COLM
Collection
Accepted papers for COLM (Conference on Language Modeling), one dataset per year. • 2 items • Updated
paper_id stringlengths 10 10 | title stringlengths 31 135 | paper_url stringlengths 42 42 | authors listlengths 1 31 | type stringclasses 0
values | primary_area stringclasses 0
values | abstract large_stringlengths 649 2.54k | keywords listlengths 1 15 | TL;DR large_stringlengths 21 250 | submission_number int64 13 1.5k | arxiv_id stringlengths 10 10 ⌀ | arxiv_id_source stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|
F7aAhfitX6 | Massive Activations in Large Language Models | https://openreview.net/forum?id=F7aAhfitX6 | [
"Mingjie Sun",
"Xinlei Chen",
"J Zico Kolter",
"Zhuang Liu"
] | null | null | We observe an empirical phenomenon in Large Language Models (LLMs) -- very few activations exhibit significantly larger values than others (e.g., 100,000 times larger). We call them massive activations. First, we demonstrate the widespread existence of massive activations across various LLMs and characterize their loca... | [
"massive activations",
"biases",
"self-attention"
] | Massive activations exist in LLMs, and they are fundamentally connected to self-attention. | 13 | 2402.17762 | title_snapshot |
dnwRScljXr | Evaluating LLMs at Detecting Errors in LLM Responses | https://openreview.net/forum?id=dnwRScljXr | [
"Ryo Kamoi",
"Sarkar Snigdha Sarathi Das",
"Renze Lou",
"Jihyun Janice Ahn",
"Yilun Zhao",
"Xiaoxin Lu",
"Nan Zhang",
"Yusen Zhang",
"Haoran Ranran Zhang",
"Sujeeth Reddy Vummanthala",
"Salika Dave",
"Shaobo Qin",
"Arman Cohan",
"Wenpeng Yin",
"Rui Zhang"
] | null | null | With Large Language Models (LLMs) being widely used across various tasks, detecting errors in their responses is increasingly crucial. However, little research has been conducted on error detection of LLM responses. Collecting error annotations on LLM responses is challenging due to the subjective nature of many NLP ta... | [
"error detection",
"evaluation"
] | We introduce the first error detection benchmark consisting of objective, realistic, and diverse errors made by GPT-4 and Llama 2 70B, and evaluate 11 LLMs on the error detection task. | 16 | 2404.03602 | title_snapshot |
3GhOWfSLrD | Will the Real Linda Please Stand up...to Large Language Models? Examining the Representativeness Heuristic in LLMs | https://openreview.net/forum?id=3GhOWfSLrD | [
"Pengda Wang",
"Zilin Xiao",
"Hanjie Chen",
"Frederick L. Oswald"
] | null | null | Although large language models (LLMs) have demonstrated remarkable proficiency in modeling text and generating human-like text, they may exhibit biases acquired from training data in doing so. Specifically, LLMs may be susceptible to a common cognitive trap in human decision-making called the representativeness heurist... | [
"Psychology",
"Representativeness Heuristic",
"Language Models"
] | Dissecting Representativeness Heuristic in Language Models | 27 | 2404.01461 | title_snapshot |
K1M3gLW0MX | On Fairness of Low-Rank Adaptation of Large Models | https://openreview.net/forum?id=K1M3gLW0MX | [
"Zhoujie Ding",
"Ken Liu",
"Pura Peetathawatchai",
"Berivan Isik",
"Sanmi Koyejo"
] | null | null | Low-rank adaptation of large models, particularly LoRA, has gained traction due to its computational efficiency. This efficiency, contrasted with the prohibitive costs of full-model fine-tuning, means that practitioners often turn to LoRA without a complete understanding of its ramifications. In this study, we focus on... | [
"Low-rank adaptation",
"LoRA",
"bias",
"fairness",
"subgroup fairness",
"evaluations",
"LLMs",
"large models"
] | Compared to full-model fine-tuning, does low-rank adaptation (LoRA) have any side effects on model fairness in terms of accuracy, calibration, privacy, and gender bias? How do we know? | 32 | 2405.17512 | title_snapshot |
KZd1EErRJ1 | IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations | https://openreview.net/forum?id=KZd1EErRJ1 | [
"Deqing Fu",
