Zhiheng Lyu

Zhijing Jin, Jiarui Liu, Zhiheng Lyu et al.

Causal inference is one of the hallmarks of human intelligence. While the field of CausalNLP has attracted much interest in the recent years, existing causal inference datasets in NLP primarily rely on discovering causality from empirical knowledge (e.g., commonsense knowledge). In this work, we propose the first benchmark dataset to test the pure causal inference skills of large language models (LLMs). Specifically, we formulate a novel task Corr2Cause, which takes a set of correlational statements and determines the causal relationship between the variables. We curate a large-scale dataset of more than 200K samples, on which we evaluate seventeen existing LLMs. Through our experiments, we identify a key shortcoming of LLMs in terms of their causal inference skills, and show that these models achieve almost close to random performance on the task. This shortcoming is somewhat mitigated when we try to re-purpose LLMs for this skill via finetuning, but we find that these models still fail to generalize -- they can only perform causal inference in in-distribution settings when variable names and textual expressions used in the queries are similar to those in the training set, but fail in out-of-distribution settings generated by perturbing these queries. Corr2Cause is a challenging task for LLMs, and would be helpful in guiding future research on improving LLMs' pure reasoning skills and generalizability. Our data is at https://huggingface.co/datasets/causalnlp/corr2cause. Our code is at https://github.com/causalNLP/corr2cause.

Jialin Yang, Dongfu Jiang, Lipeng He et al. · amazon-science, utoronto

As Large Language Models (LLMs) become integral to software development workflows, their ability to generate structured outputs has become critically important. We introduce StructEval, a comprehensive benchmark for evaluating LLMs' capabilities in producing both non-renderable (JSON, YAML, CSV) and renderable (HTML, React, SVG) structured formats. Unlike prior benchmarks, StructEval systematically evaluates structural fidelity across diverse formats through two paradigms: 1) generation tasks, producing structured output from natural language prompts, and \textbf{2)} conversion tasks, translating between structured formats. Our benchmark encompasses 18 formats and 44 types of task, with novel metrics for format adherence and structural correctness. Results reveal significant performance gaps-even state-of-the-art models like o1-mini achieve only 75.58 average score, with open-source alternatives lagging approximately 10 points behind. We find generation tasks more challenging than conversion tasks, and producing correct visual content more difficult than generating text-only structures.

Zhijing Jin, Yuen Chen, Felix Leeb et al.

The ability to perform causal reasoning is widely considered a core feature of intelligence. In this work, we investigate whether large language models (LLMs) can coherently reason about causality. Much of the existing work in natural language processing (NLP) focuses on evaluating commonsense causal reasoning in LLMs, thus failing to assess whether a model can perform causal inference in accordance with a set of well-defined formal rules. To address this, we propose a new NLP task, causal inference in natural language, inspired by the "causal inference engine" postulated by Judea Pearl et al. We compose a large dataset, CLadder, with 10K samples: based on a collection of causal graphs and queries (associational, interventional, and counterfactual), we obtain symbolic questions and ground-truth answers, through an oracle causal inference engine. These are then translated into natural language. We evaluate multiple LLMs on our dataset, and we introduce and evaluate a bespoke chain-of-thought prompting strategy, CausalCoT. We show that our task is highly challenging for LLMs, and we conduct an in-depth analysis to gain deeper insights into the causal reasoning abilities of LLMs. Our data is open-sourced at https://huggingface.co/datasets/causalNLP/cladder, and our code can be found at https://github.com/causalNLP/cladder.

MiniMax, Aili Chen, Aonian Li et al.

We introduce MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. MiniMax-M1 is powered by a hybrid Mixture-of-Experts (MoE) architecture combined with a lightning attention mechanism. The model is developed based on our previous MiniMax-Text-01 model, which contains a total of 456 billion parameters with 45.9 billion parameters activated per token. The M1 model natively supports a context length of 1 million tokens, 8x the context size of DeepSeek R1. Furthermore, the lightning attention mechanism in MiniMax-M1 enables efficient scaling of test-time compute. These properties make M1 particularly suitable for complex tasks that require processing long inputs and thinking extensively. MiniMax-M1 is trained using large-scale reinforcement learning (RL) on diverse problems including sandbox-based, real-world software engineering environments. In addition to M1's inherent efficiency advantage for RL training, we propose CISPO, a novel RL algorithm to further enhance RL efficiency. CISPO clips importance sampling weights rather than token updates, outperforming other competitive RL variants. Combining hybrid-attention and CISPO enables MiniMax-M1's full RL training on 512 H800 GPUs to complete in only three weeks, with a rental cost of just $534,700. We release two versions of MiniMax-M1 models with 40K and 80K thinking budgets respectively, where the 40K model represents an intermediate phase of the 80K training. Experiments on standard benchmarks show that our models are comparable or superior to strong open-weight models such as the original DeepSeek-R1 and Qwen3-235B, with particular strengths in complex software engineering, tool utilization, and long-context tasks. We publicly release MiniMax-M1 at https://github.com/MiniMax-AI/MiniMax-M1.

Dongfu Jiang, Yi Lu, Zhuofeng Li et al. · utoronto

Reinforcement Learning with Verifiable Rewards (RLVR) has demonstrated success in enhancing LLM reasoning capabilities, but remains limited to single-turn interactions without tool integration. While recent Agentic Reinforcement Learning with Tool use (ARLT) approaches have emerged to address multi-turn tool interactions, existing works develop task-specific codebases that suffer from fragmentation, synchronous execution bottlenecks, and limited extensibility across domains. These inefficiencies hinder broader community adoption and algorithmic innovation. We introduce VerlTool, a unified and modular framework that addresses these limitations through systematic design principles. VerlTool provides four key contributions: (1) upstream alignment with VeRL ensuring compatibility and simplified maintenance, (2) unified tool management via standardized APIs supporting diverse modalities including code execution, search, SQL databases, and vision processing, (3) asynchronous rollout execution achieving near 2$\times$ speedup by eliminating synchronization bottlenecks, and (4) comprehensive evaluation demonstrating competitive performance across 6 ARLT domains. Our framework formalizes ARLT as multi-turn trajectories with multi-modal observation tokens (text/image/video), extending beyond single-turn RLVR paradigms. We train and evaluate models on mathematical reasoning, knowledge QA, SQL generation, visual reasoning, web search, and software engineering tasks, achieving results comparable to specialized systems while providing unified training infrastructure. The modular plugin architecture enables rapid tool integration requiring only lightweight Python definitions, significantly reducing development overhead and providing a scalable foundation for tool-augmented RL research. Our code is open-sourced at https://github.com/TIGER-AI-Lab/verl-tool.

Jiarong Liang, Zhiheng Lyu, Zijie Liu et al.

Executable software engineering data is valuable for training SWE agents, but scaling it remains difficult for two reasons: only a small fraction of real repository changes yield verifiable, high-signal task instances, and naively building repository-specific environments quickly becomes the dominant systems cost. We present SWE-Next, an execution-grounded framework for scalable SWE task and trajectory collection. On the data side, SWE-Next mines real merged pull requests, executes candidate base/merged commit pairs, and retains only those that produce strict test improvements without regressions, yielding self-verifying instances. It also applies strict submission gating so that collected trajectories remain evidence-driven rather than speculative. On the systems side, SWE-Next introduces reusable repo-quarter profiles, which reuse the same environment across nearby commits in time while keeping each task run separate and reproducible. Using only 30 hours and 639GB of environment storage, SWE-Next processes 3,971 seed repositories and 102,582 candidate commit pairs mined from real merged PRs to construct a dataset of 2,308 self-verifying instances. Experiments show that SWE-Next improves downstream pass@1 with fewer or comparable training trajectories, indicating that its gains come not from a stronger trajectory generator, but from higher-signal execution-grounded supervision and more efficient data collection.

Songcheng Cai, Zhiheng Lyu, Yuansheng Ni et al.

Agentic repository-level code understanding is essential for automating complex software engineering tasks, yet the field lacks reliable benchmarks. Existing evaluations often overlook the long tail topics and rely on popular repositories where Large Language Models (LLMs) can cheat via memorized knowledge. To address this, we introduce SWE-QA-Pro, a benchmark constructed from diverse, long-tail repositories with executable environments. We enforce topical balance via issue-driven clustering to cover under-represented task types and apply a rigorous difficulty calibration process: questions solvable by direct-answer baselines are filtered out. This results in a dataset where agentic workflows significantly outperform direct answering (e.g., a ~13-point gap for Claude Sonnet 4.5), confirming the necessity of agentic codebase exploration. Furthermore, to tackle the scarcity of training data for such complex behaviors, we propose a scalable synthetic data pipeline that powers a two-stage training recipe: Supervised Fine-Tuning (SFT) followed by Reinforcement Learning from AI Feedback (RLAIF). This approach allows small open models to learn efficient tool usage and reasoning. Empirically, a Qwen3-8B model trained with our recipe surpasses GPT-4o by 2.3 points on SWE-QA-Pro and substantially narrows the gap to state-of-the-art proprietary models, demonstrating both the validity of our evaluation and the effectiveness of our agentic training workflow.

Zhiheng Lyu, Kevin Yang, Lingpeng Kong et al.

While accurately detecting and correcting factual contradictions in language model outputs has become increasingly important as their capabilities improve, doing so is highly challenging. We propose a novel method, FACTTRACK, for tracking atomic facts and addressing factual contradictions. Crucially, FACTTRACK also maintains time-aware validity intervals for each fact, allowing for change over time. At a high level, FACTTRACK consists of a four-step pipeline to update a world state data structure for each new event: (1) decompose the event into directional atomic facts; (2) determine the validity interval of each atomic fact using the world state; (3) detect contradictions with existing facts in the world state; and finally (4) add new facts to the world state and update existing atomic facts. When we apply FACTTRACK to contradiction detection on structured story outlines, we find that FACTTRACK using LLaMA2-7B-Chat substantially outperforms a fair baseline using LLaMA2-7B-Chat, and achieves performance comparable to a GPT4 baseline. Moreover, when using GPT4, FACTTRACK significantly outperforms the GPT4 baseline.

Zhiheng Lyu, Zhijing Jin, Fernando Gonzalez et al.

Sentiment analysis (SA) aims to identify the sentiment expressed in a text, such as a product review. Given a review and the sentiment associated with it, this work formulates SA as a combination of two tasks: (1) a causal discovery task that distinguishes whether a review "primes" the sentiment (Causal Hypothesis C1), or the sentiment "primes" the review (Causal Hypothesis C2); and (2) the traditional prediction task to model the sentiment using the review as input. Using the peak-end rule in psychology, we classify a sample as C1 if its overall sentiment score approximates an average of all the sentence-level sentiments in the review, and C2 if the overall sentiment score approximates an average of the peak and end sentiments. For the prediction task, we use the discovered causal mechanisms behind the samples to improve LLM performance by proposing causal prompts that give the models an inductive bias of the underlying causal graph, leading to substantial improvements by up to 32.13 F1 points on zero-shot five-class SA. Our code is at https://github.com/cogito233/causal-sa

Yuansheng Ni, Songcheng Cai, Xiangchao Chen et al.

Large language models (LLMs) have recently enabled coding agents capable of generating, executing, and revising visualization code. However, existing models often fail in practical workflows due to limited language coverage, unreliable execution, and lack of iterative correction mechanisms. Progress has been constrained by narrow datasets and benchmarks that emphasize single-round generation and single-language tasks. To address these challenges, we introduce three complementary resources for advancing visualization coding agents. VisCode-Multi-679K is a large-scale, supervised dataset containing 679K validated and executable visualization samples with multi-turn correction dialogues across 12 programming languages. VisPlotBench is a benchmark for systematic evaluation, featuring executable tasks, rendered outputs, and protocols for both initial generation and multi-round self-debug. Finally, we present VisCoder2, a family of multi-language visualization models trained on VisCode-Multi-679K. Experiments show that VisCoder2 significantly outperforms strong open-source baselines and approaches the performance of proprietary models like GPT-4.1, with further gains from iterative self-debug, reaching 82.4% overall execution pass rate at the 32B scale, particularly in symbolic or compiler-dependent languages.

Utkarsh Pratiush, Austin Houston, Kamyar Barakati et al.

Microscopy is a primary source of information on materials structure and functionality at nanometer and atomic scales. The data generated is often well-structured, enriched with metadata and sample histories, though not always consistent in detail or format. The adoption of Data Management Plans (DMPs) by major funding agencies promotes preservation and access. However, deriving insights remains difficult due to the lack of standardized code ecosystems, benchmarks, and integration strategies. As a result, data usage is inefficient and analysis time is extensive. In addition to post-acquisition analysis, new APIs from major microscope manufacturers enable real-time, ML-based analytics for automated decision-making and ML-agent-controlled microscope operation. Yet, a gap remains between the ML and microscopy communities, limiting the impact of these methods on physics, materials discovery, and optimization. Hackathons help bridge this divide by fostering collaboration between ML researchers and microscopy experts. They encourage the development of novel solutions that apply ML to microscopy, while preparing a future workforce for instrumentation, materials science, and applied ML. This hackathon produced benchmark datasets and digital twins of microscopes to support community growth and standardized workflows. All related code is available at GitHub: https://github.com/KalininGroup/Mic-hackathon-2024-codes-publication/tree/1.0.0.1

Yiwen Ding, Jiarui Liu, Zhiheng Lyu et al.

While several previous studies have analyzed gender bias in research, we are still missing a comprehensive analysis of gender differences in the AI community, covering diverse topics and different development trends. Using the AI Scholar dataset of 78K researchers in the field of AI, we identify several gender differences: (1) Although female researchers tend to have fewer overall citations than males, this citation difference does not hold for all academic-age groups; (2) There exist large gender homophily in co-authorship on AI papers; (3) Female first-authored papers show distinct linguistic styles, such as longer text, more positive emotion words, and more catchy titles than male first-authored papers. Our analysis provides a window into the current demographic trends in our AI community, and encourages more gender equality and diversity in the future. Our code and data are at https://github.com/causalNLP/ai-scholar-gender.

Zhiheng Lyu, Zhijing Jin, Justus Mattern et al.

NLP datasets are richer than just input-output pairs; rather, they carry causal relations between the input and output variables. In this work, we take sentiment classification as an example and look into the causal relations between the review (X) and sentiment (Y). As psychology studies show that language can affect emotion, different psychological processes are evoked when a person first makes a rating and then self-rationalizes their feeling in a review (where the sentiment causes the review, i.e., Y -> X), versus first describes their experience, and weighs the pros and cons to give a final rating (where the review causes the sentiment, i.e., X -> Y ). Furthermore, it is also a completely different psychological process if an annotator infers the original rating of the user by theory of mind (ToM) (where the review causes the rating, i.e., X -ToM-> Y ). In this paper, we verbalize these three causal mechanisms of human psychological processes of sentiment classification into three different causal prompts, and study (1) how differently they perform, and (2) what nature of sentiment classification data leads to agreement or diversity in the model responses elicited by the prompts. We suggest future work raise awareness of different causal structures in NLP tasks. Our code and data are at https://github.com/cogito233/psych-causal-prompt

Zhijing Jin, Abhinav Lalwani, Tejas Vaidhya et al.

Reasoning is central to human intelligence. However, fallacious arguments are common, and some exacerbate problems such as spreading misinformation about climate change. In this paper, we propose the task of logical fallacy detection, and provide a new dataset (Logic) of logical fallacies generally found in text, together with an additional challenge set for detecting logical fallacies in climate change claims (LogicClimate). Detecting logical fallacies is a hard problem as the model must understand the underlying logical structure of the argument. We find that existing pretrained large language models perform poorly on this task. In contrast, we show that a simple structure-aware classifier outperforms the best language model by 5.46% on Logic and 4.51% on LogicClimate. We encourage future work to explore this task as (a) it can serve as a new reasoning challenge for language models, and (b) it can have potential applications in tackling the spread of misinformation. Our dataset and code are available at https://github.com/causalNLP/logical-fallacy

Yubo Wang, Xueguang Ma, Ping Nie et al. · utoronto

Academic writing requires both coherent text generation and precise citation of relevant literature. Although recent Retrieval-Augmented Generation (RAG) systems have significantly improved factual accuracy in general-purpose text generation, their ability to support professional academic writing remains limited. In this work, we introduce ScholarCopilot, a unified framework designed to enhance existing large language models for generating professional academic articles with accurate and contextually relevant citations. ScholarCopilot dynamically determines when to retrieve scholarly references by generating a retrieval token [RET], which is then used to query a citation database. The retrieved references are fed into the model to augment the generation process. We jointly optimize both the generation and citation tasks within a single framework to improve efficiency. Our model is built upon Qwen-2.5-7B and trained on 500K papers from arXiv. It achieves a top-1 retrieval accuracy of 40.1% on our evaluation dataset, outperforming baselines such as E5-Mistral-7B-Instruct (15.0%) and BM25 (9.8%). On a dataset of 1,000 academic writing samples, ScholarCopilot scores 16.2/25 in generation quality -- measured across relevance, coherence, academic rigor, completeness, and innovation -- significantly surpassing all existing models, including much larger ones like the Retrieval-Augmented Qwen2.5-72B-Instruct. Human studies further demonstrate that ScholarCopilot, despite being a 7B model, significantly outperforms ChatGPT, achieving 100% preference in citation quality and over 70% in overall usefulness.

