Papers by Yilong Wu

20 papers
Uncertainty-Aware Routing for Principled Alignment with MoE Dynamics (2026.acl-long)

Copied to clipboard

Challenge: Mixture-of-Experts (MoE) is a cornerstone for scaling LLMs, yet its training dynamics remain poorly understood, often leading to sub-optimal specialization.
Approach: They propose to use Helmholtz Free Energy and Router Entropy to study the MoE lifecycle and identify a universal Three-Stage Phase Transition .
Outcome: The proposed model reduces perplexity and improves expert distinctiveness, offering a principled path toward thermodynamically aligned computation.
TransferTOD: A Generalizable Chinese Multi-Domain Task-Oriented Dialogue System with Transfer Capabilities (2024.emnlp-main)

Copied to clipboard

Challenge: Current datasets cater to user-led systems and are limited to predefined specific scenarios and slots.
Approach: They propose to use a Chinese dialogue dataset to train a model that authentically simulates human-computer dialogues in 30 popular life service scenarios.
Outcome: The proposed model achieves a joint accuracy of 75.09% in out-of-domain evaluations . it also achieves notable abilities in slot filling and questioning .
PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts (2025.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) have shown remarkable progress in dialogue and reasoning, but they struggle to solve strictly constrained dialogue tasks.
Approach: They construct a dataset that contains 12,705 high-quality Chinese dialogue instructions from 440 flowcharts containing 5,055 process nodes.
Outcome: The proposed model outperforms GPT-4o models on backward transitions and outperformed GPT-42 models on the same dataset.
LEMON: Reviving Stronger and Smaller LMs from Larger LMs with Linear Parameter Fusion (2024.acl-long)

Copied to clipboard

Challenge: Existing methods to train a stronger and smaller model with the help of large models are limited by the model size and performance.
Approach: They propose to learn competent initial points for smaller models by fusing parameters from larger models and introduce controllable receptive fields to model prior parameter characteristics.
Outcome: The proposed method outperforms baselines in terms of effectiveness and efficiency.
MulDimIF: A Multi-Dimensional Constraint Framework for Evaluating and Improving Instruction Following in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing research has focused on constraint categories, offering little guidance for improving instruction following abilities.
Approach: They propose a multi-dimensional constraint framework that allows for instruction following . they construct 9,106 code-verifiable samples and evaluate 18 LLMs .
Outcome: The proposed framework improves instruction following performance without compromising general performance.
LLMEval-Fair: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models (2026.acl-long)

Copied to clipboard

Challenge: Existing evaluation of Large Language Models on static benchmarks is vulnerable to data contamination and leaderboard overfitting.
Approach: LLMEval-Fair framework provides a framework for dynamic evaluation of Large Language Models . evaluators use a proprietary bank of 220k graduate-level questions to analyze model data .
Outcome: LLMEval-Fair provides robust and credible evaluation framework for Large Language Models . it provides a strong empirical validation for the dynamic evaluation paradigm .
Inner Thinking Transformer: Leveraging Dynamic Depth Scaling to Foster Adaptive Internal Thinking (2025.acl-long)

Copied to clipboard

Challenge: Large language models face inherent performance bottlenecks under parameter constraints . challenging tokens induce abrupt gradient spikes across layers, exposing stress points .
Approach: They propose an inner thinking transformer that reimagines layer computations as implicit thinking steps.
Outcome: Empirical results show that ITT outperforms Transformer/Loop variants in 11 benchmarks.
NACL: A General and Effective KV Cache Eviction Framework for LLM at Inference Time (2024.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) with extended context windows are expensive and infeasible on fixed memory hardware due to the surprisingly large memory consumption of KV Cache.
Approach: They propose a general framework for long-context KV cache eviction that achieves more optimal and efficient evict in a single operation during the encoding phase.
Outcome: The proposed framework improves performance on short- and long-text tasks by 80% and 76% respectively, reducing KV Cache by up to 5 with over 95% performance maintenance.
AdaRewriter: Unleashing the Power of Prompting-based Conversational Query Reformulation via Test-Time Adaptation (2025.emnlp-main)

Copied to clipboard

Challenge: Prompting-based conversational query reformulation has emerged as a powerful approach for conversational search, refining ambiguous user queries into standalone search queries.
Approach: They propose a framework for query reformulation using an outcome-supervised reward model via test-time adaptation.
Outcome: Experiments on five conversational search datasets show that AdaRewriter significantly outperforms the existing methods across most settings.
A Simple Hash-Based Early Exiting Approach For Language Understanding and Generation (2022.findings-acl)

Copied to clipboard

Challenge: Existing methods to measure instance difficulty use generalization and threshold-tuning . a new approach to learn to exit is based on hash functions to assign tokens to a fixed exiting layer.
Approach: They propose a Hash-based Early Exiting approach that replaces learn-to-exit modules with hash functions to assign each token to a fixed exiting layer.
Outcome: The proposed approach improves on learning to exit and predicting instance difficulty.
SCOPE: Optimizing Key-Value Cache Compression in Long-context Generation (2025.acl-long)