"Ruohao Guo",
"Ghazal Khalighinejad",
"Ollie Liu",
"Bhuwan Dhingra",
"Dani Yogatama",
"Robin Jia",
"Willie Neiswanger"
] | null | null | Current foundation models exhibit impressive capabilities when prompted either with text only or with both image and text inputs. But do their capabilities change depending on the input modality? In this work, we propose **IsoBench**, a benchmark dataset containing problems from four major areas: math, science, algorit... | [
"large language models",
"vision language models",
"evaluation",
"isomorphism"
] | We propose IsoBench to measure multimodal foundation models' performance on various problems with isomorphic representations | 35 | 2404.01266 | title_snapshot |
HDkNbfLQgu | Reverse Training to Nurse the Reversal Curse | https://openreview.net/forum?id=HDkNbfLQgu | [
"Olga Golovneva",
"Zeyuan Allen-Zhu",
"Jason E Weston",
"Sainbayar Sukhbaatar"
] | null | null | Large language models (LLMs) have a surprising failure: when trained on ``A has a feature B``, they do not generalize to ``B is a feature of A``, which is termed the Reversal Curse. Even when training with trillions of tokens this issue still appears due to Zipf's law -- hence even if we train on the entire internet. T... | [
"LLMs",
"Large Language Models",
"Question Answering",
"Generalization",
"Knowledge Representation",
"Logical Inference",
"Relations"
] | When trained on “A has a feature B”, LLMs do not generalize to “B is a feature of A”, which is termed the Reversal Curse. This work proposes an alternative training scheme, called reverse training, that resolves the Reversal Curse. | 42 | 2403.13799 | title_snapshot |
EEPBOB2Xww | Ferret-v2: An Improved Baseline for Referring and Grounding with Large Language Models | https://openreview.net/forum?id=EEPBOB2Xww | [
"Haotian Zhang",
"Haoxuan You",
"Philipp Dufter",
"Bowen Zhang",
"Chen Chen",
"Hong-You Chen",
"Tsu-Jui Fu",
"William Yang Wang",
"Shih-Fu Chang",
"Zhe Gan",
"Yinfei Yang"
] | null | null | While Ferret seamlessly integrates regional understanding into the Large Language Model (LLM) to facilitate its referring and grounding capability, it poses certain limitations: constrained by the pre-trained fixed visual encoder and failed to perform well on broader tasks. In this work, we unveil Ferret-v2, a signific... | [
"Multimodal Large Language Model",
"Referring",
"Grounding"
] | Ferret-v2, a significant upgrade to Ferret, which excels in detailed vision-related tasks without compromising their proficiency in global understanding. | 45 | 2404.07973 | title_snapshot |
2nTzomzjjb | ProLLM: Protein Chain-of-Thoughts Enhanced LLM for Protein-Protein Interaction Prediction | https://openreview.net/forum?id=2nTzomzjjb | [
"Mingyu Jin",
"Haochen Xue",
"Zhenting Wang",
"Boming Kang",
"Ruosong Ye",
"Kaixiong Zhou",
"Mengnan Du",
"Yongfeng Zhang"
] | null | null | The prediction of protein-protein interactions (PPIs) is crucial for understanding biological functions and diseases. Previous machine learning approaches to PPI prediction mainly focus on direct physical interactions, ignoring the broader context of nonphysical connections through intermediate proteins, thus limiting ... | [
"Protein-Protein-Interaction(PPI)",
"Large Language Model(LLM)",
"Chain of Thought"
] | Our study introduces ProLLM, a novel Large Language Model fine-tuned with ProCoT prompts reflecting singal passways, showcasing a significant leap in predicting protein-protein interactions with enhanced accuracy and generalizability. | 51 | 2405.06649 | title_snapshot |
u2vAyMeLMm | Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens | https://openreview.net/forum?id=u2vAyMeLMm | [
"Jiacheng Liu",
"Sewon Min",
"Luke Zettlemoyer",
"Yejin Choi",
"Hannaneh Hajishirzi"
] | null | null | Are $n$-gram language models still relevant in this era of neural large language models (LLMs)? Our answer is *yes*, and we showcase their values in both text analysis and improving neural LLMs. This was done by modernizing $n$-gram LMs in two aspects. First, we train them at the same data scale as neural LLMs -- **5 t... | [
