Abstract
Zero-shot text classification (ZSC) offers the promise of eliminating costly task-specific annotation by matching texts directly to human-readable label descriptions. While early approaches have predominantly relied on cross-encoder models fine-tuned for natural language inference (NLI), recent advances in text-embedding models, rerankers, and instruction-tuned large language models (LLMs) have challenged the dominance of NLI-based architectures. Yet, systematically comparing these diverse approaches remains difficult. Existing evaluations, such as MTEB, often incorporate labeled examples through supervised probes or fine-tuning, leaving genuine zero-shot capabilities underexplored. To address this, we introduce BTZSC, a comprehensive benchmark of 22 public datasets spanning sentiment, topic, intent, and emotion classification, capturing diverse domains, class cardinalities, and document lengths. Leveraging BTZSC, we conduct a systematic comparison across four major model families, NLI cross-encoders, embedding models, rerankers and instruction-tuned LLMs, encompassing 38 public and custom checkpoints. Our results show that: (i) modern rerankers, exemplified by Qwen3-Reranker-8B, set a new state-of-the-art with macro F1 = 0.72; (ii) strong embedding models such as GTE-large-en-v1.5 substantially close the accuracy gap while offering the best trade-off between accuracy and latency; (iii) instruction-tuned LLMs at 4--12B parameters achieve competitive performance (macro F1 up to 0.67), excelling particularly on topic classification but trailing specialized rerankers; (iv) NLI cross-encoders plateau even as backbone size increases; and (v) scaling primarily benefits rerankers and LLMs over embedding models. BTZSC and accompanying evaluation code are publicly released to support fair and reproducible progress in zero-shot text understanding.
Abstract (translated)
零样本文本分类（ZSC）提供了一种希望，即通过将文本直接匹配到可读的标签描述来消除耗资的任务特定标注。虽然早期的方法主要依赖于为自然语言推理（NLI）微调的交叉编码器模型，但最近在文本嵌入模型、重排序器和指令调整的大规模语言模型（LLM）方面的进步挑战了基于NLI架构的主导地位。然而，系统地比较这些多样化的方法仍然困难重重。现有的评估方法，如MTEB，通常通过监督探针或微调来整合带标签的例子，这使得真正意义上的零样本能力被忽视了。为了解决这一问题，我们介绍了BTZSC，这是一个全面的基准测试集合，包括22个公共数据集，涵盖了情绪、主题、意图和情感分类，覆盖了多样化的领域、类别数量以及文档长度。利用BTZSC，我们在四种主要模型家族之间进行了系统比较：NLI交叉编码器、嵌入模型、重排序器以及指令调整的LLM，包括38个公共和自定义检查点。 我们的结果表明： (i) 现代重排序器，如Qwen3-Reranker-8B，在宏观F1分数上达到了0.72的新最高水平； (ii) 强大的嵌入模型，例如GTE-large-en-v1.5，显著缩小了准确性的差距，并且在精度和延迟之间提供了最佳的权衡； (iii) 参数量为4至12B的指令调整LLM取得了竞争性性能（宏观F1分数最高达0.67），尤其擅长主题分类，但不及专门化的重排序器的表现； (iv) NLI交叉编码器即使随着骨干模型规模的增长也表现停滞； (v) 规模化主要有利于重排序器和LLM超过嵌入模型。 BTZSC及其伴随的评估代码已公开发布，以支持零样本文本理解中的公正且可复现的进步。
URL
https://arxiv.org/abs/2603.11991
