Papers by Bei Wang

39 papers
Adaptive Spatial and Temporal Redundancy Optimization for Efficient Reasoning in Large Language Models (2026.acl-long)

Copied to clipboard

Challenge: Existing research to improve CoT efficiency falls into three categories, each with distinct limitations.
Approach: They propose a training-free framework that addresses both dimensions of CoT reasoning by applying a progressive precision reduction strategy coupled with an entropy-based confidence mechanism for adaptive termination.
Outcome: Empirical results show that the proposed framework achieves 11.3 efficiency gain without compromising accuracy.
Detecting Causal Language Use in Science Findings (D19-1)

Copied to clipboard

Challenge: Prior studies on identifying inappropriate use of causal language relied on manual content analysis, which is not scalable for examining a large volume of science publications.
Approach: They developed a prediction model that classifies conclusion sentences into “no relationship”, “correlational”, “conditional causal” and “direct causal” categories.
Outcome: The proposed model can be used to identify the inappropriate use of causal language in scientific publications and news articles.
Weight Distillation: Transferring the Knowledge in Neural Network Parameters (2021.acl-long)

Copied to clipboard

Challenge: Knowledge distillation is an effective method for model acceleration and compression.
Approach: They propose to use parameters to distill knowledge from large neural networks to small ones . they propose to do this by using a parameter generator to transfer the knowledge to a small neural network .
Outcome: The proposed method learns a small network 1.88 2.94x faster than the large network but with competitive BLEU points.
Enhancing LLM Knowledge Learning through Generalization (2025.findings-emnlp)

Copied to clipboard

Challenge: Continued pre-training on paraphrased data has shown empirical promise for enhancing knowledge acquisition, but this approach is costly and unreliable as it relies on external models or manual effort for rewriting.
Approach: They propose formatting-based data augmentation which diversifies documents conveying the same knowledge by altering document formats rather than their content.
Outcome: The proposed methods improve generalization to diverse paraphrased contexts and enhance pre-training and instruction tuning.
ATFormer: A Learned Performance Model with Transfer Learning Across Devices for Deep Learning Tensor Programs (2023.emnlp-main)

Copied to clipboard

Challenge: Compilation-based methods with performance models have poor measurement accuracy and transferability between platforms.
Approach: They propose a compiler that automatically generates tensors and automatically tunes them for different hardware platforms.
Outcome: The proposed model reduces inference time and costs on modern DNN benchmarks.
BTC-LLM: Efficient Sub-1-Bit LLM Quantization via Learnable Transformation and Binary Codebook (2026.acl-long)

Copied to clipboard

Challenge: Recent sparsity-aware binarization approaches can achieve sub-1-bit compression, but they face performance degradation, mask-management overhead, and limited hardware compatibility.
Approach: They propose a binary quantization framework that leverages binary pattern clustering and weight transformation to overcome performance degradation and mask-management overhead.
Outcome: The proposed framework achieves state-of-the-art compression (1.11–0.7 bits) it maintains high performance with only a 3.1% accuracy drop in zero-shot benchmarks while delivering a 1.6 speedup over FP16.
In-Context Example Ordering Guided by Label Distributions (2024.findings-naacl)

Copied to clipboard

Challenge: Prior work has shown that ICL is sensitive to different natural language instructions and different orderings of in-context examples.
Approach: They propose two principles for in-context example ordering guided by model’s probability predictions.
Outcome: The proposed model outperforms baseline models on 13 text classification datasets and nine autoregressive LLMs with 700M to 13B parameters.
Selecting Demonstrations for Many-Shot In-Context Learning via Gradient Matching (2025.findings-acl)

Copied to clipboard

Challenge: In-Context Learning (ICL) empowers Large Language Models for rapid task adaptation without fine-tuning.
Approach: They propose a method that aligns fine-tuning gradients between entire training set and selected examples to enable in-context learning and fine-uning.
Outcome: The proposed method outperforms random selection on large LLMs from 4-shot to 128-shot scenarios across 9 datasets.
Does Multi-Encoder Help? A Case Study on Context-Aware Neural Machine Translation (2020.acl-main)

