Beichen Zhang

An Yang, Beichen Zhang, Binyuan Hui et al.

In this report, we present a series of math-specific large language models: Qwen2.5-Math and Qwen2.5-Math-Instruct-1.5B/7B/72B. The core innovation of the Qwen2.5 series lies in integrating the philosophy of self-improvement throughout the entire pipeline, from pre-training and post-training to inference: (1) During the pre-training phase, Qwen2-Math-Instruct is utilized to generate large-scale, high-quality mathematical data. (2) In the post-training phase, we develop a reward model (RM) by conducting massive sampling from Qwen2-Math-Instruct. This RM is then applied to the iterative evolution of data in supervised fine-tuning (SFT). With a stronger SFT model, it's possible to iteratively train and update the RM, which in turn guides the next round of SFT data iteration. On the final SFT model, we employ the ultimate RM for reinforcement learning, resulting in the Qwen2.5-Math-Instruct. (3) Furthermore, during the inference stage, the RM is used to guide sampling, optimizing the model's performance. Qwen2.5-Math-Instruct supports both Chinese and English, and possess advanced mathematical reasoning capabilities, including Chain-of-Thought (CoT) and Tool-Integrated Reasoning (TIR). We evaluate our models on 10 mathematics datasets in both English and Chinese, such as GSM8K, MATH, GaoKao, AMC23, and AIME24, covering a range of difficulties from grade school level to math competition problems.

Beichen Zhang, Frank Schilder, Kelly Helm Smith et al.

Acquiring a better understanding of drought impacts becomes increasingly vital under a warming climate. Traditional drought indices describe mainly biophysical variables and not impacts on social, economic, and environmental systems. We utilized natural language processing and bidirectional encoder representation from Transformers (BERT) based transfer learning to fine-tune the model on the data from the news-based Drought Impact Report (DIR) and then apply it to recognize seven types of drought impacts based on the filtered Twitter data from the United States. Our model achieved a satisfying macro-F1 score of 0.89 on the DIR test set. The model was then applied to California tweets and validated with keyword-based labels. The macro-F1 score was 0.58. However, due to the limitation of keywords, we also spot-checked tweets with controversial labels. 83.5% of BERT labels were correct compared to the keyword labels. Overall, the fine-tuned BERT-based recognizer provided proper predictions and valuable information on drought impacts. The interpretation and analysis of the model were consistent with experiential domain expertise.

Beichen Zhang, Fatima K. Abu Salem, Michael J. Hayes et al.

Under climate change, the increasing frequency, intensity, and spatial extent of drought events lead to higher socio-economic costs. However, the relationships between the hydro-meteorological indicators and drought impacts are not identified well yet because of the complexity and data scarcity. In this paper, we proposed a framework based on the extreme gradient model (XGBoost) for Texas to predict multi-category drought impacts and connected a typical drought indicator, Standardized Precipitation Index (SPI), to the text-based impacts from the Drought Impact Reporter (DIR). The preliminary results of this study showed an outstanding performance of the well-trained models to assess drought impacts on agriculture, fire, society & public health, plants & wildlife, as well as relief, response & restrictions in Texas. It also provided a possibility to appraise drought impacts using hydro-meteorological indicators with the proposed framework in the United States, which could help drought risk management by giving additional information and improving the updating frequency of drought impacts. Our interpretation results using the Shapley additive explanation (SHAP) interpretability technique revealed that the rules guiding the predictions of XGBoost comply with domain expertise knowledge around the role that SPI indicators play around drought impacts.

Chujie Zheng, Zhenru Zhang, Beichen Zhang et al. · tsinghua

As language models regularly make mistakes when solving math problems, automated identification of errors in the reasoning process becomes increasingly significant for their scalable oversight. In this paper, we introduce ProcessBench for measuring the ability to identify erroneous steps in mathematical reasoning. It consists of 3,400 test cases, primarily focused on competition- and Olympiad-level math problems. Each test case contains a step-by-step solution with error location annotated by human experts. Models are required to identify the earliest step that contains an error, or conclude that all steps are correct. We conduct extensive evaluation on ProcessBench, involving two types of models: process reward models (PRMs) and critic models, where for the latter we prompt general language models to critique each solution step by step. We draw two main observations: (1) Existing PRMs typically fail to generalize to more challenging math problems beyond GSM8K and MATH. They underperform both critic models (i.e., prompted general language models) and our own trained PRM that is straightforwardly fine-tuned on the PRM800K dataset. (2) The best open-source model, QwQ-32B-Preview, has demonstrated the critique capability competitive with the proprietary model GPT-4o, despite that it still lags behind the reasoning-specialized o1-mini. We hope ProcessBench can foster future research in reasoning process assessment, paving the way toward scalable oversight of language models.

Zhipeng Chen, Yingqian Min, Beichen Zhang et al.

In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore, we show that even when a model like DeepSeek-R1-Distill-Qwen-1.5B has already achieved a high performance level, it can be further refined through RL training, reaching an accuracy of 39.33% on AIME 2024. Beyond RL training, we also explore the use of tool manipulation, finding that it significantly boosts the reasoning performance of large reasoning models. This approach achieves a remarkable accuracy of 86.67% with greedy search on AIME 2024, underscoring its effectiveness in enhancing model capabilities. We release our resources at the STILL project website: https://github.com/RUCAIBox/Slow_Thinking_with_LLMs.

Beichen Zhang, Huiqi Wang, Amani Alabri et al.

The accurate characterization of the severity of the wildfire event strongly contributes to the characterization of the fuel conditions in fire-prone areas, and provides valuable information for disaster response. The aim of this study is to develop an autonomous system built on top of high-resolution multispectral satellite imagery, with an advanced deep learning method for detecting burned area change. This work proposes an initial exploration of using an unsupervised model for feature extraction in wildfire scenarios. It is based on the contrastive learning technique SimCLR, which is trained to minimize the cosine distance between augmentations of images. The distance between encoded images can also be used for change detection. We propose changes to this method that allows it to be used for unsupervised burned area detection and following downstream tasks. We show that our proposed method outperforms the tested baseline approaches.