arXiv subjects

Si Wei

Publications and source records attributed to Si Wei.

At least 19 recordsLinked to original sources

arXiv2026

ReDiPPO: Reference-Guided Value Calibration and Discrepancy-Aware Token Reweighting for Mathematical Reasoning

Reinforcement learning has emerged as an effective paradigm for enhancing the mathematical reasoning capabilities of large language models. Among existing policy optimization methods, Proximal Policy Optimization (PPO) remains particularly appealing because its learned critic can, in principle, provide token-level credit assignment. However, in mathematical reasoning tasks characterized by long reasoning horizons and sparse outcome rewards, reliable token-level credit assignment remains challenging. The standard critic often fails to accurately evaluate intermediate reasoning states, resulting in noisy advantage estimates and suboptimal policy updates. In this paper, we propose ReDiPPO, a Reference-guided and Discrepancy-aware PPO framework for mathematical reasoning. ReDiPPO introduces a reference-guided critic that uses reference answers as training-time privileged signals to provide more accurate value estimation. Meanwhile, it retains a standard critic and quantifies the token-level reference-standard discrepancy between the standard value estimate and the reference-guided value estimate. This discrepancy serves as an indicator of difficult reasoning states and is used to reweight the corresponding token-level advantages during PPO optimization. Extensive experiments on diverse mathematical reasoning benchmarks demonstrate that ReDiPPO improves value-estimation accuracy and consistently outperforms strong policy optimization baselines, including PPO, DAPO, and GSPO, in final reasoning performance. Our code is available on https://github.com/cii030/ReDiPPO.

cs.AI↗

arXiv2025

An Efficient LLM-based Evolutional Recommendation with Locate-Forget-Update Paradigm

Nowadays, Large Language Models (LLMs) have shown exceptional performance in sequential recommendations, and the adoption of LLM-based recommender systems (LLMRec) is becoming increasingly widespread in existing e-commerce platforms. Despite the impressive performance, the constant high volume of new user-item interactions makes it difficult to adapt to the evolution of user preference over time, especially for LLM-based recommender systems. The challenge arises from the large number of parameters in LLMs, which makes traditional evolution methods (i.e., Re-training or Fine-tuning) impractical. Specifically, Re-training with all interactions results in prohibitively high computational costs. On the other hand, fine-tuning with only new interactions leads to preference forgetting among inactive users, ultimately compromising overall performance. To tackle this problem, we propose EvoRec, an efficient Locate-Forget-Update framework designed for LLM-based recommender systems to model the evolution of user preferences. EvoRec identifies a small set of parameters associated with preference changes and updates them precisely, thereby saving computational resources while maintaining strong recommendation performance. Notably, the modified parameters account for only 30\% of LoRA adapter parameters, with no additional parameters introduced. Extensive experiments on two real-world datasets demonstrate that, compared to existing methods, EvoRec not only efficiently evolves LLMRec to adapt to the preferences of active users, but also preserves the interests of inactive users from being disturbed during evolution.

cs.IR↗

arXiv2025

Spark-Prover-X1: Formal Theorem Proving Through Diverse Data Training

Large Language Models (LLMs) have shown significant promise in automated theorem proving, yet progress is often constrained by the scarcity of diverse and high-quality formal language data. To address this issue, we introduce Spark-Prover-X1, a 7B parameter model trained via an three-stage framework designed to unlock the reasoning potential of more accessible and moderately-sized LLMs. The first stage infuses deep knowledge through continuous pre-training on a broad mathematical corpus, enhanced by a suite of novel data tasks. Key innovation is a "CoT-augmented state prediction" task to achieve fine-grained reasoning. The second stage employs Supervised Fine-tuning (SFT) within an expert iteration loop to specialize both the Spark-Prover-X1-7B and Spark-Formalizer-X1-7B models. Finally, a targeted round of Group Relative Policy Optimization (GRPO) is applied to sharpen the prover's capabilities on the most challenging problems. To facilitate robust evaluation, particularly on problems from real-world examinations, we also introduce ExamFormal-Bench, a new benchmark dataset of 402 formal problems. Experimental results demonstrate that Spark-Prover achieves state-of-the-art performance among similarly-sized open-source models within the "Whole-Proof Generation" paradigm. It shows exceptional performance on difficult competition benchmarks, notably solving 27 problems on PutnamBench (pass@32) and achieving 24.0\% on CombiBench (pass@32). Our work validates that this diverse training data and progressively refined training pipeline provides an effective path for enhancing the formal reasoning capabilities of lightweight LLMs. We will release both Spark-Prover-X1-7B and Spark-Formalizer-X1-7B, along with the ExamFormal-Bench dataset, in the near future.

