20260102_165827/papers.csv

Rank,ID,Title,Authors,Year,Venue,Track,Status,Primary Area,Keywords,Citations,BM25 Score,Combined Score,DOI,URL,PDF,Source,TLDR,Abstract
1,,Token-Budget-Aware LLM Reasoning,Tingxu Han; Zhenting Wang; Chunrong Fang; Shiyun Zhao; Shiqing Ma,2024,Annual Meeting of the Association for Computational Linguistics,,,,,117,0.000,0.257,10.48550/arXiv.2412.18547,https://www.semanticscholar.org/paper/3bbf541c2a3ce0ffaed3703cd486f0e1b2febb27,,semantic_scholar,,"Reasoning is critical for large language models (LLMs) to excel in a wide range of tasks. While methods like Chain-of-Thought (CoT) reasoning and enhance LLM performance by decomposing problems into intermediate steps, they also incur significant overhead in token usage, leading to increased costs. "
2,,Token-level Direct Preference Optimization,Yongcheng Zeng; Guoqing Liu; Weiyu Ma; Ning Yang; Haifeng Zhang,2024,International Conference on Machine Learning,,,,,104,0.000,0.245,10.48550/arXiv.2404.11999,https://www.semanticscholar.org/paper/8f2254fb38cfc8f79524fd1cd2609124808f2c8c,,semantic_scholar,,"Fine-tuning pre-trained Large Language Models (LLMs) is essential to align them with human values and intentions. This process often utilizes methods like pairwise comparisons and KL divergence against a reference LLM, focusing on the evaluation of full answers generated by the models. However, the "
3,,How Well do LLMs Compress Their Own Chain-of-Thought? A Token Complexity Approach,Ayeong Lee; Ethan Che; Tianyi Peng,2025,arXiv.org,,,,,61,0.000,0.244,10.48550/arXiv.2503.01141,https://www.semanticscholar.org/paper/d6d561786eb06df93e29bcecfc569a3eeb95af10,,semantic_scholar,,"Chain-of-thought prompting has emerged as a powerful technique for enabling large language models (LLMs) to solve complex reasoning tasks. However, these reasoning chains can be verbose, raising concerns about efficiency. In response, recent works have sought to decrease response lengths through sim"
4,,AIM: Adaptive Inference of Multi-Modal LLMs via Token Merging and Pruning,Yiwu Zhong; Zhuoming Liu; Yin Li; Liwei Wang,2024,arXiv.org,,,,,19,0.000,0.225,10.48550/arXiv.2412.03248,https://www.semanticscholar.org/paper/eba22fafe9a7f7cdda35539ae3dc8debd77aa714,,semantic_scholar,,"Large language models (LLMs) have enabled the creation of multi-modal LLMs that exhibit strong comprehension of visual data such as images and videos. However, these models usually rely on extensive visual tokens from visual encoders, leading to high computational demands, which limits their applica"
5,,Lockpicking LLMs: A Logit-Based Jailbreak Using Token-level Manipulation,Yuxi Li; Yi Liu; Yuekang Li; Ling Shi; Gelei Deng,2024,arXiv.org,,,,,18,0.000,0.223,10.48550/arXiv.2405.13068,https://www.semanticscholar.org/paper/9baeb262c3961f46636c9a657a6996640af6a092,,semantic_scholar,,"Large language models (LLMs) have transformed the field of natural language processing, but they remain susceptible to jailbreaking attacks that exploit their capabilities to generate unintended and potentially harmful content. Existing token-level jailbreaking techniques, while effective, face scal"
6,,Branch-Train-MiX: Mixing Expert LLMs into a Mixture-of-Experts LLM,Sainbayar Sukhbaatar; Olga Golovneva; Vasu Sharma; Hu Xu; Xi Victoria Lin,2024,arXiv.org,,,,,89,0.000,0.216,10.48550/arXiv.2403.07816,https://www.semanticscholar.org/paper/07894aeadab9158fdb97647c4792816ede1b60b9,,semantic_scholar,,"We investigate efficient methods for training Large Language Models (LLMs) to possess capabilities in multiple specialized domains, such as coding, math reasoning and world knowledge. Our method, named Branch-Train-MiX (BTX), starts from a seed model, which is branched to train experts in embarrassi"
7,,Discovering the Gems in Early Layers: Accelerating Long-Context LLMs with 1000x Input Token Reduction,Zhenmei Shi; Yifei Ming; Xuan-Phi Nguyen; Yingyu Liang; Shafiq Joty,2024,arXiv.org,,,,,41,0.000,0.204,10.48550/arXiv.2409.17422,https://www.semanticscholar.org/paper/65a1f6c82840a98d0830b40e348f7243e9afb89a,,semantic_scholar,,"Large Language Models (LLMs) have demonstrated remarkable capabilities in handling long context inputs, but this comes at the cost of increased computational resources and latency. Our research introduces a novel approach for the long context bottleneck to accelerate LLM inference and reduce GPU mem"
8,,Ada-K Routing: Boosting the Efficiency of MoE-based LLMs,Tongtian Yue; Longteng Guo; Jie Cheng; Xuange Gao; Jing Liu,2024,arXiv.org,,,,,8,0.000,0.203,10.48550/arXiv.2410.10456,https://www.semanticscholar.org/paper/b8633fd8a31bd986c35a04f3d41e44db8a43d990,,semantic_scholar,,"In the era of Large Language Models (LLMs), Mixture-of-Experts (MoE) architectures offer a promising approach to managing computational costs while scaling up model parameters. Conventional MoE-based LLMs typically employ static Top-K routing, which activates a fixed and equal number of experts for "
9,,When to Stop? Towards Efficient Code Generation in LLMs with Excess Token Prevention,Lianghong Guo; Yanlin Wang; Ensheng Shi; Wanjun Zhong; Hongyu Zhang,2024,International Symposium on Software Testing and Analysis,,,,,30,0.000,0.200,10.1145/3650212.3680343,https://www.semanticscholar.org/paper/395357c372c298ce517f64aa99fd5419aed9e2bb,https://arxiv.org/pdf/2407.20042,semantic_scholar,,"Code generation aims to automatically generate code snippets that meet given natural language requirements and plays an important role in software development. Although Code LLMs have shown excellent performance in this domain, their long generation time poses a signification limitation in practice "
10,,Enhancing Character-Level Understanding in LLMs through Token Internal Structure Learning,Zhuo Xu; Zhiqiang Zhao; Zihan Zhang; Yuchi Liu; Quanwei Shen,2024,Annual Meeting of the Association for Computational Linguistics,,,,,4,0.000,0.197,10.48550/arXiv.2411.17679,https://www.semanticscholar.org/paper/117b5eccd7163e7bb4d2f1bf05df0c08f6d79f8a,,semantic_scholar,,"Tokenization methods like Byte-Pair Encoding (BPE) enhance computational efficiency in large language models (LLMs) but often obscure internal character structures within tokens. This limitation hinders LLMs' ability to predict precise character positions, which is crucial in tasks like Chinese Spel"