"infini-gram",
"n-gram",
"language model",
"suffix array"
] | We built the largest ever n-gram LM on trillions of tokens and with unbounded n, developed a method to efficiently train and serve it, and showed its great utility in this era of neural LLMs. | 59 | 2401.17377 | title_snapshot |
kh9Zt2Ldmn | Don't throw away your value model! Generating more preferable text with Value-Guided Monte-Carlo Tree Search decoding | https://openreview.net/forum?id=kh9Zt2Ldmn | [
"Jiacheng Liu",
"Andrew Cohen",
"Ramakanth Pasunuru",
"Yejin Choi",
"Hannaneh Hajishirzi",
"Asli Celikyilmaz"
] | null | null | Inference-time search algorithms such as Monte-Carlo Tree Search (MCTS) may seem unnecessary when generating natural language text based on state-of-the-art reinforcement learning such as Proximal Policy Optimization (PPO). In this paper, we demonstrate that it is possible to get extra mileage out of PPO by integrating... | [
"text generation",
"decoding",
"search",
"reinforcement learning",
"ppo",
"monte-carlo tree search"
] | We adapt and apply AlphaGo-style MCTS decoding to post-RLHF language models, and achieve superior generation quality. | 60 | 2309.15028 | title_snapshot |
soGxskHGox | Linearizing Large Language Models | https://openreview.net/forum?id=soGxskHGox | [
"Jean Mercat",
"Igor Vasiljevic",
"Sedrick Scott Keh",
"Kushal Arora",
"Achal Dave",
"Adrien Gaidon",
"Thomas Kollar"
] | null | null | Linear transformers have emerged as a subquadratic-time alternative to softmax
attention and have garnered significant interest due to their fixed recurrent state.
However, they suffer from poor scaling and under-perform compute-matched
transformers. Prior models such as RWKV and Mamba have attempted to address
these s... | [
"linear attention",
"efficient attention",
"RNN"
] | converting LLMs into RNNs through minimal up-training | 68 | 2405.06640 | title_snapshot |
Pvn1dKreZW | "Merge Conflicts!'" Exploring the Impacts of External Knowledge Distractors to Parametric Knowledge Graphs | https://openreview.net/forum?id=Pvn1dKreZW | [
"Cheng Qian",
"Xinran Zhao",
"Tongshuang Wu"
] | null | null | Large language models (LLMs) acquire extensive knowledge during pre-training, known as their parametric knowledge. However, to remain up-to-date and align with human instructions, LLMs inevitably require external knowledge during interactions. This raises a crucial question: How will LLMs respond when external knowledg... | [
"Large Language Model",
"Knowledge Conflict"
] | We build parametric knowledge graph to reveal LLM's internal knowledge, and systematically introduce external knowledge to discover the impact of internal and external knowledge conflicts. | 69 | 2309.08594 | title_judge |
DOMP5AgwQz | CTIKG: LLM-Powered Knowledge Graph Construction from Cyber Threat Intelligence | https://openreview.net/forum?id=DOMP5AgwQz | [
"Liangyi Huang",
"Xusheng Xiao"
] | null | null | To gain visibility into evolving threat landscape, knowledge of cyber threats has been aggressively collected across organizations and is often shared through Cyber Threat Intelligence (CTI). While knowledge of CTI can be shared via structured format such as Indicators of Compromise (IOC), articles in technical blogs a... | [
"Large Language Model",
"Machine Learning and Security",
"Knowledge Graph",
"Information Extraction"
] | Collaboration of multiple LLM Agents for knowledge extraction from articles in the computer security field | 74 | null | null |
PKfAq8N4fK | AgentKit: Structured LLM Reasoning with Dynamic Graphs | https://openreview.net/forum?id=PKfAq8N4fK | [
"Yue Wu",
"Yewen Fan",
"So Yeon Min",
"Shrimai Prabhumoye",
"Stephen Marcus McAleer",
"Ruslan Salakhutdinov",
"Yonatan Bisk",
"Yuanzhi Li",
"Tom Mitchell"
] | null | null | We propose an intuitive LLM prompting framework (AgentKit) for multifunctional agents.
