity and also synthesize datasets of 12k multi-round guidance/generation trajectories and 5.5k guidance comparison pairs. We fine-tune the Llama-3-8B model with a newly designed multi-round supervised fine-tuning (SFT) and direct preference optimization (DPO). The resulting model, CodeSteerLLM, augmented with the proposed symbolic and self-answer checkers, effectively guides the code/text generation of larger models. Augmenting GPT-4o with CodeSteer raises its average performance score from 53.3 to 86.4, even outperforming the existing best LLM OpenAI o1 (82.7), o1-preview (74.8), and DeepSeek R1 (76.8) across all 37 tasks (28 seen, 9 unseen). Trained for GPT-4o, CodeSteer demonstrates superior generalizability, providing an average 41.8 performance boost on Claude, Mistral, and GPT-3.5. CodeSteer-guided LLMs fully harness symbolic computing to maintain strong performance on highly complex tasks. Models, Datasets, and Codes are available at https://github.com/yongchao98/CodeSteer-v1.0. |
2025-02-10T00:00:00 | 2502.04507 | Fast Video Generation with Sliding Tile Attention | [
"Peiyuan Zhang",
"Yongqi Chen",
"Runlong Su",
"Hangliang Ding",
"Ion Stoica",
"Zhenghong Liu",
"Hao Zhang"
] | Diffusion Transformers (DiTs) with 3D full attention power state-of-the-art video generation, but suffer from prohibitive compute cost -- when generating just a 5-second 720P video, attention alone takes 800 out of 945 seconds of total inference time. This paper introduces sliding tile attention (STA) to address this challenge. STA leverages the observation that attention scores in pretrained video diffusion models predominantly concentrate within localized 3D windows. By sliding and attending over the local spatial-temporal region, STA eliminates redundancy from full attention. Unlike traditional token-wise sliding window attention (SWA), STA operates tile-by-tile with a novel hardware-aware sliding window design, preserving expressiveness while being hardware-efficient. With careful kernel-level optimizations, STA offers the first efficient 2D/3D sliding-window-like attention implementation, achieving 58.79% MFU. Precisely, STA accelerates attention by 2.8-17x over FlashAttention-2 (FA2) and 1.6-10x over FlashAttention-3 (FA3). On the leading video DiT, HunyuanVideo, STA reduces end-to-end latency from 945s (FA3) to 685s without quality degradation, requiring no training. Enabling finetuning further lowers latency to 268s with only a 0.09% drop on VBench. |
|
2025-02-10T00:00:00 | 2502.05176 | AuraFusion360: Augmented Unseen Region Alignment for Reference-based 360° Unbounded Scene Inpainting | [
"Chung-Ho Wu",
"Yang-Jung Chen",
"Ying-Huan Chen",
"Jie-Ying Lee",
"Bo-Hsu Ke",
"Chun-Wei Tuan Mu",
"Yi-Chuan Huang",
"Chin-Yang Lin",
"Min-Hung Chen",
"Yen-Yu Lin",
"Yu-Lun Liu"
] | Three-dimensional scene inpainting is crucial for applications from virtual reality to architectural visualization, yet existing methods struggle with view consistency and geometric accuracy in 360{\deg} unbounded scenes. We present AuraFusion360, a novel reference-based method that enables high-quality object removal and hole filling in 3D scenes represented by Gaussian Splatting. Our approach introduces (1) depth-aware unseen mask generation for accurate occlusion identification, (2) Adaptive Guided Depth Diffusion, a zero-shot method for accurate initial point placement without requiring additional training, and (3) SDEdit-based detail enhancement for multi-view coherence. We also introduce 360-USID, the first comprehensive dataset for 360{\deg} unbounded scene inpainting with ground truth. Extensive experiments demonstrate that AuraFusion360 significantly outperforms existing methods, achieving superior perceptual quality while maintaining geometric accuracy across dramatic viewpoint changes. See our project page for video results and the dataset at https://kkennethwu.github.io/aurafusion360/. |
|
2025-02-10T00:00:00 | 2502.05163 | DuoGuard: A Two-Player RL-Driven Framework for Multilingual LLM Guardrails | [
"Yihe Deng",
"Yu Yang",
"Junkai Zhang",
"Wei Wang",
"Bo Li"
] | https://github.com/yihedeng9/DuoGuard | The rapid advancement of large language models (LLMs) has increased the need for guardrail models to ensure responsible use, particularly in detecting unsafe and illegal content. While substantial safety data exist in English, multilingual guardrail modeling remains underexplored due to the scarcity of open-source safety data in other languages. To address this gap, we propose a novel two-player Reinforcement Learning (RL) framework, where a generator and a guardrail model co-evolve adversarially to produce high-quality synthetic data for multilingual guardrail training. We theoretically formalize this interaction as a two-player game, proving convergence to a Nash equilibrium. Empirical evaluations show that our model \ours outperforms state-of-the-art models, achieving nearly 10% improvement over LlamaGuard3 (8B) on English benchmarks while being 4.5x faster at inference with a significantly smaller model (0.5B). We achieve substantial advancements in multilingual safety tasks, particularly in addressing the imbalance for lower-resource languages in a collected real dataset. Ablation studies emphasize the critical role of synthetic data generation in bridging the imbalance in open-source data between English and other languages. These findings establish a scalable and efficient approach to synthetic data generation, paving the way for improved multilingual guardrail models to enhance LLM safety. Code, model, and data will be open-sourced at https://github.com/yihedeng9/DuoGuard. |
2025-02-10T00:00:00 | 2502.05179 | FlashVideo:Flowing Fidelity to Detail for Efficient High-Resolution Video Generation | [
"Shilong Zhang",
"Wenbo Li",
"Shoufa Chen",
"Chongjian Ge",
"Peize Sun",
"Yida Zhang",
"Yi Jiang",
"Zehuan Yuan",
"Binyue Peng",
"Ping Luo"
] | DiT diffusion models have achieved great success in text-to-video generation, leveraging their scalability in model capacity and data scale. High content and motion fidelity aligned with text prompts, however, often require large model parameters and a substantial number of function evaluations (NFEs). Realistic and visually appealing details are typically reflected in high resolution outputs, further amplifying computational demands especially for single stage DiT models. To address these challenges, we propose a novel two stage framework, FlashVideo, which strategically allocates model capacity and NFEs across stages to balance generation fidelity and quality. In the first stage, prompt fidelity is prioritized through a low resolution generation process utilizing large parameters and sufficient NFEs to enhance computational efficiency. The second stage establishes flow matching between low and high resolutions, effectively generating fine details with minimal NFEs. Quantitative and visual results demonstrate that FlashVideo achieves state-of-the-art high resolution video generation with superior computational efficiency. Additionally, the two-stage design enables users to preview the initial output before committing to full resolution generation, thereby significantly reducing computational costs and wait times as well as enhancing commercial viability . |
|
2025-02-10T00:00:00 | 2502.05178 | QLIP: Text-Aligned Visual Tokenization Unifies Auto-Regressive Multimodal Understanding and Generation | [
"Yue Zhao",
"Fuzhao Xue",
"Scott Reed",
"Linxi Fan",
"Yuke Zhu",
"Jan Kautz",
"Zhiding Yu",
"Philipp Krähenbühl",
"De-An Huang"
] | We introduce Quantized Language-Image Pretraining (QLIP), a visual tokenization method that combines state-of-the-art reconstruction quality with state-of-the-art zero-shot image understanding. QLIP trains a binary-spherical-quantization-based autoencoder with reconstruction and language-image alignment objectives. We are the first to show that the two objectives do not need to be at odds. We balance the two loss terms dynamically during training and show that a two-stage training pipeline effectively mixes the large-batch requirements of image-language pre-training with the memory bottleneck imposed by the reconstruction objective. We validate the effectiveness of QLIP for multimodal understanding and text-conditioned image generation with a single model. Specifically, QLIP serves as a drop-in replacement for the visual encoder for LLaVA and the image tokenizer for LlamaGen with comparable or even better performance. Finally, we demonstrate that QLIP enables a unified mixed-modality auto-regressive model for understanding and generation. |
|
2025-02-10T00:00:00 | 2502.04376 | MEETING DELEGATE: Benchmarking LLMs on Attending Meetings on Our Behalf | [
"Lingxiang Hu",
"Shurun Yuan",
"Xiaoting Qin",
"Jue Zhang",
"Qingwei Lin",
"Dongmei Zhang",
"Saravan Rajmohan",
"Qi Zhang"
] | In contemporary workplaces, meetings are essential for exchanging ideas and ensuring team alignment but often face challenges such as time consumption, scheduling conflicts, and inefficient participation. Recent advancements in Large Language Models (LLMs) have demonstrated their strong capabilities in natural language generation and reasoning, prompting the question: can LLMs effectively delegate participants in meetings? To explore this, we develop a prototype LLM-powered meeting delegate system and create a comprehensive benchmark using real meeting transcripts. Our evaluation reveals that GPT-4/4o maintain balanced performance between active and cautious engagement strategies. In contrast, Gemini 1.5 Pro tends to be more cautious, while Gemini 1.5 Flash and Llama3-8B/70B display more active tendencies. Overall, about 60\% of responses address at least one key point from the ground-truth. However, improvements are needed to reduce irrelevant or repetitive content and enhance tolerance for transcription errors commonly found in real-world settings. Additionally, we implement the system in practical settings and collect real-world feedback from demos. Our findings underscore the potential and challenges of utilizing LLMs as meeting delegates, offering valuable insights into their practical application for alleviating the burden of meetings. |