Copied to clipboard

Challenge: In encoder-decoder neural models, multiple encoders are used to represent contextual information in addition to the individual sentence.
Approach: They propose to use multiple context encoders to encode the individual sentences in document-level neural machine translation (NMT) They propose a noisy dropout setup and a single-encoder approach to encode context sentences.
Outcome: The proposed approach encodes the context and the current sentence without contexts.
Benchmarking and Learning Real-World Customer Service Dialogue (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks and training pipelines for industrial intelligent customer service (ICS) focus on task completion and tool correctness.
Approach: They propose a benchmark-to-optimization loop that bridges offline gains to deployment . they propose OlaMind, which distills reusable reasoning patterns from expert dialogues .
Outcome: The proposed benchmark surpasses GPT-5.2 and Gemini 3 Pro on OlaBench . it delivers an average +23.67% issue resolution and -6.6% human transfer rate versus baseline .
Revealing the Parallel Multilingual Learning within Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Large language models (LLMs) can handle multilingual and cross-lingual text within a single input; however, previous studies focusing on using English as the pivot language to enhance language understanding and reasoning focus on using multiple languages.
Approach: They propose to use parallel multilingual input to enhance the model's comprehension of the input and to examine how multilingual processing affects prediction.
Outcome: The proposed model can handle multilingual and cross-lingual text within a single input, but previous studies focused on using English as the pivot language to enhance language understanding and reasoning.
TCPO: Thought-Centric Preference Optimization for Effective Embodied Decision-making (2025.emnlp-main)

Copied to clipboard

Challenge: Existing post-SFT methods for embodied AI are constrained by sparse rewards and action-only optimization, resulting in low sample efficiency, poor consistency, and model degradation.
Approach: They propose to integrate Thought-Centric Preference Optimization (TCPO) into embodied decision-making by transforming sparse reward signals into richer step sample pairs.
Outcome: The proposed approach achieves an average success rate of 26.67% in the ALFWorld environment, and a 6% improvement over RL4VLM.
Hybrid Alignment Training for Large Language Models (2024.findings-acl)

Copied to clipboard

Challenge: Existing approaches to align large language models with instructions and preferences are conflicting . et al., 2023b) show that hybrid alignment training can outperform baselines .
Approach: They propose a hybrid alignment training approach based on alternating alignment and modified elastic weight consolidation methods to achieve better collaboration between different alignment tasks.
Outcome: The proposed approach outperforms baseline alignment training methods on summarization and dialogue tasks.
Bit-by-Bit: Progressive QAT Strategy with Outlier Channel Splitting for Stable Low-Bit LLMs (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to training LLMs at ultra-low precisions suffer from convergence instability and substantial training costs.
Approach: They propose a progressive QAT framework with outlier channel splitting to address these issues . they use nested structure of integer quantization grids to enable a "train once, deploy any precision" paradigm .
Outcome: The proposed framework outperforms baselines on both Llama2/3 and W2A16, with an 11 speedup over BF16.
ReMamba: Equip Mamba with Effective Long-Sequence Modeling (2025.findings-emnlp)

Copied to clipboard

Challenge: Mamba models demonstrate superior inference efficiency and competitive performance on short-context tasks, but their capacity to comprehend long contexts is limited compared to transformer-based models.
Approach: They propose a model which incorporates selective compression and adaptation techniques within a two-stage re-forward process, incurring minimal additional inference costs overhead.
Outcome: The proposed model improves on the LongBench and L-Eval benchmarks by 3.2 and 1.6 points and attains performance almost on par with same-size transformer models.
Detecting Health Advice in Medical Research Literature (2021.emnlp-main)

Copied to clipboard

Challenge: Health and medical researchers often give clinical and policy recommendations to inform health practice and public health policy.
Approach: They developed a BERT-based prediction model that can predict whether a sentence gives strong advice, weak advice, or not.
Outcome: The proposed model can predict whether a sentence gives strong advice, weak advice, or not with a macro-averaged F1 score of 0.93.
On the Emotion Understanding of Synthesized Speech (2026.acl-long)

Copied to clipboard

Challenge: Existing models for emotion understanding do not capture fundamental features of synthesized speech.
Approach: They evaluate emotion recognition models on synthesized speech using SER models and generative models.
Outcome: The proposed model can't generalize to synthesized speech because of speech token prediction . generative models tend to infer emotion from textual semantics while ignoring paralinguistic cues.
Disentangling Preference Representation and Text Generation for Efficient Individual Preference Alignment (2025.coling-main)

Copied to clipboard

Challenge: Human values are inherently diverse, making it insufficient to align LLMs solely with general preferences.
Approach: They propose a flexible paradigm for individual preference alignment that disentangles preference representation from text generation in LLMs.
Outcome: The proposed method produces aligned quality and better than PEFT-based methods while reducing training time for each new individual preference by 80% to 90%.
When Language Model Meets Private Library (2022.findings-emnlp)

Copied to clipboard

Challenge: Existing language models have been pre-trained on large-scale code corpora and generate decent code snippets.
Approach: They propose a framework that can provide pre-trained language models with the ability to generate code using private libraries.
Outcome: The proposed framework can generate code using private libraries using off-the-shelf language models or pre-trained models on code corpus containing API information.
SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignment (2025.coling-main)