AgentKit offers a unified framework for explicitly constructing a complex "thought process" from simple natural language prompts. The basic building block in AgentKit is a **node**, containing a natural language prompt for a specifi... | [
"LLM Agents"
] | An intuitive LLM prompting framework for multifunctional agents, by explicitly constructing a complex "thought process" from simple natural language prompts. | 78 | 2404.11483 | title_snapshot |
dribhnhm1i | Tuning Language Models by Proxy | https://openreview.net/forum?id=dribhnhm1i | [
"Alisa Liu",
"Xiaochuang Han",
"Yizhong Wang",
"Yulia Tsvetkov",
"Yejin Choi",
"Noah A. Smith"
] | null | null | Despite the general capabilities of large pretrained language models, they consistently benefit from further adaptation to better achieve desired behaviors. However, tuning these models has become increasingly resource-intensive, or impossible when model weights are private. We introduce **proxy-tuning**, a lightweight... | [
"LM adaptation",
"inference algorithms",
"instruction-tuning"
] | Tune black-box LMs by operating only on its output logits (not its weights), by shifting them in the direction of tuning (as represented by smaller, tunable proxies). Experiments on instruction-tuning, domain adaptation, and task finetuning. | 80 | 2401.08565 | title_snapshot |
QdWhj0QZFw | LLM360: Towards Fully Transparent Open-Source LLMs | https://openreview.net/forum?id=QdWhj0QZFw | [
"Zhengzhong Liu",
"Aurick Qiao",
"Willie Neiswanger",
"Hongyi Wang",
"Bowen Tan",
"Tianhua Tao",
"Junbo Li",
"Yuqi Wang",
"Suqi Sun",
"Omkar Pangarkar",
"Richard Fan",
"Yi Gu",
"Victor Miller",
"Yonghao Zhuang",
"Guowei He",
"Haonan Li",
"Fajri Koto",
"Liping Tang",
"Nikhil Ranja... | null | null | The recent surge in open-source Large Language Models (LLMs), such as LLaMA, Falcon, and Mistral, provides diverse options for AI practitioners and researchers. However, most LLMs have only released partial artifacts, such as the final model weights or inference code, and technical reports increasingly limit their scop... | [
"open-source LLMs",
"LLM360",
"open data",
"fully open LLMs",
"open source"
] | We introduce LLM360, an initiative to open source LLMs that fosters transparency, trust, and collaborative research. | 85 | 2312.06550 | title_snapshot |
C0j44uRPcl | On Robustness-Accuracy Characterization of Language Models using Synthetic Datasets | https://openreview.net/forum?id=C0j44uRPcl | [
"Ching-Yun Ko",
"Pin-Yu Chen",
"Payel Das",
"Yung-Sung Chuang",
"Luca Daniel"
] | null | null | In recent years, language models (LMs) that were pretrained at scale on diverse data have proven to be a successful approach for solving different downstream tasks. However, new concerns about proper performance evaluation have been raised, especially for test-data leakage caused by accidentally including them during p... | [
"synthetic data",
"evaluation"
] | A real-data-free evaluation framework for language models | 89 | null | null |
t4eB3zYWBK | MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries | https://openreview.net/forum?id=t4eB3zYWBK | [
"Yixuan Tang",
"Yi Yang"