|
2025-02-10T00:00:00 | 2502.03512 | YINYANG-ALIGN: Benchmarking Contradictory Objectives and Proposing Multi-Objective Optimization based DPO for Text-to-Image Alignment | [
"Amitava Das",
"Yaswanth Narsupalli",
"Gurpreet Singh",
"Vinija Jain",
"Vasu Sharma",
"Suranjana Trivedy",
"Aman Chadha",
"Amit Sheth"
] | Precise alignment in Text-to-Image (T2I) systems is crucial to ensure that generated visuals not only accurately encapsulate user intents but also conform to stringent ethical and aesthetic benchmarks. Incidents like the Google Gemini fiasco, where misaligned outputs triggered significant public backlash, underscore the critical need for robust alignment mechanisms. In contrast, Large Language Models (LLMs) have achieved notable success in alignment. Building on these advancements, researchers are eager to apply similar alignment techniques, such as Direct Preference Optimization (DPO), to T2I systems to enhance image generation fidelity and reliability. We present YinYangAlign, an advanced benchmarking framework that systematically quantifies the alignment fidelity of T2I systems, addressing six fundamental and inherently contradictory design objectives. Each pair represents fundamental tensions in image generation, such as balancing adherence to user prompts with creative modifications or maintaining diversity alongside visual coherence. YinYangAlign includes detailed axiom datasets featuring human prompts, aligned (chosen) responses, misaligned (rejected) AI-generated outputs, and explanations of the underlying contradictions. |
|
2025-02-10T00:00:00 | 2502.03738 | Scaling Laws in Patchification: An Image Is Worth 50,176 Tokens And More | [
"Feng Wang",
"Yaodong Yu",
"Guoyizhe Wei",
"Wei Shao",
"Yuyin Zhou",
"Alan Yuille",
"Cihang Xie"
] | https://github.com/wangf3014/Patch_Scaling | Since the introduction of Vision Transformer (ViT), patchification has long been regarded as a de facto image tokenization approach for plain visual architectures. By compressing the spatial size of images, this approach can effectively shorten the token sequence and reduce the computational cost of ViT-like plain architectures. In this work, we aim to thoroughly examine the information loss caused by this patchification-based compressive encoding paradigm and how it affects visual understanding. We conduct extensive patch size scaling experiments and excitedly observe an intriguing scaling law in patchification: the models can consistently benefit from decreased patch sizes and attain improved predictive performance, until it reaches the minimum patch size of 1x1, i.e., pixel tokenization. This conclusion is broadly applicable across different vision tasks, various input scales, and diverse architectures such as ViT and the recent Mamba models. Moreover, as a by-product, we discover that with smaller patches, task-specific decoder heads become less critical for dense prediction. In the experiments, we successfully scale up the visual sequence to an exceptional length of 50,176 tokens, achieving a competitive test accuracy of 84.6% with a base-sized model on the ImageNet-1k benchmark. We hope this study can provide insights and theoretical foundations for future works of building non-compressive vision models. Code is available at https://github.com/wangf3014/Patch_Scaling. |
2025-02-10T00:00:00 | 2502.05003 | QuEST: Stable Training of LLMs with 1-Bit Weights and Activations | [
"Andrei Panferov",
"Jiale Chen",
"Soroush Tabesh",
"Roberto L. Castro",
"Mahdi Nikdan",
"Dan Alistarh"
] | https://github.com/IST-DASLab/QuEST | One approach to reducing the massive costs of large language models (LLMs) is the use of quantized or sparse representations for training or deployment. While post-training compression methods are very popular, the question of obtaining even more accurate compressed models by directly training over such representations, i.e., Quantization-Aware Training (QAT), is still open: for example, a recent study (arXiv:2411.04330v2) put the "optimal" bit-width at which models can be trained using QAT, while staying accuracy-competitive with standard FP16/BF16 precision, at 8-bits weights and activations. We advance this state-of-the-art via a new method called QuEST, which is Pareto-competitive with FP16, i.e., it provides better accuracy at lower model size, while training models with weights and activations in 4-bits or less. Moreover, QuEST allows stable training with 1-bit weights and activations. QuEST achieves this by improving two key aspects of QAT methods: (1) accurate and fast quantization of the (continuous) distributions of weights and activations via Hadamard normalization and MSE-optimal fitting; (2) a new trust gradient estimator based on the idea of explicitly minimizing the error between the noisy gradient computed over quantized states and the "true" (but unknown) full-precision gradient. Experiments on Llama-type architectures show that QuEST induces stable scaling laws across the entire range of hardware-supported precisions, and can be extended to sparse representations. We provide GPU kernel support showing that models produced by QuEST can be executed efficiently. Our code is available at https://github.com/IST-DASLab/QuEST. |
2025-02-10T00:00:00 | 2502.04689 | ARR: Question Answering with Large Language Models via Analyzing, Retrieving, and Reasoning | [
"Yuwei Yin",
"Giuseppe Carenini"
] | Large language models (LLMs) achieve remarkable performance on challenging benchmarks that are often structured as multiple-choice question-answering (QA) tasks. Zero-shot Chain-of-Thought (CoT) prompting enhances reasoning in LLMs but provides only vague and generic guidance ("think step by step"). This paper introduces ARR, an intuitive and effective zero-shot prompting method that explicitly incorporates three key steps in QA solving: analyzing the intent of the question, retrieving relevant information, and reasoning step by step. Comprehensive experiments across diverse and challenging QA tasks demonstrate that ARR consistently improves the Baseline (without ARR prompting) and outperforms CoT. Ablation and case studies further validate the positive contributions of each component: analyzing, retrieving, and reasoning. Notably, intent analysis plays a vital role in ARR. Additionally, extensive evaluations across various model sizes, LLM series, and generation settings solidify the effectiveness, robustness, and generalizability of ARR. |
|
2025-02-10T00:00:00 | 2502.04416 | CMoE: Fast Carving of Mixture-of-Experts for Efficient LLM Inference | [
"Zehua Pei",
"Lancheng Zou",
"Hui-Ling Zhen",
"Xianzhi Yu",
"Wulong Liu",
"Sinno Jialin Pan",
"Mingxuan Yuan",
"Bei Yu"
] | https://github.com/JarvisPei/CMoE | Large language models (LLMs) achieve impressive performance by scaling model parameters, but this comes with significant inference overhead. Feed-forward networks (FFNs), which dominate LLM parameters, exhibit high activation sparsity in hidden neurons. To exploit this, researchers have proposed using a mixture-of-experts (MoE) architecture, where only a subset of parameters is activated. However, existing approaches often require extensive training data and resources, limiting their practicality. We propose CMoE (Carved MoE), a novel framework to efficiently carve MoE models from dense models. CMoE achieves remarkable performance through efficient expert grouping and lightweight adaptation. First, neurons are grouped into shared and routed experts based on activation rates. Next, we construct a routing mechanism without training from scratch, incorporating a differentiable routing process and load balancing. Using modest data, CMoE produces a well-designed, usable MoE from a 7B dense model within five minutes. With lightweight fine-tuning, it achieves high-performance recovery in under an hour. We make our code publicly available at https://github.com/JarvisPei/CMoE. |
2025-02-10T00:00:00 | 2502.04959 | No Task Left Behind: Isotropic Model Merging with Common and Task-Specific Subspaces | [
"Daniel Marczak",
"Simone Magistri",
"Sebastian Cygert",
"Bartłomiej Twardowski",
"Andrew D. Bagdanov",
"Joost van de Weijer"
] | https://github.com/danielm1405/iso-merging | Model merging integrates the weights of multiple task-specific models into a single multi-task model. Despite recent interest in the problem, a significant performance gap between the combined and single-task models remains. In this paper, we investigate the key characteristics of task matrices -- weight update matrices applied to a pre-trained model -- that enable effective merging. We show that alignment between singular components of task-specific and merged matrices strongly correlates with performance improvement over the pre-trained model. Based on this, we propose an isotropic merging framework that flattens the singular value spectrum of task matrices, enhances alignment, and reduces the performance gap. Additionally, we incorporate both common and task-specific subspaces to further improve alignment and performance. Our proposed approach achieves state-of-the-art performance across multiple scenarios, including various sets of tasks and model scales. This work advances the understanding of model merging dynamics, offering an effective methodology to merge models without requiring additional training. Code is available at https://github.com/danielm1405/iso-merging . |