cs.CL↗

arXiv2025

CARE-Bench: A Benchmark of Diverse Client Simulations Guided by Expert Principles for Evaluating LLMs in Psychological Counseling

The mismatch between the growing demand for psychological counseling and the limited availability of services has motivated research into the application of Large Language Models (LLMs) in this domain. Consequently, there is a need for a robust and unified benchmark to assess the counseling competence of various LLMs. Existing works, however, are limited by unprofessional client simulation, static question-and-answer evaluation formats, and unidimensional metrics. These limitations hinder their effectiveness in assessing a model's comprehensive ability to handle diverse and complex clients. To address this gap, we introduce \textbf{CARE-Bench}, a dynamic and interactive automated benchmark. It is built upon diverse client profiles derived from real-world counseling cases and simulated according to expert guidelines. CARE-Bench provides a multidimensional performance evaluation grounded in established psychological scales. Using CARE-Bench, we evaluate several general-purpose LLMs and specialized counseling models, revealing their current limitations. In collaboration with psychologists, we conduct a detailed analysis of the reasons for LLMs' failures when interacting with clients of different types, which provides directions for developing more comprehensive, universal, and effective counseling models.

cs.CL↗

arXiv2025

WeaveRec: An LLM-Based Cross-Domain Sequential Recommendation Framework with Model Merging

Cross-Domain Sequential Recommendation (CDSR) seeks to improve user preference modeling by transferring knowledge from multiple domains. Despite the progress made in CDSR, most existing methods rely on overlapping users or items to establish cross-domain correlations-a requirement that rarely holds in real-world settings. The advent of large language models (LLM) and model-merging techniques appears to overcome this limitation by unifying multi-domain data without explicit overlaps. Yet, our empirical study shows that naively training an LLM on combined domains-or simply merging several domain-specific LLMs-often degrades performance relative to a model trained solely on the target domain. To address these challenges, we first experimentally investigate the cause of suboptimal performance in LLM-based cross-domain recommendation and model merging. Building on these insights, we introduce WeaveRec, which cross-trains multiple LoRA modules with source and target domain data in a weaving fashion, and fuses them via model merging. WeaveRec can be extended to multi-source domain scenarios and notably does not introduce additional inference-time cost in terms of latency or memory. Furthermore, we provide a theoretical guarantee that WeaveRec can reduce the upper bound of the expected error in the target domain. Extensive experiments on single-source, multi-source, and cross-platform cross-domain recommendation scenarios validate that WeaveRec effectively mitigates performance degradation and consistently outperforms baseline approaches in real-world recommendation tasks.

cs.IR↗

arXiv2025

MoRE: A Mixture of Low-Rank Experts for Adaptive Multi-Task Learning

With the rapid development of Large Language Models (LLMs), Parameter-Efficient Fine-Tuning (PEFT) methods have gained significant attention, which aims to achieve efficient fine-tuning of LLMs with fewer parameters. As a representative PEFT method, Low-Rank Adaptation (LoRA) introduces low-rank matrices to approximate the incremental tuning parameters and achieves impressive performance over multiple scenarios. After that, plenty of improvements have been proposed for further improvement. However, these methods either focus on single-task scenarios or separately train multiple LoRA modules for multi-task scenarios, limiting the efficiency and effectiveness of LoRA in multi-task scenarios. To better adapt to multi-task fine-tuning, in this paper, we propose a novel Mixture of Low-Rank Experts (MoRE) for multi-task PEFT. Specifically, instead of using an individual LoRA for each task, we align different ranks of LoRA module with different tasks, which we named low-rank experts. Moreover, we design a novel adaptive rank selector to select the appropriate expert for each task. By jointly training low-rank experts, MoRE can enhance the adaptability and efficiency of LoRA in multi-task scenarios. Finally, we conduct extensive experiments over multiple multi-task benchmarks along with different LLMs to verify model performance. Experimental results demonstrate that compared to traditional LoRA and its variants, MoRE significantly improves the performance of LLMs in multi-task scenarios and incurs no additional inference cost. We also release the model and code to facilitate the community.

cs.LG↗

arXiv2025

How Does Sequence Modeling Architecture Influence Base Capabilities of Pre-trained Language Models? Exploring Key Architecture Design Principles to Avoid Base Capabilities Degradation