] | null | null | Retrieval-augmented generation (RAG) augments large language models (LLM) by retrieving relevant knowledge, showing promising potential in mitigating LLM hallucinations and enhancing response quality, thereby facilitating the great adoption of LLMs in practice. However, we find that existing RAG systems are inadequate ... | [
"Retrieval-Augmented Generation",
"Benchmark",
"Multi-Hop Reasoning"
] | Evaluate retrieval and reasoning across documents in the RAG pipelines. | 95 | 2401.15391 | title_snapshot |
W8Rv1jVycX | Description-Based Text Similarity | https://openreview.net/forum?id=W8Rv1jVycX | [
"Shauli Ravfogel",
"Valentina Pyatkin",
"Amir David Nissan Cohen",
"Avshalom Manevich",
"Yoav Goldberg"
] | null | null | Identifying texts with a given semantics is central for many information seeking scenarios. Similarity search over vector embeddings appear to be central to this ability, yet the similarity reflected in current text embeddings is corpus-driven, and is inconsistent and sub-optimal for many use cases. What, then, is a go... | [
"semantics",
"similarity",
"description based similarity",
"retrieval"
] | We define the notion of description-based similarity, use LLMs to create a tailored dataset for this kind of similarity, and train a corresponding retrieval model. | 96 | 2305.12517 | title_snapshot |
UPE6WYE8vg | A Language Agent for Autonomous Driving | https://openreview.net/forum?id=UPE6WYE8vg | [
"Jiageng Mao",
"Junjie Ye",
"Yuxi Qian",
"Marco Pavone",
"Yue Wang"
] | null | null | Human-level driving is an ultimate goal of autonomous driving. Conventional approaches formulate autonomous driving as a perception-prediction-planning framework, yet their systems do not capitalize on the inherent reasoning ability and experiential knowledge of humans. In this paper, we propose a fundamental paradigm ... | [
"Language Agent",
"Autonomous Driving"
] | Agent-Driver transforms the conventional perception-prediction-planning framework by introducing LLMs as an agent for autonomous driving. | 98 | 2311.10813 | title_snapshot |
IBCBMeAhmC | Evaluating Language Models for Efficient Code Generation | https://openreview.net/forum?id=IBCBMeAhmC | [
"Jiawei Liu",
"Songrun Xie",
"Junhao Wang",
"Yuxiang Wei",
"Yifeng Ding",
"LINGMING ZHANG"
] | null | null | We introduce Differential Performance Evaluation (DPE), a framework designed to reliably evaluate Large Language Models (LLMs) for efficient code generation. Traditional coding benchmarks often fail to provide reliable insights into code efficiency, due to their reliance on simplistic test inputs and the absence of eff... | [
"Code Generation",
"Evaluation",
"Code Efficiency"
] | We propose a new methodology and dataset to effectively evaluate the efficiency of LLM produced code. | 99 | 2408.06450 | title_snapshot |
ZZzXpyv65G | Language Models as Critical Thinking Tools: A Case Study of Philosophers | https://openreview.net/forum?id=ZZzXpyv65G | [
"Andre Ye",
"Jared Moore",
"Rose Novick",
"Amy X Zhang"
] | null | null | Current work in language models (LMs) helps us speed up or even skip thinking by accelerating and automating cognitive work.
But can LMs help us with critical thinking -- thinking in deeper, more reflective ways which challenge assumptions, clarify ideas, and engineer new concepts?