2025-02-10T00:00:00 | 2502.05092 | Lost in Time: Clock and Calendar Understanding Challenges in Multimodal LLMs | [
"Rohit Saxena",
"Aryo Pradipta Gema",
"Pasquale Minervini"
] | Understanding time from visual representations is a fundamental cognitive skill, yet it remains a challenge for multimodal large language models (MLLMs). In this work, we investigate the capabilities of MLLMs in interpreting time and date through analogue clocks and yearly calendars. To facilitate this, we curated a structured dataset comprising two subsets: 1) ClockQA, which comprises various types of clock styles-standard, black-dial, no-second-hand, Roman numeral, and arrow-hand clocks-paired with time related questions; and 2) CalendarQA, which consists of yearly calendar images with questions ranging from commonly known dates (e.g., Christmas, New Year's Day) to computationally derived ones (e.g., the 100th or 153rd day of the year). We aim to analyse how MLLMs can perform visual recognition, numerical reasoning, and temporal inference when presented with time-related visual data. Our evaluations show that despite recent advancements, reliably understanding time remains a significant challenge for MLLMs. |
|
2025-02-10T00:00:00 | 2502.04327 | Value-Based Deep RL Scales Predictably | [
"Oleh Rybkin",
"Michal Nauman",
"Preston Fu",
"Charlie Snell",
"Pieter Abbeel",
"Sergey Levine",
"Aviral Kumar"
] | Scaling data and compute is critical to the success of machine learning. However, scaling demands predictability: we want methods to not only perform well with more compute or data, but also have their performance be predictable from small-scale runs, without running the large-scale experiment. In this paper, we show that value-based off-policy RL methods are predictable despite community lore regarding their pathological behavior. First, we show that data and compute requirements to attain a given performance level lie on a Pareto frontier, controlled by the updates-to-data (UTD) ratio. By estimating this frontier, we can predict this data requirement when given more compute, and this compute requirement when given more data. Second, we determine the optimal allocation of a total resource budget across data and compute for a given performance and use it to determine hyperparameters that maximize performance for a given budget. Third, this scaling behavior is enabled by first estimating predictable relationships between hyperparameters, which is used to manage effects of overfitting and plasticity loss unique to RL. We validate our approach using three algorithms: SAC, BRO, and PQL on DeepMind Control, OpenAI gym, and IsaacGym, when extrapolating to higher levels of data, compute, budget, or performance. |
|
2025-02-10T00:00:00 | 2501.12387 | Continuous 3D Perception Model with Persistent State | [
"Qianqian Wang",
"Yifei Zhang",
"Aleksander Holynski",
"Alexei A. Efros",
"Angjoo Kanazawa"
] | We present a unified framework capable of solving a broad range of 3D tasks. Our approach features a stateful recurrent model that continuously updates its state representation with each new observation. Given a stream of images, this evolving state can be used to generate metric-scale pointmaps (per-pixel 3D points) for each new input in an online fashion. These pointmaps reside within a common coordinate system, and can be accumulated into a coherent, dense scene reconstruction that updates as new images arrive. Our model, called CUT3R (Continuous Updating Transformer for 3D Reconstruction), captures rich priors of real-world scenes: not only can it predict accurate pointmaps from image observations, but it can also infer unseen regions of the scene by probing at virtual, unobserved views. Our method is simple yet highly flexible, naturally accepting varying lengths of images that may be either video streams or unordered photo collections, containing both static and dynamic content. We evaluate our method on various 3D/4D tasks and demonstrate competitive or state-of-the-art performance in each. Project Page: https://cut3r.github.io/ |
|
2025-02-10T00:00:00 | 2502.03771 | Adaptive Semantic Prompt Caching with VectorQ | [
"Luis Gaspar Schroeder",
"Shu Liu",
"Alejandro Cuadron",
"Mark Zhao",
"Stephan Krusche",
"Alfons Kemper",
"Matei Zaharia",
"Joseph E. Gonzalez"
] | Semantic prompt caches reduce the latency and cost of large language model (LLM) inference by reusing cached LLM-generated responses for semantically similar prompts. Vector similarity metrics assign a numerical score to quantify the similarity between an embedded prompt and its nearest neighbor in the cache. Existing systems rely on a static threshold to classify whether the similarity score is sufficiently high to result in a cache hit. We show that this one-size-fits-all threshold is insufficient across different prompts. We propose VectorQ, a framework to learn embedding-specific threshold regions that adapt to the complexity and uncertainty of an embedding. Through evaluations on a combination of four diverse datasets, we show that VectorQ consistently outperforms state-of-the-art systems across all static thresholds, achieving up to 12x increases in cache hit rate and error rate reductions up to 92%. |
|
2025-02-10T00:00:00 | 2502.02909 | SPARC: Subspace-Aware Prompt Adaptation for Robust Continual Learning in LLMs | [
"Dinithi Jayasuriya",
"Sina Tayebati",
"Davide Ettori",
"Ranganath Krishnan",
"Amit Ranjan Trivedi"
] | We propose SPARC, a lightweight continual learning framework for large language models (LLMs) that enables efficient task adaptation through prompt tuning in a lower-dimensional space. By leveraging principal component analysis (PCA), we identify a compact subspace of the training data. Optimizing prompts in this lower-dimensional space enhances training efficiency, as it focuses updates on the most relevant features while reducing computational overhead. Furthermore, since the model's internal structure remains unaltered, the extensive knowledge gained from pretraining is fully preserved, ensuring that previously learned information is not compromised during adaptation. Our method achieves high knowledge retention in both task-incremental and domain-incremental continual learning setups while fine-tuning only 0.04% of the model's parameters. Additionally, by integrating LoRA, we enhance adaptability to computational constraints, allowing for a tradeoff between accuracy and training cost. Experiments on the SuperGLUE benchmark demonstrate that our PCA-based prompt tuning combined with LoRA maintains full knowledge retention while improving accuracy, utilizing only 1% of the model's parameters. These results establish our approach as a scalable and resource-efficient solution for continual learning in LLMs. |
|
2025-02-10T00:00:00 | 2502.02692 | Intelligent Sensing-to-Action for Robust Autonomy at the Edge: Opportunities and Challenges | [
"Amit Ranjan Trivedi",
"Sina Tayebati",
"Hemant Kumawat",
"Nastaran Darabi",
"Divake Kumar",
"Adarsh Kumar Kosta",
"Yeshwanth Venkatesha",
"Dinithi Jayasuriya",
"Nethmi Jayasinghe",
"Priyadarshini Panda",
"Saibal Mukhopadhyay",
"Kaushik Roy"
] | Autonomous edge computing in robotics, smart cities, and autonomous vehicles relies on the seamless integration of sensing, processing, and actuation for real-time decision-making in dynamic environments. At its core is the sensing-to-action loop, which iteratively aligns sensor inputs with computational models to drive adaptive control strategies. These loops can adapt to hyper-local conditions, enhancing resource efficiency and responsiveness, but also face challenges such as resource constraints, synchronization delays in multi-modal data fusion, and the risk of cascading errors in feedback loops. This article explores how proactive, context-aware sensing-to-action and action-to-sensing adaptations can enhance efficiency by dynamically adjusting sensing and computation based on task demands, such as sensing a very limited part of the environment and predicting the rest. By guiding sensing through control actions, action-to-sensing pathways can improve task relevance and resource use, but they also require robust monitoring to prevent cascading errors and maintain reliability. Multi-agent sensing-action loops further extend these capabilities through coordinated sensing and actions across distributed agents, optimizing resource use via collaboration. Additionally, neuromorphic computing, inspired by biological systems, provides an efficient framework for spike-based, event-driven processing that conserves energy, reduces latency, and supports hierarchical control--making it ideal for multi-agent optimization. This article highlights the importance of end-to-end co-design strategies that align algorithmic models with hardware and environmental dynamics and improve cross-layer interdependencies to improve throughput, precision, and adaptability for energy-efficient edge autonomy in complex environments. |
|