Copied to clipboard

Challenge: Large language models (LLMs) have demonstrated significant improvements in reasoning abilities, but these improvements are primarily focused on English, leading to inferior performance in non-English scenarios.
Approach: They propose a multilingual reasoning alignment approach that fine-tunes the layers responsible for multilingual comprehension in one stage.
Outcome: The proposed method fine-tunes 6 of the 9 layers responsible for multilingual comprehension, while reducing training time by 4.1-11.9 compared to the two-stage method.
Measuring Correlation-to-Causation Exaggeration in Press Releases (2020.coling-main)

Copied to clipboard

Challenge: Recent studies have found that press releases are a major source of exaggeration in science communication, which is later spread to mainstream media.
Approach: They propose an NLP approach to identify exaggerated causal claims in health press releases that report on observational studies.
Outcome: The proposed approach can identify causal claims in press releases that report on observational studies.
MTR-Suite: A Framework for Evaluating and Synthesizing Conversational Retrieval Benchmarks (2026.acl-long)

Copied to clipboard

Challenge: Existing conversational retrieval benchmarks suffer from costly, sparse human annotation or rigid, unnatural automated heuristics.
Approach: They propose a framework for auditing, synthesizing, and benchmarking conversational retrieval.
Outcome: The proposed framework is based on three LLM-based auditors and a multi-agent system . it mimics production-style challenges (hard topic switching, verbosity) and offers superior discriminative power.
IIET: Efficient Numerical Transformer via Implicit Iterative Euler Method (2025.emnlp-main)

Copied to clipboard

Challenge: High-order numerical methods enhance performance in tasks like NLP but introduce a performance-efficiency trade-off due to increased computational overhead.
Approach: They propose an iterative implicit Euler Transformer which simplifies high-order numerical methods by iterating implicit Eule.
Outcome: The proposed method improves accuracy and reduces inference overhead by 55% while retaining 99.4% of the original task accuracy.
Can LLM Agents Simulate Multi-Turn Human Behavior? Evidence from Real Online Customer Behavior Data (2026.acl-long)

Copied to clipboard

Challenge: Recent research shows that LLM Agents can generate “believable” human behaviors via prompt-only methods, leaving open questions of whether they can accurately generate step-by-step actions in multi-turn interaction tasks.
Approach: They propose to use shopping data to evaluate LLMs' ability to accurately generate step-by-step actions in a multi-turn interaction task.
Outcome: The proposed model achieves 17.26% action generation accuracy and 33.86% F1 score on final purchase prediction, representing improvements of 5.4% and 13.85% over baselines.
Large Language Models Meet NL2Code: A Survey (2023.acl-long)

Copied to clipboard

Challenge: generating code from a natural language description is a pressing and significant challenge in code intelligence.
Approach: They propose to survey 27 existing large language models for NL2Code and compare them to humanEval benchmarks.
Outcome: The proposed model is compared with existing models on the HumanEval benchmark.
BAPO: Boundary-Aware Policy Optimization for Reliable Agentic Search (2026.findings-acl)

Copied to clipboard

Challenge: Existing RL-based agentic search models fail to recognize reasoning boundaries and rarely admit "I DON'T KNOW" lack of reliability leads to plausible but unreliable answers, introducing significant risks .
Approach: They propose a framework to cultivate reliable boundary awareness without compromising accuracy.
Outcome: Experiments show that the proposed framework improves the reliability of agentic search models.
Language-Specific Layer Matters: Efficient Multilingual Enhancement for Large Vision-Language Models (2025.findings-emnlp)

Copied to clipboard

Challenge: Large vision-language models exhibit an imbalance in multilingual capabilities .
Approach: They propose a training recipe that achieves efficient multilingual enhancement for LVLMs by Precise Language Specific layers fine-tuning.
Outcome: The proposed training recipe achieves efficient multilingual enhancement for LVLMs by fine-tuning language specific layers.
Forgetting Curve: A Reliable Method for Evaluating Memorization Capability for Long-Context Models (2024.emnlp-main)

Copied to clipboard

Challenge: Existing methods, tasks and benchmarks to measure model’s effective memory length are limited.
Approach: They propose a method called forgetting curve to measure the memorization capability of long-context models.
Outcome: The proposed method is robust to the tested corpus and experimental settings, and can be applied to any model size.
PartialFormer: Modeling Part Instead of Whole for Machine Translation (2024.findings-acl)

Copied to clipboard

Challenge: Existing feed-forward neural networks have significant computational and parametric overhead.
Approach: They propose a parameter-efficient Transformer architecture that utilizes multiple smaller FFNs to reduce parameters and computation while maintaining essential hidden dimensions.
Outcome: The proposed architecture reduces computational and parameter overhead while maintaining essential hidden dimensions.
QaRL: Rollout-Aligned Quantization-Aware RL for Fast and Stable Training under Training–Inference Mismatch (2026.findings-acl)

