Papers with optimization process
Overcoming Catastrophic Forgetting by Exemplar Selection in Task-oriented Dialogue System (2024.findings-acl)
Copied to clipboard
| Challenge: | Experimental results show that HESIT effectively alleviates catastrophic forgetting by exemplar selection, and achieves state-of-the-art performance on the largest CL benchmark of ToDs in terms of all metrics. |
| Approach: | They propose a method to overcome catastrophic forgetting in task-oriented dialogue systems by tracing their hyper-gradients and a retraining strategy that uses influential exemplars for periodic retrains. |
| Outcome: | The proposed method achieves state-of-the-art on the largest CL benchmark of ToDs in terms of all metrics. |
Reflection in the Dark: Exposing and Escaping the Black Box in Reflective Prompt Optimization (2026.acl-srw)
Copied to clipboard
| Challenge: | Automatic prompt optimization (APO) is a powerful paradigm for improving LLM performance without manual prompt engineering. |
| Approach: | They propose a framework that decouples hypothesis generation from prompt rewriting . they propose VISTA framework that recovers accuracy to 87.57% on same defective seed . |
| Outcome: | The proposed framework outperforms baselines on GSM8K and AIME2025 on a defective seed. |
iPrOp: Interactive Prompt Optimization for Large Language Models with a Human in the Loop (2025.acl-srw)
Copied to clipboard
| Challenge: | Prompt engineering has made significant contributions to the era of large language models, yet its effectiveness depends on the skills of a prompt author. |
| Approach: | They propose a novel approach to prompt optimization that bridges manual prompt engineering and automatic prompt optimization by providing task-specific guidance. |
| Outcome: | The proposed approach bridges manual prompt engineering and automatic prompt optimization while offering users the flexibility to assess evolving prompts. |
Emergence of Syntax Needs Minimal Supervision (2020.acl-main)
Copied to clipboard
| Challenge: | Syntax is the essence of human linguistic capacity that makes it possible to produce and understand unheard sentences. |
| Approach: | They propose to use a corpus to define and isolate grammaticality and meaning/pragmatics information without explicit syntax-specific guidance. |
| Outcome: | The proposed model can search for syntax-based lexical categories without prior hypothesis on the form of the model. |
Gradient Inversion Attack in Federated Learning: Exposing Text Data through Discrete Optimization (2025.coling-main)
Copied to clipboard
| Challenge: | federated learning could overcome the bottleneck of public text data in large language models . a novel attack method is proposed to fully expose text data from gradients . |
| Approach: | They propose a method to fully expose text data from gradients by using a network of clients and a server. |
| Outcome: | The proposed method shows it is possible to Fully Expose Text data from gradients. |
On Surgical Fine-tuning for Language Encoders (2023.findings-emnlp)
Copied to clipboard
Abhilasha Lodha, Gayatri Belapurkar, Saloni Chalkapurkar, Yuanming Tao, Reshmi Ghosh, Samyadeep Basu, Dmitrii Petrov, Soundararajan Srinivasan
| Challenge: | preserving knowledge of target distribution by fine-tuning all layers can be expensive and may increase data volume requirements. |
| Approach: | They propose an efficient metric based on the diagonal of the Fisher information matrix (FIM score) to select the candidate layers for selective fine-tuning. |
| Outcome: | The proposed metric can select layers leading to strong performance on GLUE and SuperGLUE tasks and across distinct language encoders. |
ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning (2025.findings-acl)
Copied to clipboard
| Challenge: | Recent advances have extended DPO to multimodal scenarios, achieving strong performance. |
| Approach: | They propose to use a sentence-level preference optimization technique to optimize individual sentences for more precise preference optimization without additional models or parameters. |
| Outcome: | Experiments show that Adaptive Sentence-level Preference Optimization significantly improves the alignment of multimodal models. |
Exploiting the Index Gradients for Optimization-Based Jailbreaking on Large Language Models (2025.coling-main)
Copied to clipboard
| Challenge: | Despite advances in training Large Language Models, they remain vulnerable to jailbreak, an adversarial attack method. |
| Approach: | They propose an adversarial jailbreak algorithm that exploits the gradient information of the suffix tokens to accelerate the optimization process. |
| Outcome: | The proposed model achieves 1.5x speedup while maintaining high attack success rates. |
Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent (2025.findings-acl)
Copied to clipboard
Xueyang Feng, Jingsen Zhang, Jiakai Tang, Wei Li, Guohao Cai, Xu Chen, Quanyu Dai, Yue Zhu, Zhenhua Dong
| Challenge: | Recent advances in Large Language Models (LLMs) have propelled the development of Conversational Recommendation Agents (CRAs). |
| Approach: | They propose a multi-turn preference optimization paradigm that leverages Expectation Confirmation Theory to explicitly model the evolution of user satisfaction throughout multi-turned dialogues. |
| Outcome: | The proposed paradigm eliminates the significant sampling overhead of existing MTPO methods while ensuring the optimization process drives meaningful improvements. |