2025-02-11T00:00:00 | 2502.03628 | The Hidden Life of Tokens: Reducing Hallucination of Large Vision-Language Models via Visual Information Steering | [
"Zhuowei Li",
"Haizhou Shi",
"Yunhe Gao",
"Di Liu",
"Zhenting Wang",
"Yuxiao Chen",
"Ting Liu",
"Long Zhao",
"Hao Wang",
"Dimitris N. Metaxas"
] | Large Vision-Language Models (LVLMs) can reason effectively over both textual and visual inputs, but they tend to hallucinate syntactically coherent yet visually ungrounded contents. In this paper, we investigate the internal dynamics of hallucination by examining the tokens logits rankings throughout the generation process, revealing three key patterns in how LVLMs process information: (1) gradual visual information loss -- visually grounded tokens gradually become less favored throughout generation, and (2) early excitation -- semantically meaningful tokens achieve peak activation in the layers earlier than the final layer. (3) hidden genuine information -- visually grounded tokens though not being eventually decided still retain relatively high rankings at inference. Based on these insights, we propose VISTA (Visual Information Steering with Token-logit Augmentation), a training-free inference-time intervention framework that reduces hallucination while promoting genuine information. VISTA works by combining two complementary approaches: reinforcing visual information in activation space and leveraging early layer activations to promote semantically meaningful decoding. Compared to existing methods, VISTA requires no external supervision and is applicable to various decoding strategies. Extensive experiments show that VISTA on average reduces hallucination by abount 40% on evaluated open-ended generation task, and it consistently outperforms existing methods on four benchmarks across four architectures under three decoding strategies. |
|
2025-02-11T00:00:00 | 2502.06635 | Steel-LLM:From Scratch to Open Source -- A Personal Journey in Building a Chinese-Centric LLM | [
"Qingshui Gu",
"Shu Li",
"Tianyu Zheng",
"Zhaoxiang Zhang"
] | https://github.com/zhanshijinwat/Steel-LLM | Steel-LLM is a Chinese-centric language model developed from scratch with the goal of creating a high-quality, open-source model despite limited computational resources. Launched in March 2024, the project aimed to train a 1-billion-parameter model on a large-scale dataset, prioritizing transparency and the sharing of practical insights to assist others in the community. The training process primarily focused on Chinese data, with a small proportion of English data included, addressing gaps in existing open-source LLMs by providing a more detailed and practical account of the model-building journey. Steel-LLM has demonstrated competitive performance on benchmarks such as CEVAL and CMMLU, outperforming early models from larger institutions. This paper provides a comprehensive summary of the project's key contributions, including data collection, model design, training methodologies, and the challenges encountered along the way, offering a valuable resource for researchers and practitioners looking to develop their own LLMs. The model checkpoints and training script are available at https://github.com/zhanshijinwat/Steel-LLM. |
2025-02-11T00:00:00 | 2502.06155 | Efficient-vDiT: Efficient Video Diffusion Transformers With Attention Tile | [
"Hangliang Ding",
"Dacheng Li",
"Runlong Su",
"Peiyuan Zhang",
"Zhijie Deng",
"Ion Stoica",
"Hao Zhang"
] | Despite the promise of synthesizing high-fidelity videos, Diffusion Transformers (DiTs) with 3D full attention suffer from expensive inference due to the complexity of attention computation and numerous sampling steps. For example, the popular Open-Sora-Plan model consumes more than 9 minutes for generating a single video of 29 frames. This paper addresses the inefficiency issue from two aspects: 1) Prune the 3D full attention based on the redundancy within video data; We identify a prevalent tile-style repetitive pattern in the 3D attention maps for video data, and advocate a new family of sparse 3D attention that holds a linear complexity w.r.t. the number of video frames. 2) Shorten the sampling process by adopting existing multi-step consistency distillation; We split the entire sampling trajectory into several segments and perform consistency distillation within each one to activate few-step generation capacities. We further devise a three-stage training pipeline to conjoin the low-complexity attention and few-step generation capacities. Notably, with 0.1% pretraining data, we turn the Open-Sora-Plan-1.2 model into an efficient one that is 7.4x -7.8x faster for 29 and 93 frames 720p video generation with a marginal performance trade-off in VBench. In addition, we demonstrate that our approach is amenable to distributed inference, achieving an additional 3.91x speedup when running on 4 GPUs with sequence parallelism. |
|
2025-02-11T00:00:00 | 2502.06772 | ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates | [
"Ling Yang",
"Zhaochen Yu",
"Bin Cui",
"Mengdi Wang"
] | https://github.com/Gen-Verse/ReasonFlux | We present that hierarchical LLM reasoning via scaling thought templates can effectively optimize the reasoning search space and outperform the mathematical reasoning capabilities of powerful LLMs like OpenAI o1-preview and DeepSeek V3. We train our ReasonFlux-32B model with only 8 GPUs and introduces three innovations: (i) a structured and generic thought template library, containing around 500 high-level thought templates capable of generalizing to similar or relevant reasoning problems; (ii) performing hierarchical reinforcement learning on a sequence of thought templates instead of long CoTs, optimizing a base LLM to plan out an optimal template trajectory for gradually handling complex problems; (iii) a brand new inference scaling system that enables hierarchical LLM reasoning by adaptively scaling thought templates at inference time. With a template trajectory containing sequential thought templates, our ReasonFlux-32B significantly advances math reasoning capabilities to state-of-the-art levels. Notably, on the MATH benchmark, it achieves an accuracy of 91.2% and surpasses o1-preview by 6.7%. On the USA Math Olympiad (AIME) benchmark, ReasonFlux-32B solves an average of 56.7% of problems, surpassing o1-preview and DeepSeek-V3 by 27% and 45%, respectively. Code: https://github.com/Gen-Verse/ReasonFlux |
2025-02-11T00:00:00 | 2502.06788 | EVEv2: Improved Baselines for Encoder-Free Vision-Language Models | [
"Haiwen Diao",
"Xiaotong Li",
"Yufeng Cui",
"Yueze Wang",
"Haoge Deng",
"Ting Pan",
"Wenxuan Wang",
"Huchuan Lu",
"Xinlong Wang"
] | https://github.com/baaivision/EVE | Existing encoder-free vision-language models (VLMs) are rapidly narrowing the performance gap with their encoder-based counterparts, highlighting the promising potential for unified multimodal systems with structural simplicity and efficient deployment. We systematically clarify the performance gap between VLMs using pre-trained vision encoders, discrete tokenizers, and minimalist visual layers from scratch, deeply excavating the under-examined characteristics of encoder-free VLMs. We develop efficient strategies for encoder-free VLMs that rival mainstream encoder-based ones. After an in-depth investigation, we launch EVEv2.0, a new and improved family of encoder-free VLMs. We show that: (i) Properly decomposing and hierarchically associating vision and language within a unified model reduces interference between modalities. (ii) A well-designed training strategy enables effective optimization for encoder-free VLMs. Through extensive evaluation, our EVEv2.0 represents a thorough study for developing a decoder-only architecture across modalities, demonstrating superior data efficiency and strong vision-reasoning capability. Code is publicly available at: https://github.com/baaivision/EVE. |
2025-02-11T00:00:00 | 2502.06049 | LM2: Large Memory Models | [
"Jikun Kang",
"Wenqi Wu",
"Filippos Christianos",
"Alex J. Chan",
"Fraser Greenlee",
"George Thomas",
"Marvin Purtorab",
"Andy Toulis"
] | This paper introduces the Large Memory Model (LM2), a decoder-only Transformer architecture enhanced with an auxiliary memory module that aims to address the limitations of standard Transformers in multi-step reasoning, relational argumentation, and synthesizing information distributed over long contexts. The proposed LM2 incorporates a memory module that acts as a contextual representation repository, interacting with input tokens via cross attention and updating through gating mechanisms. To preserve the Transformers general-purpose capabilities, LM2 maintains the original information flow while integrating a complementary memory pathway. Experimental results on the BABILong benchmark demonstrate that the LM2model outperforms both the memory-augmented RMT model by 37.1% and the baseline Llama-3.2 model by 86.3% on average across tasks. LM2 exhibits exceptional capabilities in multi-hop inference, numerical reasoning, and large-context question-answering. On the MMLU dataset, it achieves a 5.0% improvement over a pre-trained vanilla model, demonstrating that its memory module does not degrade performance on general tasks. Further, in our analysis, we explore the memory interpretability, effectiveness of memory modules, and test-time behavior. Our findings emphasize the importance of explicit memory in enhancing Transformer architectures. |
|
2025-02-11T00:00:00 | 2502.06023 | Dual Caption Preference Optimization for Diffusion Models | [