Copied to clipboard

Challenge: Recent work has shown that reinforcement learning with simple rule-based reward functions (RLVR) can induce emergent reasoning behaviors and yield gains in challenging domains such as math problem solving.
Approach: They propose a rollout-alignment-quantization-aware RL which aligns training-side forward with the quantized rollout to minimize mismatch.
Outcome: The proposed approach outperforms quantized-rollout training by +5.5 on Qwen3-30B-A3B MoE for math problems while maintaining low-bit throughput.
Shallow-to-Deep Training for Neural Machine Translation (2020.emnlp-main)

Copied to clipboard

Challenge: Experimental results show that deep training is 1:4 faster than training from scratch.
Approach: They propose a shallow-to-deep training method that learns deep models by stacking shallow models.
Outcome: The proposed method is 1:4 faster than training from scratch and achieves BLEU scores of 30:33 and 43:29 on two translation tasks.
Scaling Laws Across Model Architectures: A Comparative Analysis of Dense and MoE Models in Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: a study of large language models (LLMs) reveals the transferability and discrepancies of scaling laws between Dense and MoE models.
Approach: They investigate the transferability and discrepancies of scaling laws between Dense Models and Mixture of Experts models.
Outcome: The results show that the power-law scaling framework also applies to MoE Models .
RouteLMT: Learned Sample Routing for Hybrid LLM Translation Deployment (2026.acl-industry)

Copied to clipboard

Challenge: Existing routing strategies rely on heuristics, external predictors, or absolute quality estimation to capture whether the large model provides a worthwhile improvement over the small one.
Approach: They propose a budget allocation problem for routing large model to large model . they propose heuristics, external predictors, or absolute quality estimation to determine the optimal signal for budgeted decisions.
Outcome: The proposed model outperforms heuristics, quality/difficulty estimation baselines and achieves a superior quality–budget Pareto frontier.
Learning Deep Transformer Models for Machine Translation (P19-1)

Copied to clipboard

Challenge: Neural machine translation models have advanced the previous state-of-the-art by learning mappings between sequences via neural networks and attention mechanisms.
Approach: They propose to use layer normalization to pass the combination of previous layers to the next layer to improve the model.
Outcome: The proposed model outperforms the shallow Transformer-Big/Base baseline model on English-German and Chinese-English tasks by 0.4-2.4 BLEU points.
CONSTRUCTURE: Benchmarking CONcept STRUCTUre REasoning for Multimodal Large Language Models (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing multimodal large language models lack the ability to perceive the visual world with a deep concept structure cognition.
Approach: They propose a concept-level benchmark to assess MLLMs’ hierarchical concept understanding and reasoning abilities.
Outcome: The proposed model outperforms state-of-the-art models in concept structure reasoning evaluation.
Self Promotion in US Congressional Tweets (2021.naacl-main)

Copied to clipboard

Challenge: Prior studies have found that women self-promote less than men due to gender stereotypes.
Approach: They built a BERT-based NLP model to predict whether a Congressional tweet shows self-promotion and then used it to examine whether he gender gap exists among Congressional Tweets.
Outcome: The model predicts whether a Congressional tweet shows self-promotion and then tests it against 2 million tweets from 2017 to 2021.
CodeM: Less Data Yields More Versatility via Ability Matrix (2024.findings-acl)

Copied to clipboard

Challenge: Recent efforts to train code large language models have been booming recently . however, this will incur significant costs in constructing data and training model considering the countless downstream scenarios.
Approach: They propose a data construction strategy which decouples code LLMs’ abilities into two dimensions and constructs a lightweight training corpus that only covers a subset of target scenarios.
Outcome: The proposed model can train a multilingual multitasking model using less data and training data.
Aria-UI: Visual Grounding for GUI Instructions (2025.findings-acl)

Copied to clipboard

Challenge: Using a multimodal model, GUI agents can ground from language instructions to target elements . relying on HTML or AXTree inputs is a challenge for GUI agents .
Approach: They propose a large multimodal model specifically designed for GUI grounding that adopts a pure vision approach instead of auxiliary inputs.
Outcome: The proposed model outperforms vision-only and AXTree-reliant models on offline and online agents.
LANG: Reinforcement Learning for Multilingual Reasoning with Language-Adaptive Hint Guidance (2026.acl-long)

Copied to clipboard

Challenge: Existing methods for enhancing multi-step reasoning have not fully translated to multilingual contexts.
Approach: They propose a framework that leverages language-conditioned hints to guide exploration in non-English reasoning tasks.
Outcome: Empirical results show that the proposed framework improves reasoning performance without compromising language consistency.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations