Zehua Li

Neel Guha, Julian Nyarko, Daniel E. Ho et al.

The advent of large language models (LLMs) and their adoption by the legal community has given rise to the question: what types of legal reasoning can LLMs perform? To enable greater study of this question, we present LegalBench: a collaboratively constructed legal reasoning benchmark consisting of 162 tasks covering six different types of legal reasoning. LegalBench was built through an interdisciplinary process, in which we collected tasks designed and hand-crafted by legal professionals. Because these subject matter experts took a leading role in construction, tasks either measure legal reasoning capabilities that are practically useful, or measure reasoning skills that lawyers find interesting. To enable cross-disciplinary conversations about LLMs in the law, we additionally show how popular legal frameworks for describing legal reasoning -- which distinguish between its many forms -- correspond to LegalBench tasks, thus giving lawyers and LLM developers a common vocabulary. This paper describes LegalBench, presents an empirical evaluation of 20 open-source and commercial LLMs, and illustrates the types of research explorations LegalBench enables.

Peisong Wang, Bowen Liu, Zehua Li et al.

Large language models still struggle with contest-level programming, while many agentic remedies rely on massive inference-time sampling or expensive multi-stage post-training. We study when execution feedback reliably helps an LLM CP solver and which mechanisms govern the gains. We model feedback-driven solving as a calibrated stopped process and identify three quantities: false-admission risk, program-level evidence against bad programs, and the active-state success hazard. Under held-out trace calibration and selection from a pre-declared finite controller manifest, the resulting structural certificate lower-bounds the clean success probability before false admission. We instantiate mechanisms targeting these quantities as Dual-Granularity Verification, Test Augmentation, and Experience-Driven Self-Evolving, yielding CP-Agent. Without updating any parameters, CP-Agent raises Pass@1 from 25.8\% to 48.5\% on LiveCodeBench Pro and improves Refine@5 by 11.0\% on ICPC-Eval. Across three LLM backbones, CP-Agent lies on the cost--accuracy efficiency frontier, and ablations show that each component primarily affects its corresponding certificate quantity.

Nuo Chen, Zehua Li, Keqin Bao et al.

Building robust and general reasoning ability is a central goal in the development of large language models (LLMs). Recent efforts increasingly turn to code as a rich training source, given its inherent logical structure and diverse reasoning paradigms such as divide-and-conquer, topological ordering, and enumeration. However, reasoning in code is often expressed implicitly and entangled with syntactic or implementation noise, making direct training on raw code suboptimal.To address this, we introduce TracePile, a large-scale corpus of 2.6 million samples that transforms code execution into explicit, step-by-step chain-of-thought-style rationales, which we call Chain of Execution (CoE). The corpus spans domains including mathematics, classical algorithms and algorithmic competition, and is enriched with variable-tracing questions and code rewritings to enhance logical granularity and code diversity. We evaluate TracePile using three training setups: continue-pretraining, instruction tuning after pretraining, and two-stage finetuning. Experiments across four base models (LLaMA 3, LLaMA 3.1, Qwen-2.5, and Qwen-2.5 Coder) and 20 benchmarks covering math, code, logic, and algorithms demonstrate consistent improvements. Notably, TracePile boosts LLaMA3.1-8B by 7.1\% on average across nine math datasets and delivers clear gains on LiveCodeBench, CRUX, and MMLU under two-stage fine-tuning.

Zehua Li

This paper presents a configuration-first framework for evaluating cross-backend compatibility in deep learning systems deployed on CPU, GPU, and compiled runtimes. The framework decouples experiments from code using YAML, supports both library and repository models, and employs a three-tier verification protocol covering tensor-level closeness, activation alignment, and task-level metrics. Through 672 checks across multiple models and tolerance settings, we observe that 72.0% of runs pass, with most discrepancies occurring under stricter thresholds. Our results show that detection models and compiled backends are particularly prone to drift, often due to nondeterministic post-processing. We further demonstrate that deterministic adapters and selective fallbacks can substantially improve agreement without significant performance loss. To our knowledge, this is the first unified framework that systematically quantifies and mitigates cross-backend drift in deep learning, providing a reproducible methodology for dependable deployment across heterogeneous runtimes.

Qifan Zhang, Nuo Chen, Zehua Li et al.

Large Language Models (LLMs) have made remarkable strides in reasoning tasks, yet their performance often falters on novel and complex problems. Domain-specific continued pretraining (CPT) methods, such as those tailored for mathematical reasoning, have shown promise but lack transferability to broader reasoning tasks. In this work, we pioneer the use of Graph Problem Reasoning (GPR) to enhance the general reasoning capabilities of LLMs. GPR tasks, spanning pathfinding, network analysis, numerical computation, and topological reasoning, require sophisticated logical and relational reasoning, making them ideal for teaching diverse reasoning patterns. To achieve this, we introduce GraphPile, the first large-scale corpus specifically designed for CPT using GPR data. Spanning 10.9 billion tokens across 23 graph tasks, the dataset includes chain-of-thought, program-of-thought, trace of execution, and real-world graph data. Using GraphPile, we train GraphMind on popular base models Llama 3 and 3.1, as well as Gemma 2, achieving up to 4.9 percent higher accuracy in mathematical reasoning and up to 21.2 percent improvement in non-mathematical reasoning tasks such as logical and commonsense reasoning. By being the first to harness GPR for enhancing reasoning patterns and introducing the first dataset of its kind, our work bridges the gap between domain-specific pretraining and universal reasoning capabilities, advancing the adaptability and robustness of LLMs.

Ashkan Kazemi, Zehua Li, Verónica Pérez-Rosas et al.

An important challenge for news fact-checking is the effective dissemination of existing fact-checks. This in turn brings the need for reliable methods to detect previously fact-checked claims. In this paper, we focus on automatically finding existing fact-checks for claims made in social media posts (tweets). We conduct both classification and retrieval experiments, in monolingual (English only), multilingual (Spanish, Portuguese), and cross-lingual (Hindi-English) settings using multilingual transformer models such as XLM-RoBERTa and multilingual embeddings such as LaBSE and SBERT. We present promising results for "match" classification (86% average accuracy) in four language pairs. We also find that a BM25 baseline outperforms or is on par with state-of-the-art multilingual embedding models for the retrieval task during our monolingual experiments. We highlight and discuss NLP challenges while addressing this problem in different languages, and we introduce a novel curated dataset of fact-checks and corresponding tweets for future research.

Ashkan Kazemi, Zehua Li, Verónica Pérez-Rosas et al.

In this paper, we explore the construction of natural language explanations for news claims, with the goal of assisting fact-checking and news evaluation applications. We experiment with two methods: (1) an extractive method based on Biased TextRank -- a resource-effective unsupervised graph-based algorithm for content extraction; and (2) an abstractive method based on the GPT-2 language model. We perform comparative evaluations on two misinformation datasets in the political and health news domains, and find that the extractive method shows the most promise.