Middo: Model-Informed Dynamic Data Optimization for Enhanced LLM Fine-Tuning via Closed-Loop Learning (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to improve data quality face limitations in static dataset curation that fail to adapt to evolving model capabilities. |
| Approach: | They propose a self-evolving framework that uses model-aware data selection and context-preserving data refinement to improve LLM performance. |
| Outcome: | The proposed framework improves the quality of seed data and boosts LLM’s performance with improving accuracy by 7.15% on average while maintaining the original dataset scale. |
When Evolution Strategy Meets Language Models Tuning (2025.coling-main)
Copied to clipboard
| Challenge: | Autoregressive language models with pretraining often display limited capability in effectively following instructions. |
| Approach: | They propose an on-policy approach to optimize models by harnessing the principle of biological evolution, namely survival of the fittest. |
| Outcome: | The proposed method can achieve superior performance in various tasks and comparable performance in the human alignment task. |
WPO: Enhancing RLHF with Weighted Preference Optimization (2024.emnlp-main)
Copied to clipboard
Wenxuan Zhou, Ravi Agrawal, Shujian Zhang, Sathish Reddy Indurthi, Sanqiang Zhao, Kaiqiang Song, Silei Xu, Chenguang Zhu
| Challenge: | Off-policy preference optimization suffers from a distributional gap between the policy used for data collection and the target policy, leading to suboptimal optimization. |
| Approach: | They propose a method to simulate on-policy learning with off-police preference data. |
| Outcome: | The proposed method outperforms Direct Preference Optimization (DPO) by up to 5.6% on Alpaca Eval 2 and MT-bench. |
Logic Unveils Truth, While Disguise Obscures It: Transition Logic Augmented Response Selection for Multi-Turn Dialogue (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods of negative samples tend to yield false negatives due to one-to-many property in open-domain dialogue. |
| Approach: | They propose a sequential variational ladder auto-encoder to capture one-to-many transition pattern of multiple characteristics in open-domain dialogue. |
| Outcome: | The proposed approach improves the performance of a retrieval dialogue system on two benchmarks. |
Character-level White-Box Adversarial Attacks against Transformers via Attachable Subwords Substitution (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to attack transformer models are not effective at character level, but they are a natural attack scenario. |
| Approach: | They propose a character-level adversarial attack method against transformer models . they use a gradient-based method to find the most vulnerable words in a sentence . |
| Outcome: | The proposed method outperforms previous methods on sentence-level and token-level tasks. |
WeightLoRA: Keep Only Necessary Adapters (2026.acl-long)
Copied to clipboard
| Challenge: | Low-rank adaptation (LoRA) adds trainable adapters to selected layers, but requires significant memory to train large models and intuition on which layers to add adapters. |
| Approach: | They propose a method which adds trainable adapters to selected layers . they compare weightLoRA with different adaptive approaches to reduce trainable parameters while maintaining consistent or even superior metric values. |
| Outcome: | The proposed method reduces the number of trainable parameters while maintaining the capability to obtain consistent or even superior metric values. |
Efficient Adversarial Training with Robust Early-Bird Tickets (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to improve the robustness of pre-trained language models are expensive because of the need to generate adversarial examples via gradient descent. |
| Approach: | They propose an adversarial optimization method that searches for robust tickets with structured sparsity in the early stage and fine-tunes tickets in the remaining time. |
| Outcome: | The proposed method achieves up to 7 13 training speedups while maintaining comparable or even better robustness compared to the most competitive state-of-the-art methods. |
Highway Transformer: Self-Gating Enhanced Self-Attentive Networks (2020.acl-main)
Copied to clipboard
| Challenge: | Self-attention mechanisms have made striking state-of-the-art (SOTA) progress in various sequence learning tasks, attending to all the global contexts at different locations. |
| Approach: | They propose a gated component self-dependency units (SDU) that incorporates LSTM-styled gating units to replenish internal semantic importance within the multi-dimensional latent space of individual representations. |
| Outcome: | The proposed system could boost the performance of the Transformer modules by allowing them to skip connections and step towards suboptimal points during the optimization process. |
SGDPO: Self-Guided Direct Preference Optimization for Language Model Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for aligning Large Language Models with human values are limited and results of DPO are not resilient. |
| Approach: | They propose a self-guided direct preference optimization algorithm that incorporates a pilot term to steer the gradient flow during the optimization process. |
| Outcome: | The proposed method can generate human-preferred response up to 9.19% higher than previous methods. |
