The rapid progress of large language models (LLMs) has increased the demand for efficient and reliable evaluation of question answering (QA) systems. Existing evaluation methods either rely on rule-based matching with shallow semantic understanding or adopt LLM-as-a-Judge approaches that incur high cost and latency while offering limited error interpretability. Accordingly, we propose HiEval, a curriculum learning based hierarchical framework for QA task evaluation that supports both quick scoring and fine-grained error analysis. HiEval contains a quick scoring model (HiEval-QS) that predicts three-level correctness labels, and an error analysis model (HiEval-EA) that identifies incorrect responses into five error types. HiEval incorporates a class-balanced focal loss to handle label imbalance, experience replay to prevent forgetting, and contrastive unlikelihood optimization to improve error discrimination. We also construct two large-scale human-annotated evaluation datasets collected from 50 QA-related datasets, covering 8 task types and release two challenging benchmarks. Extensive experiments show that HiEval achieves state-of-the-art performance on both quick scoring and error analysis tasks, outperforming all baseline methods, including GPT-5, while being approximately 25× faster.

Figure-to-Text (F2T) tasks aim to convert structured figure information into natural language text, serving as a bridge between visual perception and language understanding.However, existing evaluation methods remain limited: 1) Reference-based methods can only capture shallow semantic similarities and rely on costly labeled reference text; 2) Reference-free methods depend on multimodal large language models, which suffer from low efficiency and instruction sensitivity; 3) Existing methods provide only sample-level evaluations, lacking interpretability and alignment with expert-level multi-dimensional evaluation criteria.Accordingly, we propose F2TEval, a five-dimensional reference-free evaluation method aligned with expert criteria, covering faithfulness, completeness, conciseness, logicality, and analysis, to support fine-grained evaluation. We design a lightweight mixture-of-experts model that incorporates independent scoring heads and applies the Hilbert-Schmidt Independence Criterion to optimize the disentanglement of scoring representations across dimensions. Furthermore, we construct F2TBenchmark, a human-annotated benchmark dataset covering 21 chart types and 35 application domains, to support research on F2T evaluation. Experimental results demonstrate our model’s superior performance and efficiency, outperforming Gemini-2.0 and Claude-3.5 with only 0.9B parameters.

Question answering (QA) tasks serve as a key benchmark for evaluating generation systems. Traditional rule-based metrics, such as accuracy and relaxed-accuracy, struggle with open-ended and unstructured responses. LLM-based evaluation methods offer greater flexibility but suffer from sensitivity to instructions, robustness issues, and high computational costs. To overcome these challenges, we introduce QAEval, a hybrid framework combining rule-based reliability with LLM-based adaptability. QAEval utilizes two high-quality datasets: QAExtract for short-answer extraction and QAScore for scoring model training. By integrating a Mixture of Evaluators model with Dynamic Load Balancing Optimization, QAEval enables accurate, cost-effective QA evaluation. Experimental results show it outperforms models like GPT-4o and Claude-3, achieving 92.3% accuracy with only 0.6B parameters.

Sarcasm poses a challenge in linguistic analysis due to its implicit nature, involving an intended meaning that contradicts the literal expression. The advent of social networks has propelled the utilization of multimodal data to enhance sarcasm detection performance. In prior multimodal sarcasm detection datasets, a single label is assigned to a multimodal instance. Subsequent experiments often highlight the superiority of multimodal models by demonstrating their improvements compared to unimodal models based on these unified labels across multiple modalities. However, our investigation revealed that numerous instances of sarcasm cannot be identified using a single modality. Humans employ the conflict between a statement and factual information as a cue to detect sarcasm, and these cues can stem from different modalities. Then, a unified label for a multimodal instance may be not suitable for the associated text or image. In this work, we introduce SarcNet, a multilingual and multimodal sarcasm detection dataset in English and Chinese, consisting of 3,335 image-text pair samples. We provide annotations for sarcasm in visual, textual, and multimodal data, respectively, resulting in over 10,000 labeled instances. The separated annotation schema for unimodal and multimodal data facilitates a more accurate and reasonable assessment of unimodal and multimodal models.