Copied to clipboard

Challenge: Excessive compression during the prefill phase impairs comprehension of reasoning tasks . SCOPE is a framework that performs KV cache optimization during the decoding and prefill phases .
Approach: They propose a framework that performs optimization during the prefill and decoding phases . they propose enabling a sliding strategy to select essential heavy hitters for the decoding phase .
Outcome: Experiments show that SCOPE can optimize key-value cache for long-context generation tasks . the framework can preserve essential information while minimizing memory usage and transfer .
ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages (2024.acl-long)

Copied to clipboard

Challenge: Existing research focuses on enhancing LLMs capabilities through tool utilization.
Approach: They propose a framework to investigate safety issues in large language models in tool learning . they propose malicious queries and jailbreak attacks in the input stage .
Outcome: The proposed framework investigates six safety scenarios for LLMs in tool learning . the data will be released upon acceptance of the proposed framework .
Large Language Model Agents in Finance: A Survey Bridging Research, Practice, and Real-World Deployment (2025.findings-emnlp)

Copied to clipboard

Challenge: a systematic review of large language models (LLMs) is conducted to better align their capabilities with real-world demands.
Approach: They propose a functional taxonomy mapping financial domains to tasks, datasets, and institutional constraints. they catalog over 30 financial benchmarks and 20 representative models.
Outcome: The proposed model frameworks are bridging financial practice and LLM research.
TL-Training: A Task-Feature-Based Framework for Training Large Language Models in Tool Use (2025.findings-emnlp)

Copied to clipboard

Challenge: a new approach to training large language models (LLMs) overlooks task-specific characteristics in tool use, leading to performance bottlenecks.
Approach: They propose a task-feature-based framework that mitigates the effects of suboptimal training data . they use a dataset to train large-scale LLMs and a reward mechanism tailored to error categories .
Outcome: The proposed framework matches or surpasses open- and closed-source LLMs in tool-use performance using only 1,217 training data points.
ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios (2025.coling-main)

Copied to clipboard

Challenge: Existing evaluations of tool learning focus on validation of tools for large language models with expected outcomes, but this focus ignores the complex capabilities required for LLMs to effectively use tools.
Approach: They propose a fine-grained system for evaluation of large language models’ tool learning capabilities in authentic scenarios.
Outcome: The proposed system examines seven real-world scenarios, analyzing five dimensions crucial to LLMs in tool learning: format alignment, intent comprehension, behavior planning, tool selection, and answer organization.
RoTBench: A Multi-Level Benchmark for Evaluating the Robustness of Large Language Models in Tool Learning (2024.emnlp-main)

Copied to clipboard

Challenge: Current research emphasizes LLMs’ capacity to utilize tools in well-structured environments while overlooking their stability when confronted with the inevitable noise of the real world.
Approach: They propose a multi-level benchmark to evaluate the robustness of large language models in tool learning by establishing five external environments with varying levels of noise.
Outcome: The proposed model outperforms the GPT-4 model in tool learning in three critical phases: tool selection, parameter identification, and content filling.
Sparse Growing Transformer: Training-Time Sparse Depth Allocation via Progressive Attention Looping (2026.findings-acl)

Copied to clipboard

Challenge: Existing approaches to increasing effective depth of LLMs rely on parameter reuse, extending computation through recursive execution.
Approach: They propose a training-time sparse depth allocation framework that progressively increases depth for a small subset of parameters as training evolves.
Outcome: The proposed model outperforms existing approaches to increasing the effective depth of language models while reducing training FLOPs overhead from approximately 16–20% to only 1–3% relative to a standard Transformer backbone.
SEED: Accelerating Reasoning Tree Construction via Scheduled Speculative Decoding (2025.coling-main)

Copied to clipboard

Challenge: Large Language Models (LLMs) have remarkable emergent abilities across various tasks, yet their performance on complex reasoning and planning tasks remains suboptimal.
Approach: They propose a tree-search-based reasoning framework that encourages the exploration of intermediate steps and a round-scheduled strategy to manage draft model dispatching.
Outcome: The proposed framework improves runtime speed and GPU memory management concurrently and handles multiple iterations for thought generation and state evaluation.
GoViG: Goal-Conditioned Visual Navigation Instruction Generation via Multimodal Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Current methods for instruction generation depend on privileged inputs such as semantic maps, landmark annotations, and panoramic views.
Approach: They propose a task that generates coherent navigation instructions from egocentric visual observations.
Outcome: The proposed task generates coherent navigation instructions from egocentric visual data . the proposed task improves performance over state-of-the-art methods in BLEU-4 and CIDEr scores .
AdaCQR: Enhancing Query Reformulation for Conversational Search via Sparse and Dense Retrieval Alignment (2025.coling-main)

Copied to clipboard

Challenge: Existing methods to address conversational search challenges are limited by one specific retrieval system.
Approach: They propose a framework to enhance generalizability of information-seeking queries by aligning reformulation models with term-based and semantic retrieval systems.
Outcome: The proposed framework outperforms existing methods in a more efficient framework.

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