Model Merging and Safety Alignment: One Bad Model Spoils the Bunch (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for merging large language models often overlook safety alignment during merging, leading to misaligned models. |
| Approach: | They propose to combine safety and domain-specific data to optimize model merging techniques . they propose to use this data to maximize model alignment . |
| Outcome: | The proposed method allows for models that excel in both domain expertise and alignment. |
Debate4MATH: Multi-Agent Debate for Fine-Grained Reasoning in Math (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing data annotation methods suffer from high annotation cost and lack of effective automatic validation. |
| Approach: | They propose a Fine-grained Multi-Agent Debate framework and a dataset that prompts multiple agents to debate and then a Multi-agent Debates Reward Model (MRM) to improve its mathematical reasoning capabilities. |
| Outcome: | The proposed model outperforms the state-of-the-art methods by 1.2% and 3.5% on a GSM8K dataset and 45.1% on the MATH dataset. |
Guiding not Forcing: Enhancing the Transferability of Jailbreaking Attacks on LLMs via Removing Superfluous Constraints (2025.acl-long)
Copied to clipboard
| Challenge: | Jailbreaking attacks can effectively induce unsafe behaviors in Large Language Models (LLMs). |
| Approach: | They propose a conceptual framework to elucidate transferability of gradient-based jailbreaking methods . they identify superfluous constraints as significant barriers to improved transferability . |
| Outcome: | The proposed method increases the overall transfer attack success rate (T-ASR) across target and source models with varying safety levels from 18.4% to 50.3% while improving stability and controllability of jailbreak behaviors. |
Impartial Multi-task Representation Learning via Variance-invariant Probabilistic Decoding (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods focus on balancing loss or gradients but fail to address this issue due to the representation discrepancy in latent space. |
| Approach: | They propose a framework that harmonizes representation spaces across tasks to ensure impartial learning by harmonizing representation spaces. |
| Outcome: | The proposed framework outperforms 12 representative methods under the same multi-task settings, especially in heterogeneous task combinations and data-constrained scenarios. |
Embedding Domain Knowledge for Large Language Models via Reinforcement Learning from Augmented Generation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to embed knowledge into large language models have some limitations . static nature of training data and lack of knowledge in domains create knowledge gaps . |
| Approach: | They propose a method that iteratively cycles between sampling generations and optimizing the model through calculated rewards. |
| Outcome: | The proposed method outperforms baseline approaches on medical, legal, astronomy, and current events datasets. |
RiOT: Efficient Prompt Refinement with Residual Optimization Tree (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for automatic prompt optimization face two challenges: lack of diversity and semantic drift. |
| Approach: | They propose a framework for automatic prompt optimization that iteratively refines prompts through text gradients and selects the best prompt using perplexity. |
| Outcome: | The proposed framework outperforms existing prompt optimization methods and manual prompting on commonsense, mathematical, logical, temporal, and semantic reasoning benchmarks. |
C2PO: Diagnosing and Disentangling Bias Shortcuts in LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to solve large language models address stereotypical and structural biases in isolation . however, prior paradigms address these in isolation, often at the expense of exacerbating the other . |
| Approach: | They propose a framework to tackle latent spurious feature correlations within input that drive erroneous reasoning shortcuts. |
| Outcome: | The proposed framework mitigates stereotypical and structural biases while preserving robust general reasoning capabilities. |
Reverse Prompt Engineering: A Zero-Shot, Genetic Algorithm Approach to Language Model Inversion (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods rely on a large number of outputs for training and inference, and they can produce garbled text. |
| Approach: | They propose a training-free framework that reconstructs prompts using only a limited number of text outputs from a language model. |
| Outcome: | The proposed framework generates high-quality prompt recovery and more semantically and functionally aligned with the originals than current state-of-the-art methods. |
SEE: Strategic Exploration and Exploitation for Cohesive In-Context Prompt Optimization (2025.acl-long)
Copied to clipboard
Wendi Cui, Jiaxin Zhang, Zhuohang Li, Hao Sun, Damien Lopez, Kamalika Das, Bradley A. Malin, Sricharan Kumar
| Challenge: | Existing approaches separate the optimization of prompt instructions and in-context learning examples, leading to incohesive, suboptimal results. |
| Approach: | They propose a framework that refines both prompt instructions and in-context learning examples. |
| Outcome: | The proposed framework outperforms state-of-the-art prompt optimization methods on 35 benchmark tasks. |