"Amir Saeidi",
"Yiran Luo",
"Agneet Chatterjee",
"Shamanthak Hegde",
"Bimsara Pathiraja",
"Yezhou Yang",
"Chitta Baral"
] | Recent advancements in human preference optimization, originally developed for Large Language Models (LLMs), have shown significant potential in improving text-to-image diffusion models. These methods aim to learn the distribution of preferred samples while distinguishing them from less preferred ones. However, existing preference datasets often exhibit overlap between these distributions, leading to a conflict distribution. Additionally, we identified that input prompts contain irrelevant information for less preferred images, limiting the denoising network's ability to accurately predict noise in preference optimization methods, known as the irrelevant prompt issue. To address these challenges, we propose Dual Caption Preference Optimization (DCPO), a novel approach that utilizes two distinct captions to mitigate irrelevant prompts. To tackle conflict distribution, we introduce the Pick-Double Caption dataset, a modified version of Pick-a-Pic v2 with separate captions for preferred and less preferred images. We further propose three different strategies for generating distinct captions: captioning, perturbation, and hybrid methods. Our experiments show that DCPO significantly improves image quality and relevance to prompts, outperforming Stable Diffusion (SD) 2.1, SFT_Chosen, Diffusion-DPO, and MaPO across multiple metrics, including Pickscore, HPSv2.1, GenEval, CLIPscore, and ImageReward, fine-tuned on SD 2.1 as the backbone. |
|
2025-02-11T00:00:00 | 2502.05431 | APE: Faster and Longer Context-Augmented Generation via Adaptive Parallel Encoding | [
"Xinyu Yang",
"Tianqi Chen",
"Beidi Chen"
] | Context-augmented generation (CAG) techniques, including RAG and ICL, require the efficient combination of multiple contexts to generate responses to user queries. Directly inputting these contexts as a sequence introduces a considerable computational burden by re-encoding the combined selection of contexts for every request. To address this, we explore the promising potential of parallel encoding to independently pre-compute and cache each context's KV states. This approach enables the direct loading of cached states during inference while accommodating more contexts through position reuse across contexts. However, due to misalignments in attention distribution, directly applying parallel encoding results in a significant performance drop. To enable effective and efficient CAG, we propose Adaptive Parallel Encoding (APE), which brings shared prefix, attention temperature, and scaling factor to align the distribution of parallel encoding with sequential encoding. Results on RAG and ICL tasks demonstrate that APE can preserve 98% and 93% sequential encoding performance using the same inputs while outperforming parallel encoding by 3.6% and 7.9%, respectively. It also scales to many-shot CAG, effectively encoding hundreds of contexts in parallel. Efficiency evaluation shows that APE can achieve an end-to-end 4.5times speedup by reducing 28times prefilling time for a 128K-length context. |
|
2025-02-11T00:00:00 | 2502.06781 | Exploring the Limit of Outcome Reward for Learning Mathematical Reasoning | [
"Chengqi Lyu",
"Songyang Gao",
"Yuzhe Gu",
"Wenwei Zhang",
"Jianfei Gao",
"Kuikun Liu",
"Ziyi Wang",
"Shuaibin Li",
"Qian Zhao",
"Haian Huang",
"Weihan Cao",
"Jiangning Liu",
"Hongwei Liu",
"Junnan Liu",
"Songyang Zhang",
"Dahua Lin",
"Kai Chen"
] | https://github.com/InternLM/OREAL | Reasoning abilities, especially those for solving complex math problems, are crucial components of general intelligence. Recent advances by proprietary companies, such as o-series models of OpenAI, have made remarkable progress on reasoning tasks. However, the complete technical details remain unrevealed, and the techniques that are believed certainly to be adopted are only reinforcement learning (RL) and the long chain of thoughts. This paper proposes a new RL framework, termed OREAL, to pursue the performance limit that can be achieved through Outcome REwArd-based reinforcement Learning for mathematical reasoning tasks, where only binary outcome rewards are easily accessible. We theoretically prove that behavior cloning on positive trajectories from best-of-N (BoN) sampling is sufficient to learn the KL-regularized optimal policy in binary feedback environments. This formulation further implies that the rewards of negative samples should be reshaped to ensure the gradient consistency between positive and negative samples. To alleviate the long-existing difficulties brought by sparse rewards in RL, which are even exacerbated by the partial correctness of the long chain of thought for reasoning tasks, we further apply a token-level reward model to sample important tokens in reasoning trajectories for learning. With OREAL, for the first time, a 7B model can obtain 94.0 pass@1 accuracy on MATH-500 through RL, being on par with 32B models. OREAL-32B also surpasses previous 32B models trained by distillation with 95.0 pass@1 accuracy on MATH-500. Our investigation also indicates the importance of initial policy models and training queries for RL. Code, models, and data will be released to benefit future researchhttps://github.com/InternLM/OREAL. |
2025-02-11T00:00:00 | 2502.05609 | Lossless Acceleration of Large Language Models with Hierarchical Drafting based on Temporal Locality in Speculative Decoding | [
"Sukmin Cho",
"Sangjin Choi",
"Taeho Hwang",
"Jeongyeon Seo",
"Soyeong Jeong",
"Huije Lee",
"Hoyun Song",
"Jong C. Park",
"Youngjin Kwon"
] | Accelerating inference in Large Language Models (LLMs) is critical for real-time interactions, as they have been widely incorporated into real-world services. Speculative decoding, a fully algorithmic solution, has gained attention for improving inference speed by drafting and verifying tokens, thereby generating multiple tokens in a single forward pass. However, current drafting strategies usually require significant fine-tuning or have inconsistent performance across tasks. To address these challenges, we propose Hierarchy Drafting (HD), a novel lossless drafting approach that organizes various token sources into multiple databases in a hierarchical framework based on temporal locality. In the drafting step, HD sequentially accesses multiple databases to obtain draft tokens from the highest to the lowest locality, ensuring consistent acceleration across diverse tasks and minimizing drafting latency. Our experiments on Spec-Bench using LLMs with 7B and 13B parameters demonstrate that HD outperforms existing database drafting methods, achieving robust inference speedups across model sizes, tasks, and temperatures. |
|
2025-02-11T00:00:00 | 2502.06703 | Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling | [
"Runze Liu",
"Junqi Gao",
"Jian Zhao",
"Kaiyan Zhang",
"Xiu Li",
"Biqing Qi",
"Wanli Ouyang",
"Bowen Zhou"
] | Test-Time Scaling (TTS) is an important method for improving the performance of Large Language Models (LLMs) by using additional computation during the inference phase. However, current studies do not systematically analyze how policy models, Process Reward Models (PRMs), and problem difficulty influence TTS. This lack of analysis limits the understanding and practical use of TTS methods. In this paper, we focus on two core questions: (1) What is the optimal approach to scale test-time computation across different policy models, PRMs, and problem difficulty levels? (2) To what extent can extended computation improve the performance of LLMs on complex tasks, and can smaller language models outperform larger ones through this approach? Through comprehensive experiments on MATH-500 and challenging AIME24 tasks, we have the following observations: (1) The compute-optimal TTS strategy is highly dependent on the choice of policy model, PRM, and problem difficulty. (2) With our compute-optimal TTS strategy, extremely small policy models can outperform larger models. For example, a 1B LLM can exceed a 405B LLM on MATH-500. Moreover, on both MATH-500 and AIME24, a 0.5B LLM outperforms GPT-4o, a 3B LLM surpasses a 405B LLM, and a 7B LLM beats o1 and DeepSeek-R1, while with higher inference efficiency. These findings show the significance of adapting TTS strategies to the specific characteristics of each task and model and indicate that TTS is a promising approach for enhancing the reasoning abilities of LLMs. |
|
2025-02-11T00:00:00 | 2502.06527 | CustomVideoX: 3D Reference Attention Driven Dynamic Adaptation for Zero-Shot Customized Video Diffusion Transformers | [
"D. She",
"Mushui Liu",
"Jingxuan Pang",
"Jin Wang",
"Zhen Yang",
"Wanggui He",
"Guanghao Zhang",
"Yi Wang",
"Qihan Huang",
"Haobin Tang",
"Yunlong Yu",
"Siming Fu"