Pre-trained language models represented by the Transformer have been proven to possess strong base capabilities, and the representative self-attention mechanism in the Transformer has become a classic in sequence modeling architectures. Different from the work of proposing sequence modeling architecture to improve the efficiency of attention mechanism, this work focuses on the impact of sequence modeling architectures on base capabilities. Specifically, our concern is: How exactly do sequence modeling architectures affect the base capabilities of pre-trained language models? In this work, we first point out that the mixed domain pre-training setting commonly adopted in existing architecture design works fails to adequately reveal the differences in base capabilities among various architectures. To address this, we propose a limited domain pre-training setting with out-of-distribution testing, which successfully uncovers significant differences in base capabilities among architectures at an early stage. Next, we analyze the base capabilities of stateful sequence modeling architectures, and find that they exhibit significant degradation in base capabilities compared to the Transformer. Then, through a series of architecture component analysis, we summarize a key architecture design principle: A sequence modeling architecture need possess full-sequence arbitrary selection capability to avoid degradation in base capabilities. Finally, we empirically validate this principle using an extremely simple Top-1 element selection architecture and further generalize it to a more practical Top-1 chunk selection architecture. Experimental results demonstrate our proposed sequence modeling architecture design principle and suggest that our work can serve as a valuable reference for future architecture improvements and novel designs.

cs.CL↗

arXiv2021

SemEval-2021 Task 4: Reading Comprehension of Abstract Meaning

This paper introduces the SemEval-2021 shared task 4: Reading Comprehension of Abstract Meaning (ReCAM). This shared task is designed to help evaluate the ability of machines in representing and understanding abstract concepts. Given a passage and the corresponding question, a participating system is expected to choose the correct answer from five candidates of abstract concepts in a cloze-style machine reading comprehension setup. Based on two typical definitions of abstractness, i.e., the imperceptibility and nonspecificity, our task provides three subtasks to evaluate the participating models. Specifically, Subtask 1 aims to evaluate how well a system can model concepts that cannot be directly perceived in the physical world. Subtask 2 focuses on models' ability in comprehending nonspecific concepts located high in a hypernym hierarchy given the context of a passage. Subtask 3 aims to provide some insights into models' generalizability over the two types of abstractness. During the SemEval-2021 official evaluation period, we received 23 submissions to Subtask 1 and 28 to Subtask 2. The participating teams additionally made 29 submissions to Subtask 3. The leaderboard and competition website can be found at https://competitions.codalab.org/competitions/26153. The data and baseline code are available at https://github.com/boyuanzheng010/SemEval2021-Reading-Comprehension-of-Abstract-Meaning.

cs.CL↗

arXiv2020

Speaker-Aware BERT for Multi-Turn Response Selection in Retrieval-Based Chatbots

In this paper, we study the problem of employing pre-trained language models for multi-turn response selection in retrieval-based chatbots. A new model, named Speaker-Aware BERT (SA-BERT), is proposed in order to make the model aware of the speaker change information, which is an important and intrinsic property of multi-turn dialogues. Furthermore, a speaker-aware disentanglement strategy is proposed to tackle the entangled dialogues. This strategy selects a small number of most important utterances as the filtered context according to the speakers' information in them. Finally, domain adaptation is performed to incorporate the in-domain knowledge into pre-trained language models. Experiments on five public datasets show that our proposed model outperforms the present models on all metrics by large margins and achieves new state-of-the-art performances for multi-turn response selection.

cs.CL↗

arXiv2020

DialBERT: A Hierarchical Pre-Trained Model for Conversation Disentanglement

Disentanglement is a problem in which multiple conversations occur in the same channel simultaneously, and the listener should decide which utterance is part of the conversation he will respond to. We propose a new model, named Dialogue BERT (DialBERT), which integrates local and global semantics in a single stream of messages to disentangle the conversations that mixed together. We employ BERT to capture the matching information in each utterance pair at the utterance-level, and use a BiLSTM to aggregate and incorporate the context-level information. With only a 3% increase in parameters, a 12% improvement has been attained in comparison to BERT, based on the F1-Score. The model achieves a state-of-the-art result on the a new dataset proposed by IBM and surpasses previous work by a substantial margin.

cs.CL↗

arXiv2019

Several Experiments on Investigating Pretraining and Knowledge-Enhanced Models for Natural Language Inference

Natural language inference (NLI) is among the most challenging tasks in natural language understanding. Recent work on unsupervised pretraining that leverages unsupervised signals such as language-model and sentence prediction objectives has shown to be very effective on a wide range of NLP problems. It would still be desirable to further understand how it helps NLI; e.g., if it learns artifacts in data annotation or instead learn true inference knowledge. In addition, external knowledge that does not exist in the limited amount of NLI training data may be added to NLI models in two typical ways, e.g., from human-created resources or an unsupervised pretraining paradigm. We runs several experiments here to investigate whether they help NLI in the same way, and if not,how?

cs.CL↗

arXiv2019

Exploring Unsupervised Pretraining and Sentence Structure Modelling for Winograd Schema Challenge

