arXiv subjects

Zijiao Zhang

Publications and source records attributed to Zijiao Zhang.

4 recordsLinked to original sources

arXiv2026

Beyond Benchmarks: LLM Evaluation with an Anthropomorphic and Lifecycle-oriented Roadmap

Despite their rapid advancement, large language models (LLMs) suffer from a critical disconnect between benchmark scores and real-world utility. Current evaluation remains fragmented, prioritizing isolated technical metrics over the holistic, developmental, and societal aspects essential for deployment. Rather than serving merely as a descriptive catalog, this work establishes a diagnostic ontology that causally maps evaluation dimensions to the canonical LLM training pipeline, transforming evaluation from static ranking into a diagnostic tool for root-cause analysis. In this paper, we introduce an anthropomorphic evaluation framework that re-conceptualizes LLM capabilities through a four-dimensional lens: Intelligence Quotient (IQ), Professional Quotient (PQ), Emotional Quotient (EQ), and Value-oriented Quotient (VQ). We operationalize these concepts through a modular evaluation architecture and validate the framework's diagnostic claims through meta-analysis of public benchmark trends. Analyzing over 200 benchmarks, we synthesize key challenges and future directions. This work offers a strategic compass for developing LLMs that are not only technically proficient but also contextually relevant and ethically sound. A curated repository is available at: https://github.com/onejune2018/Awesome-LLM-Eval.

cs.AI↗

arXiv2026

FinixDoc: Rethinking Financial Document Parsing Beyond Saturated Benchmarks

Financial document parsing requires accuracy, structural consistency, and verifiability that current benchmarks often fail to reflect. We present FinixDoc, an end-to-end agentic parsing system for real-world financial documents, with FinixDoc-VL, a 4B-scale vision-language model built on Qwen3-VL-4B, as its core parser. To characterize the gap between benchmark and deployment performance, we introduce a Document Parsing Capability Matrix organized along two practical axes: visual quality and document scale. Guided by this matrix, FinixDoc-VL is trained with a domain-adapted recipe combining homoglyph-aware contrastive learning and multi-stage reinforcement learning with composite domain-specific rewards. To better leverage our accumulated advantage in low-quality financial-document data and support large-scale, high-quality data production, we further build a human-in-the-loop Data Factory pipeline with confidence-aware expert review. For evaluation, we construct FinixDocBench, a financial-domain evaluation suite covering digital-native, camera-captured, ultra-large-page, and internal-workflow scenarios, with a compliance-reviewed subset released alongside this technical report. On its main subsets, FinixDoc-VL achieves the highest overall score (81.43) among evaluated baselines, outperforming the next-best open-source model by 5.13 points, with the largest gains on internal financial workflows (FinixInner: 84.08 vs. 78.73).

cs.AI↗

arXiv2026

Entropy-Tree: Tree-Based Decoding with Entropy-Guided Exploration

Large language models achieve strong reasoning performance, yet existing decoding strategies either explore blindly (random sampling) or redundantly (independent multi-sampling). We propose Entropy-Tree, a tree-based decoding method that exploits entropy as a signal for branching decisions--expanding the search tree only at positions where the model exhibits genuine uncertainty. Entropy-Tree shows superior accuracy and calibration in reasoning tasks: it achieves better pass@k than Multi-chain across multiple models and datasets, and its predictive entropy demonstrates better AUROC compared to several traditional metrics. Entropy-Tree unifies efficient structured exploration and reliable uncertainty estimation within a single decoding procedure.

cs.CL↗

arXiv2020

Cost-effectiveness Analysis of Antiepidemic Policies and Global Situation Assessment of COVID-19

With a two-layer contact-dispersion model and data in China, we analyze the cost-effectiveness of three types of antiepidemic measures for COVID-19: regular epidemiological control, local social interaction control, and inter-city travel restriction. We find that: 1) intercity travel restriction has minimal or even negative effect compared to the other two at the national level; 2) the time of reaching turning point is independent of the current number of cases, and only related to the enforcement stringency of epidemiological control and social interaction control measures; 3) strong enforcement at the early stage is the only opportunity to maximize both antiepidemic effectiveness and cost-effectiveness; 4) mediocre stringency of social interaction measures is the worst choice. Subsequently, we cluster countries/regions into four groups based on their control measures and provide situation assessment and policy suggestions for each group.

physics.soc-ph↗