] | Customized generation has achieved significant progress in image synthesis, yet personalized video generation remains challenging due to temporal inconsistencies and quality degradation. In this paper, we introduce CustomVideoX, an innovative framework leveraging the video diffusion transformer for personalized video generation from a reference image. CustomVideoX capitalizes on pre-trained video networks by exclusively training the LoRA parameters to extract reference features, ensuring both efficiency and adaptability. To facilitate seamless interaction between the reference image and video content, we propose 3D Reference Attention, which enables direct and simultaneous engagement of reference image features with all video frames across spatial and temporal dimensions. To mitigate the excessive influence of reference image features and textual guidance on generated video content during inference, we implement the Time-Aware Reference Attention Bias (TAB) strategy, dynamically modulating reference bias over different time steps. Additionally, we introduce the Entity Region-Aware Enhancement (ERAE) module, aligning highly activated regions of key entity tokens with reference feature injection by adjusting attention bias. To thoroughly evaluate personalized video generation, we establish a new benchmark, VideoBench, comprising over 50 objects and 100 prompts for extensive assessment. Experimental results show that CustomVideoX significantly outperforms existing methods in terms of video consistency and quality. |
|
2025-02-11T00:00:00 | 2502.06782 | Lumina-Video: Efficient and Flexible Video Generation with Multi-scale Next-DiT | [
"Dongyang Liu",
"Shicheng Li",
"Yutong Liu",
"Zhen Li",
"Kai Wang",
"Xinyue Li",
"Qi Qin",
"Yufei Liu",
"Yi Xin",
"Zhongyu Li",
"Bin Fu",
"Chenyang Si",
"Yuewen Cao",
"Conghui He",
"Ziwei Liu",
"Yu Qiao",
"Qibin Hou",
"Hongsheng Li",
"Peng Gao"
] | Recent advancements have established Diffusion Transformers (DiTs) as a dominant framework in generative modeling. Building on this success, Lumina-Next achieves exceptional performance in the generation of photorealistic images with Next-DiT. However, its potential for video generation remains largely untapped, with significant challenges in modeling the spatiotemporal complexity inherent to video data. To address this, we introduce Lumina-Video, a framework that leverages the strengths of Next-DiT while introducing tailored solutions for video synthesis. Lumina-Video incorporates a Multi-scale Next-DiT architecture, which jointly learns multiple patchifications to enhance both efficiency and flexibility. By incorporating the motion score as an explicit condition, Lumina-Video also enables direct control of generated videos' dynamic degree. Combined with a progressive training scheme with increasingly higher resolution and FPS, and a multi-source training scheme with mixed natural and synthetic data, Lumina-Video achieves remarkable aesthetic quality and motion smoothness at high training and inference efficiency. We additionally propose Lumina-V2A, a video-to-audio model based on Next-DiT, to create synchronized sounds for generated videos. Codes are released at https://www.github.com/Alpha-VLLM/Lumina-Video. |
|
2025-02-11T00:00:00 | 2502.05957 | MetaChain: A Fully-Automated and Zero-Code Framework for LLM Agents | [
"Jiabin Tang",
"Tianyu Fan",
"Chao Huang"
] | Large Language Model (LLM) Agents have demonstrated remarkable capabilities in task automation and intelligent decision-making, driving the widespread adoption of agent development frameworks such as LangChain and AutoGen. However, these frameworks predominantly serve developers with extensive technical expertise - a significant limitation considering that only 0.03 % of the global population possesses the necessary programming skills. This stark accessibility gap raises a fundamental question: Can we enable everyone, regardless of technical background, to build their own LLM agents using natural language alone? To address this challenge, we introduce MetaChain-a Fully-Automated and highly Self-Developing framework that enables users to create and deploy LLM agents through Natural Language Alone. Operating as an autonomous Agent Operating System, MetaChain comprises four key components: i) Agentic System Utilities, ii) LLM-powered Actionable Engine, iii) Self-Managing File System, and iv) Self-Play Agent Customization module. This lightweight yet powerful system enables efficient and dynamic creation and modification of tools, agents, and workflows without coding requirements or manual intervention. Beyond its code-free agent development capabilities, MetaChain also serves as a versatile multi-agent system for General AI Assistants. Comprehensive evaluations on the GAIA benchmark demonstrate MetaChain's effectiveness in generalist multi-agent tasks, surpassing existing state-of-the-art methods. Furthermore, MetaChain's Retrieval-Augmented Generation (RAG)-related capabilities have shown consistently superior performance compared to many alternative LLM-based solutions. |
|
2025-02-11T00:00:00 | 2502.06786 | Matryoshka Quantization | [
"Pranav Nair",
"Puranjay Datta",
"Jeff Dean",
"Prateek Jain",
"Aditya Kusupati"
] | Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a trade-off in model quality; int2, in particular, is known to severely degrade model quality. Consequently, practitioners are often forced to maintain multiple models with different quantization levels or serve a single model that best satisfies the quality-latency trade-off. On the other hand, integer data types, such as int8, inherently possess a nested (Matryoshka) structure where smaller bit-width integers, like int4 or int2, are nested within the most significant bits. This paper proposes Matryoshka Quantization (MatQuant), a novel multi-scale quantization technique that addresses the challenge of needing multiple quantized models. It allows training and maintaining just one model, which can then be served at different precision levels. Furthermore, due to the co-training and co-distillation regularization provided by MatQuant, the int2 precision models extracted by MatQuant can be up to 10% more accurate than standard int2 quantization (using techniques like QAT or OmniQuant). This represents significant progress in model quantization, demonstrated by the fact that, with the same recipe, an int2 FFN-quantized Gemma-2 9B model is more accurate than an int8 FFN-quantized Gemma-2 2B model. |
|
2025-02-11T00:00:00 | 2502.06764 | History-Guided Video Diffusion | [
"Kiwhan Song",
"Boyuan Chen",
"Max Simchowitz",
"Yilun Du",
"Russ Tedrake",
"Vincent Sitzmann"
] | Classifier-free guidance (CFG) is a key technique for improving conditional generation in diffusion models, enabling more accurate control while enhancing sample quality. It is natural to extend this technique to video diffusion, which generates video conditioned on a variable number of context frames, collectively referred to as history. However, we find two key challenges to guiding with variable-length history: architectures that only support fixed-size conditioning, and the empirical observation that CFG-style history dropout performs poorly. To address this, we propose the Diffusion Forcing Transformer (DFoT), a video diffusion architecture and theoretically grounded training objective that jointly enable conditioning on a flexible number of history frames. We then introduce History Guidance, a family of guidance methods uniquely enabled by DFoT. We show that its simplest form, vanilla history guidance, already significantly improves video generation quality and temporal consistency. A more advanced method, history guidance across time and frequency further enhances motion dynamics, enables compositional generalization to out-of-distribution history, and can stably roll out extremely long videos. Website: https://boyuan.space/history-guidance |
|
2025-02-11T00:00:00 | 2502.05415 | Show-o Turbo: Towards Accelerated Unified Multimodal Understanding and Generation | [
"Chenkai Xu",
"Xu Wang",
"Zhenyi Liao",
"Yishun Li",
"Tianqi Hou",
"Zhijie Deng"
] | https://github.com/zhijie-group/Show-o-Turbo | There has been increasing research interest in building unified multimodal understanding and generation models, among which Show-o stands as a notable representative, demonstrating great promise for both text-to-image and image-to-text generation. The inference of Show-o involves progressively denoising image tokens and autoregressively decoding text tokens, and hence, unfortunately, suffers from inefficiency issues from both sides. This paper introduces Show-o Turbo to bridge the gap. We first identify a unified denoising perspective for the generation of images and text in Show-o based on the parallel decoding of text tokens. We then propose to extend consistency distillation (CD), a qualified approach for shortening the denoising process of diffusion models, to the multimodal denoising trajectories of Show-o. We introduce a trajectory segmentation strategy and a curriculum learning procedure to improve the training convergence. Empirically, in text-to-image generation, Show-o Turbo displays a GenEval score of 0.625 at 4 sampling steps without using classifier-free guidance (CFG), outperforming that of the original Show-o with 8 steps and CFG; in image-to-text generation, Show-o Turbo exhibits a 1.5x speedup without significantly sacrificing performance. The code is available at https://github.com/zhijie-group/Show-o-Turbo. |
2025-02-11T00:00:00 | 2502.04370 | DreamDPO: Aligning Text-to-3D Generation with Human Preferences via Direct Preference Optimization | [
"Zhenglin Zhou",
"Xiaobo Xia",
"Fan Ma",
"Hehe Fan",
"Yi Yang",
"Tat-Seng Chua"
