Author name cluster

Yaxuan Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

5 papers

2 author rows

EAAI Journal 2026 Journal Article

Dual-channel machine learning proxy for pseudo-two-dimensional model with enhanced extrapolation correction

Yue Cui
Yaxuan Wang
Shilong Guo
Liang Deng
Junfu Li
Lei Zhao
Zhenbo Wang

Physics-based electrochemical models are essential for analyzing and predicting the performance of lithium metal batteries (LMBs), yet their high computational cost restricts their use in real-time applications. To overcome this limitation, a dual-channel agent model (DCAM) is proposed, which decouples the mapping between electrochemical parameters and discharge duration and voltage profile, serving as an efficient surrogate for the pseudo-two-dimensional (P2D) model. Unlike physics-constrained or purely data-driven approaches, DCAM does not rely on explicit physical equations but rather bypasses regions with poor generalization, enabling fast and accurate extrapolation from partial discharge data. Furthermore, a hybrid modeling framework is developed by embedding a multilayer perceptron (MLP) into the Butler–Volmer (B–V) kinetics, replacing the iterative Newton process and thereby improving computational efficiency. Experimental and simulation results demonstrate that the proposed framework accurately reproduces the P2D voltage behavior and achieves high-fidelity extrapolation and robustness under limited-data conditions.

Details DOI

AAAI Conference 2026 Conference Paper

Stabilizing Self-Consuming Diffusion Models with Latent Space Filtering

Zhongteng Cai
Yaxuan Wang
Yang Liu
Xueru Zhang

As synthetic data proliferates across the Internet, it is often reused to train successive generations of generative models. This creates a "self-consuming loop" that can lead to training instability or *model collapse*. Common strategies to address the issue---such as accumulating historical training data or injecting fresh real data---either increase computational cost or require expensive human annotation. In this paper, we empirically analyze the latent space dynamics of self-consuming diffusion models and observe that the low-dimensional structure of latent representations extracted from synthetic data degrade over generations. Based on this insight, we propose *Latent Space Filtering* (LSF), a novel approach that mitigates model collapse by filtering out less realistic synthetic data from mixed datasets. Theoretically, we present a framework that connects latent space degradation to empirical observations. Experimentally, we show that LSF consistently outperforms existing baselines across multiple real-world datasets, effectively mitigating model collapse without increasing training cost or relying on human annotation.

PDF Details DOI

ICLR Conference 2025 Conference Paper

Improving Data Efficiency via Curating LLM-Driven Rating Systems

Jinlong Pang
Jiaheng Wei
Ankit Shah 0001
Zhaowei Zhu
Yaxuan Wang
Chen Qian 0001
Yang Liu 0018
Yujia Bao

Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce $DS^2$, a **D**iversity-aware **S**core curation method for **D**ata **S**election. By systematically modeling error patterns through a score transition matrix, $DS^2$ corrects LLM-based scores and promotes diversity in the selected data samples. Our approach shows that a curated subset (just 3.3\% of the original dataset) outperforms full-scale datasets (300k samples) across various machine-alignment benchmarks, and matches or surpasses human-aligned datasets such as LIMA with the same sample size (1k samples). These findings challenge conventional data scaling assumptions, highlighting that redundant, low-quality samples can degrade performance and reaffirming that ``more can be less''.

Details

ICLR Conference 2025 Conference Paper

LLM Unlearning via Loss Adjustment with Only Forget Data

Yaxuan Wang
Jiaheng Wei
Chris Yuhao Liu
Jinlong Pang
Quan Liu
Ankit Shah 0001
Yujia Bao
Yang Liu 0018

Unlearning in Large Language Models (LLMs) is essential for ensuring ethical and responsible AI use, especially in addressing privacy leak, bias, safety, and evolving regulations. Existing approaches to LLM unlearning often rely on retain data or a reference LLM, yet they struggle to adequately balance unlearning performance with overall model utility. This challenge arises because leveraging explicit retain data or implicit knowledge of retain data from a reference LLM to fine-tune the model tends to blur the boundaries between the forgotten and retain data, as different queries often elicit similar responses. In this work, we propose eliminating the need to retain data or the reference LLM for response calibration in LLM unlearning. Recognizing that directly applying gradient ascent on the forget data often leads to optimization instability and poor performance, our method guides the LLM on what not to respond to, and importantly, how to respond, based on the forget data. Hence, we introduce Forget data only Loss AjustmenT (FLAT), a "flat" loss adjustment approach which addresses these issues by maximizing $f$-divergence between the available template answer and the forget answer only w.r.t. the forget data. The variational form of the defined $f$-divergence theoretically provides a way of loss adjustment by assigning different importance weights for the learning w.r.t. template responses and the forgetting of responses subject to unlearning. Empirical results demonstrate that our approach not only achieves superior unlearning performance compared to existing methods but also minimizes the impact on the model’s retained capabilities, ensuring high utility across diverse tasks, including copyrighted content unlearning on Harry Potter dataset and MUSE Benchmark, and entity unlearning on the TOFU dataset.

Details

NeurIPS Conference 2024 Conference Paper

Large Language Model Unlearning via Embedding-Corrupted Prompts

Chris Y. Liu
Yaxuan Wang
Jeffrey Flanigan
Yang Liu

Large language models (LLMs) have advanced to encompass extensive knowledge across diverse domains. Yet controlling what a large language model should not know is important for ensuring alignment and thus safe use. However, accurately and efficiently unlearning knowledge from an LLM remains challenging due to the potential collateral damage caused by the fuzzy boundary between retention and forgetting, and the large computational requirements for optimization across state-of-the-art models with hundreds of billions of parameters. In this work, we present \textbf{Embedding-COrrupted (ECO) Prompts}, a lightweight unlearning framework for large language models to address both the challenges of knowledge entanglement and unlearning efficiency. Instead of relying on the LLM itself to unlearn, we enforce an unlearned state during inference by employing a prompt classifier to identify and safeguard prompts to forget. We learn corruptions added to prompt embeddings via zeroth order optimization toward the unlearning objective offline and corrupt prompts flagged by the classifier during inference. We find that these embedding-corrupted prompts not only lead to desirable outputs that satisfy the unlearning objective but also closely approximate the output from a model that has never been trained on the data intended for forgetting. Through extensive experiments on unlearning, we demonstrate the superiority of our method in achieving promising unlearning at \textit{nearly zero side effects} in general domains and domains closely related to the unlearned ones. Additionally, we highlight the scalability of our method to 100 LLMs, ranging from 0. 5B to 236B parameters, incurring no additional cost as the number of parameters increases. We have made our code publicly available at \url{https: //github. com/chrisliu298/llm-unlearn-eco}.

PDF Details DOI