Recent studies have raised significant concerns regarding the reliability of current mathematical benchmarks, highlighting key limitations such as simplistic design and potential data contamination that undermine evaluation accuracy. Consequently, developing a reliable benchmark that effectively evaluates large language models' (LLMs) genuine capabilities in mathematical reasoning remains a critical challenge. To address these concerns, we propose RV-Bench, a novel evaluation methodology for Benchmarking LLMs with Random Variables in mathematical reasoning. Specifically, we develop question-generating functions to produce random variable questions (RVQs), whose background content mirrors the original benchmark problems, but with randomized variable combinations, rendering them "unseen" to LLMs. Models must completely understand the inherent question pattern to correctly answer RVQs with diverse variable combinations. Thus, an LLMs' genuine reasoning capability is reflected through its accuracy and robustness on RV-Bench. We conducted extensive experiments on over 30 representative LLMs across more than 1,000 RVQs. Our findings reveal that LLMs exhibit a proficiency imbalance between encountered and "unseen" data distributions. Furthermore, RV-Bench reveals that proficiency generalization across similar mathematical reasoning tasks is limited, but we verified that it can still be effectively elicited through test-time scaling.

Chuang Zhou, Junnan Dong, Yilin Xiao, Shengyuan Chen, Su Dong, di Yin, Xing Sun, Zhaozhuo Xu, Xiao Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Pretrained large models, particularly large language models, have garnered increasing attention, as they have demonstrated remarkable abilities through contextual learning. Pretrained large models are increasingly recognized as fundamental tools for solving various tasks. However, the substantial computational demands of large models have dissuaded most product teams and individuals from running them. In such scenarios, to leverage the exceptional performance of large models, one must solely depend on costly APIs, further burdening product teams and individuals. On the other hand, despite the overall inferior performance of small models compared to large models, there are certain distributions where small models can achieve comparable or even superior results. For instance, during training, small models may become trapped in a local optimum that is unique to certain distributions, leading to superior performance. Hence, we propose Data Shunt (DS), a general paradigm for collaboration of small and large models. DS not only substantially reduces the cost associated with deploying large models but also effectively enhances overall performance. Specifically, DS determines the shunting direction by evaluating the confidence level of small models. When the confidence level falls below a specific threshold, the input data is forwarded to large models. To further leverage the advantages of the small and large models, we introduce Prompt Pruning (PP) and 2-Stage Confidence Distillation (2CD), which facilitate mutual collaboration, leading to better results and less cost. The remarkable performance across diverse modalities and tasks demonstrates the superiority of the proposed DS over large models. For instance, ChatGPT achieves an accuracy of 94.43% on Amazon Product sentiment analysis, and DS achieves an accuracy of 95.64%, while the cost has been reduced to only 31.18%. The code for the proposed method are provided for research purposes https://github.com/Anfeather/Data-Shunt.

Recently, speech assistants have brought convenience to our lives from many aspects. In the education field, speech assistants can also help teachers to reduce their burdens. However, there is no suitable solution to synthesize speeches for mathematical formulas although there have been lots of good techniques for text-to-speech (TTS) in the general domain. One possible solution is that we can convert mathematical formulas expressed in the LaTeX format to spelling texts and synthesize them into speech. In this paper, we investigated text generation methods that translate mathematical formulas in LaTex into Chinese spelling texts. For this purpose, we first constructed a parallel corpus of mathematical formulas and Chinese spelling texts, then compared the existing commonly used text generation methods, such as rule-based, Seq2Seq, Transformer and Graph2Seq, and finally proposed a novel model. As far as we know, this is the first study for Chinese spelling text generation of mathematical formulas. Experiment results on the annotated corpus show that our proposed model significantly outperforms the existing commonly used generation models.