] | Text-to-3D generation automates 3D content creation from textual descriptions, which offers transformative potential across various fields. However, existing methods often struggle to align generated content with human preferences, limiting their applicability and flexibility. To address these limitations, in this paper, we propose DreamDPO, an optimization-based framework that integrates human preferences into the 3D generation process, through direct preference optimization. Practically, DreamDPO first constructs pairwise examples, then compare their alignment with human preferences using reward or large multimodal models, and lastly optimizes the 3D representation with a preference-driven loss function. By leveraging pairwise comparison to reflect preferences, DreamDPO reduces reliance on precise pointwise quality evaluations while enabling fine-grained controllability through preference-guided optimization. Experiments demonstrate that DreamDPO achieves competitive results, and provides higher-quality and more controllable 3D content compared to existing methods. The code and models will be open-sourced. |
|
2025-02-11T00:00:00 | 2502.06394 | SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators | [
"Daniil Moskovskiy",
"Nikita Sushko",
"Sergey Pletenev",
"Elena Tutubalina",
"Alexander Panchenko"
] | Existing approaches to multilingual text detoxification are hampered by the scarcity of parallel multilingual datasets. In this work, we introduce a pipeline for the generation of multilingual parallel detoxification data. We also introduce SynthDetoxM, a manually collected and synthetically generated multilingual parallel text detoxification dataset comprising 16,000 high-quality detoxification sentence pairs across German, French, Spanish and Russian. The data was sourced from different toxicity evaluation datasets and then rewritten with nine modern open-source LLMs in few-shot setting. Our experiments demonstrate that models trained on the produced synthetic datasets have superior performance to those trained on the human-annotated MultiParaDetox dataset even in data limited setting. Models trained on SynthDetoxM outperform all evaluated LLMs in few-shot setting. We release our dataset and code to help further research in multilingual text detoxification. |
|
2025-02-11T00:00:00 | 2502.05795 | The Curse of Depth in Large Language Models | [
"Wenfang Sun",
"Xinyuan Song",
"Pengxiang Li",
"Lu Yin",
"Yefeng Zheng",
"Shiwei Liu"
] | In this paper, we introduce the Curse of Depth, a concept that highlights, explains, and addresses the recent observation in modern Large Language Models(LLMs) where nearly half of the layers are less effective than expected. We first confirm the wide existence of this phenomenon across the most popular families of LLMs such as Llama, Mistral, DeepSeek, and Qwen. Our analysis, theoretically and empirically, identifies that the underlying reason for the ineffectiveness of deep layers in LLMs is the widespread usage of Pre-Layer Normalization (Pre-LN). While Pre-LN stabilizes the training of Transformer LLMs, its output variance exponentially grows with the model depth, which undesirably causes the derivative of the deep Transformer blocks to be an identity matrix, and therefore barely contributes to the training. To resolve this training pitfall, we propose LayerNorm Scaling, which scales the variance of output of the layer normalization inversely by the square root of its depth. This simple modification mitigates the output variance explosion of deeper Transformer layers, improving their contribution. Our experimental results, spanning model sizes from 130M to 1B, demonstrate that LayerNorm Scaling significantly enhances LLM pre-training performance compared to Pre-LN. Moreover, this improvement seamlessly carries over to supervised fine-tuning. All these gains can be attributed to the fact that LayerNorm Scaling enables deeper layers to contribute more effectively during training. |
|
2025-02-11T00:00:00 | 2502.06060 | Training Language Models for Social Deduction with Multi-Agent Reinforcement Learning | [
"Bidipta Sarkar",
"Warren Xia",
"C. Karen Liu",
"Dorsa Sadigh"
] | Communicating in natural language is a powerful tool in multi-agent settings, as it enables independent agents to share information in partially observable settings and allows zero-shot coordination with humans. However, most prior works are limited as they either rely on training with large amounts of human demonstrations or lack the ability to generate natural and useful communication strategies. In this work, we train language models to have productive discussions about their environment in natural language without any human demonstrations. We decompose the communication problem into listening and speaking. Our key idea is to leverage the agent's goal to predict useful information about the world as a dense reward signal that guides communication. Specifically, we improve a model's listening skills by training them to predict information about the environment based on discussions, and we simultaneously improve a model's speaking skills with multi-agent reinforcement learning by rewarding messages based on their influence on other agents. To investigate the role and necessity of communication in complex social settings, we study an embodied social deduction game based on Among Us, where the key question to answer is the identity of an adversarial imposter. We analyze emergent behaviors due to our technique, such as accusing suspects and providing evidence, and find that it enables strong discussions, doubling the win rates compared to standard RL. We release our code and models at https://socialdeductionllm.github.io/ |
|
2025-02-11T00:00:00 | 2502.05664 | CODESIM: Multi-Agent Code Generation and Problem Solving through Simulation-Driven Planning and Debugging | [
"Md. Ashraful Islam",
"Mohammed Eunus Ali",
"Md Rizwan Parvez"
] | Large Language Models (LLMs) have made significant strides in code generation and problem solving. Current approaches employ external tool-based iterative debuggers that use compiler or other tool-based runtime feedback to refine coarse programs generated by various methods. However, the effectiveness of these approaches heavily relies on the quality of the initial code generation, which remains an open challenge. In this paper, we introduce CodeSim, a novel multi-agent code generation framework that comprehensively addresses the stages of program synthesis-planning, coding, and debugging-through a human-like perception approach. As human verifies their understanding of any algorithms through visual simulation, CodeSim uniquely features a method of plan verification and internal debugging through the step-by-step simulation of input/output. Extensive experiments across seven challenging competitive problem-solving and program synthesis benchmarks demonstrate CodeSim's remarkable code generation capabilities. Our framework achieves new state-of-the-art (pass@1) results-(HumanEval 95.1%, MBPP 90.7%, APPS 22%, and CodeContests 29.1%). Furthermore, our method shows potential for even greater enhancement when cascaded with external debuggers. To facilitate further research and development in this area, we have open-sourced our framework in this link (https://kagnlp.github.io/codesim.github.io/). |
|
2025-02-11T00:00:00 | 2411.18676 | Embodied Red Teaming for Auditing Robotic Foundation Models | [
"Sathwik Karnik",
"Zhang-Wei Hong",
"Nishant Abhangi",
"Yen-Chen Lin",
"Tsun-Hsuan Wang",
"Christophe Dupuy",
"Rahul Gupta",
"Pulkit Agrawal"
] | Language-conditioned robot models have the potential to enable robots to perform a wide range of tasks based on natural language instructions. However, assessing their safety and effectiveness remains challenging because it is difficult to test all the different ways a single task can be phrased. Current benchmarks have two key limitations: they rely on a limited set of human-generated instructions, missing many challenging cases, and focus only on task performance without assessing safety, such as avoiding damage. To address these gaps, we introduce Embodied Red Teaming (ERT), a new evaluation method that generates diverse and challenging instructions to test these models. ERT uses automated red teaming techniques with Vision Language Models (VLMs) to create contextually grounded, difficult instructions. Experimental results show that state-of-the-art language-conditioned robot models fail or behave unsafely on ERT-generated instructions, underscoring the shortcomings of current benchmarks in evaluating real-world performance and safety. Code and videos are available at: https://s-karnik.github.io/embodied-red-team-project-page. |
|
2025-02-11T00:00:00 | 2502.06776 | Towards Internet-Scale Training For Agents | [
"Brandon Trabucco",
"Gunnar Sigurdsson",
"Robinson Piramuthu",
"Ruslan Salakhutdinov"
] | The predominant approach for training web navigation agents gathers human demonstrations for a set of popular websites and hand-written tasks, but it is becoming clear that human data are an inefficient resource. We develop a pipeline to facilitate Internet-scale training for agents without laborious human annotations. In the first stage, an LLM generates tasks for 150k diverse websites. In the next stage, LLM agents complete tasks and produce trajectories. In the final stage, an LLM reviews the trajectories and judges their success. Language models are competitive with human annotators, detecting and filtering out harmful content with an accuracy of 97%, generating feasible tasks with an 89% rate, and judging successful trajectories with an 82.6% accuracy. Scaling the pipeline, agents based on Llama 3.1 70B solve 16.7% of tasks for 150k sites. Training on the data generated by our pipeline is competitive with training on human demonstrations. In data-limited settings derived from Mind2Web and WebLINX, we improve Step Accuracy by up to +89.5% and +122.1% respectively for agents trained on mixtures of data from our pipeline, and human data. When training agents with all available human data from these benchmarks, agents fail to generalize to diverse real sites, and adding our data improves their generalization by +149.0% for WebLINX and +156.3% for Mind2Web. Code will be available at: data-for-agents.github.io. |