Winograd Schema Challenge (WSC) was proposed as an AI-hard problem in testing computers' intelligence on common sense representation and reasoning. This paper presents the new state-of-theart on WSC, achieving an accuracy of 71.1%. We demonstrate that the leading performance benefits from jointly modelling sentence structures, utilizing knowledge learned from cutting-edge pretraining models, and performing fine-tuning. We conduct detailed analyses, showing that fine-tuning is critical for achieving the performance, but it helps more on the simpler associative problems. Modelling sentence dependency structures, however, consistently helps on the harder non-associative subset of WSC. Analysis also shows that larger fine-tuning datasets yield better performances, suggesting the potential benefit of future work on annotating more Winograd schema sentences.

cs.CL↗

arXiv2018

Spelling Error Correction Using a Nested RNN Model and Pseudo Training Data

We propose a nested recurrent neural network (nested RNN) model for English spelling error correction and generate pseudo data based on phonetic similarity to train it. The model fuses orthographic information and context as a whole and is trained in an end-to-end fashion. This avoids feature engineering and does not rely on a noisy channel model as in traditional methods. Experiments show that the proposed method is superior to existing systems in correcting spelling errors.

cs.CL↗

arXiv2018

Neural Natural Language Inference Models Enhanced with External Knowledge

Modeling natural language inference is a very challenging task. With the availability of large annotated data, it has recently become feasible to train complex models such as neural-network-based inference models, which have shown to achieve the state-of-the-art performance. Although there exist relatively large annotated data, can machines learn all knowledge needed to perform natural language inference (NLI) from these data? If not, how can neural-network-based NLI models benefit from external knowledge and how to build NLI models to leverage it? In this paper, we enrich the state-of-the-art neural natural language inference models with external knowledge. We demonstrate that the proposed models improve neural NLI models to achieve the state-of-the-art performance on the SNLI and MultiNLI datasets.

cs.CL↗

arXiv2017

Recurrent Neural Network-Based Sentence Encoder with Gated Attention for Natural Language Inference

The RepEval 2017 Shared Task aims to evaluate natural language understanding models for sentence representation, in which a sentence is represented as a fixed-length vector with neural networks and the quality of the representation is tested with a natural language inference task. This paper describes our system (alpha) that is ranked among the top in the Shared Task, on both the in-domain test set (obtaining a 74.9% accuracy) and on the cross-domain test set (also attaining a 74.9% accuracy), demonstrating that the model generalizes well to the cross-domain data. Our model is equipped with intra-sentence gated-attention composition which helps achieve a better performance. In addition to submitting our model to the Shared Task, we have also tested it on the Stanford Natural Language Inference (SNLI) dataset. We obtain an accuracy of 85.5%, which is the best reported result on SNLI when cross-sentence attention is not allowed, the same condition enforced in RepEval 2017.

cs.CL↗

arXiv2017

Attention-over-Attention Neural Networks for Reading Comprehension

Cloze-style queries are representative problems in reading comprehension. Over the past few months, we have seen much progress that utilizing neural network approach to solve Cloze-style questions. In this paper, we present a novel model called attention-over-attention reader for the Cloze-style reading comprehension task. Our model aims to place another attention mechanism over the document-level attention, and induces "attended attention" for final predictions. Unlike the previous works, our neural network model requires less pre-defined hyper-parameters and uses an elegant architecture for modeling. Experimental results show that the proposed attention-over-attention model significantly outperforms various state-of-the-art systems by a large margin in public datasets, such as CNN and Children's Book Test datasets.

cs.CL↗

arXiv2017

Enhanced LSTM for Natural Language Inference

Reasoning and inference are central to human and artificial intelligence. Modeling inference in human language is very challenging. With the availability of large annotated data (Bowman et al., 2015), it has recently become feasible to train neural network based inference models, which have shown to be very effective. In this paper, we present a new state-of-the-art result, achieving the accuracy of 88.6% on the Stanford Natural Language Inference Dataset. Unlike the previous top models that use very complicated network architectures, we first demonstrate that carefully designing sequential inference models based on chain LSTMs can outperform all previous models. Based on this, we further show that by explicitly considering recursive architectures in both local inference modeling and inference composition, we achieve additional improvement. Particularly, incorporating syntactic parsing information contributes to our best result---it further improves the performance even when added to the already very strong model.

cs.CL↗

arXiv2017

Exploring Question Understanding and Adaptation in Neural-Network-Based Question Answering

The last several years have seen intensive interest in exploring neural-network-based models for machine comprehension (MC) and question answering (QA). In this paper, we approach the problems by closely modelling questions in a neural network framework. We first introduce syntactic information to help encode questions. We then view and model different types of questions and the information shared among them as an adaptation task and proposed adaptation models for them. On the Stanford Question Answering Dataset (SQuAD), we show that these approaches can help attain better results over a competitive baseline.

cs.CL↗