Tao Yu, Zhengbo Zhang, Zhiheng Lyu et al.

Efficiently solving real-world problems with LLMs increasingly hinges on their ability to interact with dynamic web environments and autonomously acquire external information. While recent research like Search-R1 and WebDancer demonstrates strong performance in solving web tasks, they heavily rely on additional tools to convert the interactive web environment into static text content. This is in contrast to human browsing behaviors, which involve diverse interactions with the browser, such as scrolling, clicking, and typing. In this paper, we propose BrowserAgent, a more interactive agent that solves complex tasks through human-inspired browser actions. BrowserAgent operates directly on raw web pages via Playwright through a set of predefined browser actions. We adopt a two-stage training (Supervised Fine-Tuning (SFT) and Rejection Fine-Tuning (RFT)) to improve the model's generalization abilities. Despite using significantly less training data than Search-R1, BrowserAgent achieves more competitive results across different Open-QA tasks. Additionally, we introduce an explicit memory mechanism to store key conclusions across steps, further enhancing the model's reasoning capabilities for long-horizon tasks. Notably, BrowserAgent-7B can achieve around 20\% improvement over Search-R1 on multi-hop QA tasks like HotpotQA, 2Wiki, and Bamboogle. These results indicate that BrowserAgent can serve as a more advanced framework for more interactive and scalable web agents.

Zhiheng Lyu, Xueguang Ma, Wenhu Chen

Recent agentic language models increasingly need to interact with real-world environments that contain tightly intertwined visual and textual information, often through raw camera pixels rather than separately processed images and tokenized text. This shift highlights the need for a unified perception paradigm. To investigate this idea, we explore Perceive Everything as Pixels (PEAP) and introduce PixelWorld, a benchmark that renders natural-language, tabular, mathematical, and diagrammatic inputs into a shared pixel space. Experiments across multiple benchmarks show that PEAP achieves comparable performance to token-based approaches on semantic understanding tasks, suggesting that vision transformers can partially capture global textual semantics without explicit tokenization. In contrast, reasoning-intensive tasks such as mathematics and code show notable performance degradation, although Chain-of-Thought prompting helps mitigate this gap by compensating for missing symbolic structure. We further find that when visual and textual information are closely integrated, representing everything as pixels simplifies preprocessing and avoids cross-modal misalignment. PixelWorld thus provides a systematic and practical framework for evaluating unified vision--language models and facilitates further exploration of pixel-based multimodal learning.

Xuan He, Dongfu Jiang, Ge Zhang et al.

The recent years have witnessed great advances in video generation. However, the development of automatic video metrics is lagging significantly behind. None of the existing metric is able to provide reliable scores over generated videos. The main barrier is the lack of large-scale human-annotated dataset. In this paper, we release VideoFeedback, the first large-scale dataset containing human-provided multi-aspect score over 37.6K synthesized videos from 11 existing video generative models. We train VideoScore (initialized from Mantis) based on VideoFeedback to enable automatic video quality assessment. Experiments show that the Spearman correlation between VideoScore and humans can reach 77.1 on VideoFeedback-test, beating the prior best metrics by about 50 points. Further result on other held-out EvalCrafter, GenAI-Bench, and VBench show that VideoScore has consistently much higher correlation with human judges than other metrics. Due to these results, we believe VideoScore can serve as a great proxy for human raters to (1) rate different video models to track progress (2) simulate fine-grained human feedback in Reinforcement Learning with Human Feedback (RLHF) to improve current video generation models.

Hongyin Zhu, Hao Peng, Zhiheng Lyu et al.

Existing technologies expand BERT from different perspectives, e.g. designing different pre-training tasks, different semantic granularities, and different model architectures. Few models consider expanding BERT from different text formats. In this paper, we propose a heterogeneous knowledge language model (\textbf{HKLM}), a unified pre-trained language model (PLM) for all forms of text, including unstructured text, semi-structured text, and well-structured text. To capture the corresponding relations among these multi-format knowledge, our approach uses masked language model objective to learn word knowledge, uses triple classification objective and title matching objective to learn entity knowledge and topic knowledge respectively. To obtain the aforementioned multi-format text, we construct a corpus in the tourism domain and conduct experiments on 5 tourism NLP datasets. The results show that our approach outperforms the pre-training of plain text using only 1/4 of the data. We further pre-train the domain-agnostic HKLM and achieve performance gains on the XNLI dataset.