|
2025-02-11T00:00:00 | 2502.06282 | Jakiro: Boosting Speculative Decoding with Decoupled Multi-Head via MoE | [
"Haiduo Huang",
"Fuwei Yang",
"Zhenhua Liu",
"Yixing Xu",
"Jinze Li",
"Yang Liu",
"Xuanwu Yin",
"Dong Li",
"Pengju Ren",
"Emad Barsoum"
] | https://github.com/haiduo/Jakiro | Speculative decoding (SD) accelerates large language model inference by using a smaller draft model to predict multiple tokens, which are then verified in parallel by the larger target model. However, the limited capacity of the draft model often necessitates tree-based sampling to improve prediction accuracy, where multiple candidates are generated at each step. We identify a key limitation in this approach: the candidates at the same step are derived from the same representation, limiting diversity and reducing overall effectiveness. To address this, we propose Jakiro, leveraging Mixture of Experts (MoE), where independent experts generate diverse predictions, effectively decoupling correlations among candidates. Furthermore, we introduce a hybrid inference strategy, combining autoregressive decoding for initial tokens with parallel decoding for subsequent stages, and enhance the latter with contrastive mechanism in features to improve accuracy. Our method significantly boosts prediction accuracy and achieves higher inference speedups. Extensive experiments across diverse models validate the effectiveness and robustness of our approach, establishing a new SOTA in speculative decoding. Our codes are available at https://github.com/haiduo/Jakiro. |
2025-02-11T00:00:00 | 2502.06867 | Forbidden Science: Dual-Use AI Challenge Benchmark and Scientific Refusal Tests | [
"David Noever",
"Forrest McKee"
] | The development of robust safety benchmarks for large language models requires open, reproducible datasets that can measure both appropriate refusal of harmful content and potential over-restriction of legitimate scientific discourse. We present an open-source dataset and testing framework for evaluating LLM safety mechanisms across mainly controlled substance queries, analyzing four major models' responses to systematically varied prompts. Our results reveal distinct safety profiles: Claude-3.5-sonnet demonstrated the most conservative approach with 73% refusals and 27% allowances, while Mistral attempted to answer 100% of queries. GPT-3.5-turbo showed moderate restriction with 10% refusals and 90% allowances, and Grok-2 registered 20% refusals and 80% allowances. Testing prompt variation strategies revealed decreasing response consistency, from 85% with single prompts to 65% with five variations. This publicly available benchmark enables systematic evaluation of the critical balance between necessary safety restrictions and potential over-censorship of legitimate scientific inquiry, while providing a foundation for measuring progress in AI safety implementation. Chain-of-thought analysis reveals potential vulnerabilities in safety mechanisms, highlighting the complexity of implementing robust safeguards without unduly restricting desirable and valid scientific discourse. |
|
2025-02-12T00:00:00 | 2502.06807 | Competitive Programming with Large Reasoning Models | [
"OpenAI",
"Ahmed El-Kishky",
"Alexander Wei",
"Andre Saraiva",
"Borys Minaev",
"Daniel Selsam",
"David Dohan",
"Francis Song",
"Hunter Lightman",
"Ignasi Clavera",
"Jakub Pachocki",
"Jerry Tworek",
"Lorenz Kuhn",
"Lukasz Kaiser",
"Mark Chen",
"Max Schwarzer",
"Mostafa Rohaninejad",
"Nat McAleese",
"o3 contributors",
"Oleg Mürk",
"Rhythm Garg",
"Rui Shu",
"Szymon Sidor",
"Vineet Kosaraju",
"Wenda Zhou"
] | We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early checkpoint of o3 - with a domain-specific system, o1-ioi, which uses hand-engineered inference strategies designed for competing in the 2024 International Olympiad in Informatics (IOI). We competed live at IOI 2024 with o1-ioi and, using hand-crafted test-time strategies, placed in the 49th percentile. Under relaxed competition constraints, o1-ioi achieved a gold medal. However, when evaluating later models such as o3, we find that o3 achieves gold without hand-crafted domain-specific strategies or relaxed constraints. Our findings show that although specialized pipelines such as o1-ioi yield solid improvements, the scaled-up, general-purpose o3 model surpasses those results without relying on hand-crafted inference heuristics. Notably, o3 achieves a gold medal at the 2024 IOI and obtains a Codeforces rating on par with elite human competitors. Overall, these results indicate that scaling general-purpose reinforcement learning, rather than relying on domain-specific techniques, offers a robust path toward state-of-the-art AI in reasoning domains, such as competitive programming. |
|
2025-02-12T00:00:00 | 2502.07316 | CodeI/O: Condensing Reasoning Patterns via Code Input-Output Prediction | [
"Junlong Li",
"Daya Guo",
"Dejian Yang",
"Runxin Xu",
"Yu Wu",
"Junxian He"
] | https://github.com/hkust-nlp/CodeIO | Reasoning is a fundamental capability of Large Language Models. While prior research predominantly focuses on enhancing narrow skills like math or code generation, improving performance on many other reasoning tasks remains challenging due to sparse and fragmented training data. To address this issue, we propose CodeI/O, a novel approach that systematically condenses diverse reasoning patterns inherently embedded in contextually-grounded codes, through transforming the original code into a code input-output prediction format. By training models to predict inputs/outputs given code and test cases entirely in natural language as Chain-of-Thought (CoT) rationales, we expose them to universal reasoning primitives -- like logic flow planning, state-space searching, decision tree traversal, and modular decomposition -- while decoupling structured reasoning from code-specific syntax and preserving procedural rigor. Experimental results demonstrate CodeI/O leads to consistent improvements across symbolic, scientific, logic, math & numerical, and commonsense reasoning tasks. By matching the existing ground-truth outputs or re-executing the code with predicted inputs, we can verify each prediction and further enhance the CoTs through multi-turn revision, resulting in CodeI/O++ and achieving higher performance. Our data and models are available at https://github.com/hkust-nlp/CodeIO. |
2025-02-12T00:00:00 | 2502.06589 | Hephaestus: Improving Fundamental Agent Capabilities of Large Language Models through Continual Pre-Training | [
"Yuchen Zhuang",
"Jingfeng Yang",
"Haoming Jiang",
"Xin Liu",
"Kewei Cheng",
"Sanket Lokegaonkar",
"Yifan Gao",
"Qing Ping",
"Tianyi Liu",
"Binxuan Huang",
"Zheng Li",
"Zhengyang Wang",
"Pei Chen",
"Ruijie Wang",
"Rongzhi Zhang",
"Nasser Zalmout",
"Priyanka Nigam",
"Bing Yin",
"Chao Zhang"
] | Due to the scarcity of agent-oriented pre-training data, LLM-based autonomous agents typically rely on complex prompting or extensive fine-tuning, which often fails to introduce new capabilities while preserving strong generalizability. We introduce Hephaestus-Forge, the first large-scale pre-training corpus designed to enhance the fundamental capabilities of LLM agents in API function calling, intrinsic reasoning and planning, and adapting to environmental feedback. Hephaestus-Forge comprises 103B agent-specific data encompassing 76,537 APIs, including both tool documentation to introduce knowledge of API functions and function calling trajectories to strengthen intrinsic reasoning. To explore effective training protocols, we investigate scaling laws to identify the optimal recipe in data mixing ratios. By continual pre-training on Hephaestus-Forge, Hephaestus outperforms small- to medium-scale open-source LLMs and rivals commercial LLMs on three agent benchmarks, demonstrating the effectiveness of our pre-training corpus in enhancing fundamental agentic capabilities and generalization of LLMs to new tasks or environments. |
|
2025-02-12T00:00:00 | 2502.07617 | Scaling Pre-training to One Hundred Billion Data for Vision Language Models | [
"Xiao Wang",
"Ibrahim Alabdulmohsin",
"Daniel Salz",
"Zhe Li",
"Keran Rong",
"Xiaohua Zhai"
] | We provide an empirical investigation of the potential of pre-training vision-language models on an unprecedented scale: 100 billion examples. We find that model performance tends to saturate at this scale on many common Western-centric classification and retrieval benchmarks, such as COCO Captions. Nevertheless, tasks of cultural diversity achieve more substantial gains from the 100-billion scale web data, thanks to its coverage of long-tail concepts. Furthermore, we analyze the model's multilinguality and show gains in low-resource languages as well. In addition, we observe that reducing the size of the pretraining dataset via quality filters like using CLIP, typically used to enhance performance, may inadvertently reduce the cultural diversity represented even in large-scale datasets. Our results highlight that while traditional benchmarks may not benefit significantly from scaling noisy, raw web data to 100 billion examples, this data scale is vital for building truly inclusive multimodal systems. |
Subsets and Splits