A Unified Agentic Framework for Evaluating Conditional Image Generation

Conditional image generation has gained significant attention for its ability to personalize content. However, the field faces challenges in developing task-agnostic, reliable, and explainable evaluation metrics. This paper introduces CIGEval, a unified agentic framework for comprehensive evaluation of conditional image generation tasks. CIGEval utilizes large multimodal models (LMMs) as its core, integrating a multi-functional toolbox and establishing a fine-grained evaluation framework. Additionally, we synthesize evaluation trajectories for fine-tuning, empowering smaller LMMs to autonomously select appropriate tools and conduct nuanced analyses based on tool outputs. Experiments across seven prominent conditional image generation tasks demonstrate that CIGEval (GPT-4o version) achieves a high correlation of 0.4625 with human assessments, closely matching the inter-annotator correlation of 0.47. Moreover, when implemented with 7B open-source LMMs using only 2.3K training trajectories, CIGEval surpasses the previous GPT-4o-based state-of-the-art method. Case studies on GPT-4o image generation highlight CIGEval's capability in identifying subtle issues related to subject consistency and adherence to control guidance, indicating its great potential for automating evaluation of image generation tasks with human-level reliability.

Towards Reasoning in Large Language Models via Multi-Agent Peer Review Collaboration

Large Language Models (LLMs) have shown remarkable capabilities in general natural language processing tasks but often fall short in complex reasoning tasks. Recent studies have explored human-like problem-solving strategies, such as self-correct, to push further the boundary of single-model reasoning ability. In this work, we let a single model "step outside the box" by engaging multiple models to correct each other. We introduce a multi-agent collaboration strategy that emulates the academic peer review process. Each agent independently constructs its own solution, provides reviews on the solutions of others, and assigns confidence levels to its reviews. Upon receiving peer reviews, agents revise their initial solutions. Extensive experiments on three different types of reasoning tasks show that our collaboration approach delivers superior accuracy across all ten datasets compared to existing methods. Further study demonstrates the effectiveness of integrating confidence in the reviews for math reasoning, and suggests a promising direction for human-mimicking multi-agent collaboration process.

A Read-and-Select Framework for Zero-shot Entity Linking

Zero-shot entity linking (EL) aims at aligning entity mentions to unseen entities to challenge the generalization ability. Previous methods largely focus on the candidate retrieval stage and ignore the essential candidate ranking stage, which disambiguates among entities and makes the final linking prediction. In this paper, we propose a read-and-select (ReS) framework by modeling the main components of entity disambiguation, i.e., mention-entity matching and cross-entity comparison. First, for each candidate, the reading module leverages mention context to output mention-aware entity representations, enabling mention-entity matching. Then, in the selecting module, we frame the choice of candidates as a sequence labeling problem, and all candidate representations are fused together to enable cross-entity comparison. Our method achieves the state-of-the-art performance on the established zero-shot EL dataset ZESHEL with a 2.55% micro-average accuracy gain, with no need for laborious multi-phase pre-training used in most of the previous work, showing the effectiveness of both mention-entity and cross-entity interaction.

Revisiting Sparse Retrieval for Few-shot Entity Linking

Entity linking aims to link ambiguous mentions to their corresponding entities in a knowledge base. One of the key challenges comes from insufficient labeled data for specific domains. Although dense retrievers have achieved excellent performance on several benchmarks, their performance decreases significantly when only a limited amount of in-domain labeled data is available. In such few-shot setting, we revisit the sparse retrieval method, and propose an ELECTRA-based keyword extractor to denoise the mention context and construct a better query expression. For training the extractor, we propose a distant supervision method to automatically generate training data based on overlapping tokens between mention contexts and entity descriptions. Experimental results on the ZESHEL dataset demonstrate that the proposed method outperforms state-of-the-art models by a significant margin across all test domains, showing the effectiveness of keyword-enhanced sparse retrieval.

Generative Multimodal Entity Linking

Multimodal Entity Linking (MEL) is the task of mapping mentions with multimodal contexts to the referent entities from a knowledge base (e.g., Wikipedia). Prior MEL methods mainly focus on designing complex multimodal interaction mechanisms and require fine-tuning all model parameters, which can be prohibitively costly and difficult to scale in the era of Large Language Models (LLMs). In this work, we propose GEMEL, a simple yet effective Generative Multimodal Entity Linking method, which leverages the capabilities of LLMs from large-scale pre-training to directly generate target entity names. We keep the vision and language model frozen and only train a linear layer to enable cross-modality interactions. To adapt LLMs to the MEL task, we take advantage of the emerging in-context learning (ICL) capability of LLMs by retrieving multimodal instances as demonstrations. Extensive experiments show that with only ~0.3% of the model parameters fine-tuned, GEMEL achieves state-of-the-art results on two well-established MEL datasets (4.1% accuracy gains on WikiDiverse and 15.4% accuracy gains on WikiMEL). Our approach is compatible with any off-the-shelf language model, paving the way towards an efficient and general solution for utilizing LLMs in the MEL task.

ExplainCPE: A Free-text Explanation Benchmark of Chinese Pharmacist Examination

As ChatGPT and GPT-4 spearhead the development of Large Language Models (LLMs), more researchers are investigating their performance across various tasks. But more research needs to be done on the interpretability capabilities of LLMs, that is, the ability to generate reasons after an answer has been given. Existing explanation datasets are mostly English-language general knowledge questions, which leads to insufficient thematic and linguistic diversity. To address the language bias and lack of medical resources in generating rationales QA datasets, we present ExplainCPE (over 7k instances), a challenging medical benchmark in Simplified Chinese. We analyzed the errors of ChatGPT and GPT-4, pointing out the limitations of current LLMs in understanding text and computational reasoning. During the experiment, we also found that different LLMs have different preferences for in-context learning. ExplainCPE presents a significant challenge, but its potential for further investigation is promising, and it can be used to evaluate the ability of a model to generate explanations. AI safety and trustworthiness need more attention, and this work makes the first step to explore the medical interpretability of LLMs.The dataset is available at https://github.com/HITsz-TMG/ExplainCPE.

Hansel: A Chinese Few-Shot and Zero-Shot Entity Linking Benchmark

Modern Entity Linking (EL) systems entrench a popularity bias, yet there is no dataset focusing on tail and emerging entities in languages other than English. We present Hansel, a new benchmark in Chinese that fills the vacancy of non-English few-shot and zero-shot EL challenges. The test set of Hansel is human annotated and reviewed, created with a novel method for collecting zero-shot EL datasets. It covers 10K diverse documents in news, social media posts and other web articles, with Wikidata as its target Knowledge Base. We demonstrate that the existing state-of-the-art EL system performs poorly on Hansel (R@1 of 36.6% on Few-Shot). We then establish a strong baseline that scores a R@1 of 46.2% on Few-Shot and 76.6% on Zero-Shot on our dataset. We also show that our baseline achieves competitive results on TAC-KBP2015 Chinese Entity Linking task.