We treat philosophy as a case study i... | [
"philosophy",
"critical thinking",
"human-computer interaction"
] | We interview philosophers to discover how LMs can be critical thinking tools. | 102 | 2404.04516 | title_snapshot |
6vEfyp0o68 | MANGO: A Benchmark for Evaluating Mapping and Navigation Abilities of Large Language Models | https://openreview.net/forum?id=6vEfyp0o68 | [
"Peng Ding",
"Jiading Fang",
"Peng Li",
"Kangrui Wang",
"Xiaochen Zhou",
"Mo Yu",
"Jing Li",
"Hongyuan Mei",
"Matthew Walter"
] | null | null | Large language models such as ChatGPT and GPT-4 have recently achieved astonishing performance on a variety of natural language processing tasks. In this paper, we propose MANGO, a benchmark to evaluate their capabilities to perform text-based mapping and navigation. Our benchmark includes 53 mazes taken from a suite o... | [
"large language models",
"robotics",
"mapping",
"navigation",
"textgame"
] | We propose MANGO, a benchmark to evaluate the capabilities of large language models to perform text-based mapping and navigation. | 115 | 2403.19913 | title_snapshot |
MI52iXSSNy | Commonsense-T2I Challenge: Can Text-to-Image Generation Models Understand Commonsense? | https://openreview.net/forum?id=MI52iXSSNy | [
"Xingyu Fu",
"Muyu He",
"Yujie Lu",
"William Yang Wang",
"Dan Roth"
] | null | null | We present a novel task and benchmark for evaluating the ability of text-to-image(T2I) generation models to produce images that align with commonsense in real life, which we call Commonsense-T2I. Given two adversarial text prompts containing an identical set of action words with minor differences, such as *a lightbulb ... | [
"text to image generation",
"alignment with reality",
"commonsense reasoning"
] | This paper finds that text-to-image generation models do not align with the reality, in real world commonsense. | 120 | 2406.07546 | title_snapshot |
t3z6UlV09o | How bad is training on synthetic data? A statistical analysis of language model collapse | https://openreview.net/forum?id=t3z6UlV09o | [
"Mohamed El Amine Seddik",
"Suei-Wen Chen",
"Soufiane Hayou",
"Pierre Youssef",
"Merouane Abdelkader DEBBAH"
] | null | null | Model collapse, as introduced in (Shumailov et al., 2023), refers to the phenomenon where training models on synthetic data generated from previously trained models leads to a deterioration in performance. This recursive training loop makes the tails of the original distribution disappear, thereby making future-generat... | [
"model collapse",
"recursive training",
"generative models"
] | Study of model collapse through an introduced toy language model. | 125 | 2404.05090 | title_snapshot |
nT6fQIidrQ | Learning to Plan for Language Modeling from Unlabeled Data | https://openreview.net/forum?id=nT6fQIidrQ | [
"Nathan Cornille",
"Marie-Francine Moens",
"Florian Mai"
] | null | null | By training to predict the next token in an unlabeled corpus, large language models learn to perform many tasks without any labeled data. However, their next-token-prediction objective arguably limits their performance in scenarios that require planning, such as writing a coherent article. In this paper, we train a mod... | [
"planning",
"representation learning",
"hierarchical models"
] | We propose to condition LMs on writing actions predicted via an external planner trained in a self-supervised way. | 128 | 2404.00614 | title_snapshot |
oRXPiSOGH9 | Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking | https://openreview.net/forum?id=oRXPiSOGH9 | [
"Eric Zelikman",
"Georges Raif Harik",
"Yijia Shao",
"Varuna Jayasiri",
"Nick Haber",
"Noah Goodman"
] | null | null | When writing and talking, people sometimes pause to think. Although reasoning-focused works have often framed reasoning as a method of answering questions or completing agentic tasks, reasoning is implicit in almost all written text. For example, this applies to the steps not stated between the lines of a proof or to t... | [
"reasoning",
"internal monologue"
] | Language models can teach themselves to reason using internal monologue. | 129 | 2403.09629 | title_snapshot |
egVSgtJJAx | VisualWebBench: How Far Have Multimodal LLMs Evolved in Web Page Understanding and Grounding? | https://openreview.net/forum?id=egVSgtJJAx | [
"Junpeng Liu",
"Yifan Song",
"Bill Yuchen Lin",
"Wai Lam",
"Graham Neubig",
"Yuanzhi Li",
"Xiang Yue"
] | null | null | Multimodal Large Language models (MLLMs) have shown promise in web-related tasks, but evaluating their performance in the web domain remains a challenge due to the lack of comprehensive benchmarks. Existing benchmarks are either designed for general multimodal tasks, failing to capture the unique characteristics of web... | [
"multimodal large language model",
"evaluation",
"web understanding",
"grounding"
] | VisualWebBench: How Far Have LMMs Evolved in Web Page Understanding and Grounding? | 145 | 2404.05955 | title_snapshot |
pYEnhZ6NAv | How Far Are We from Intelligent Visual Deductive Reasoning? | https://openreview.net/forum?id=pYEnhZ6NAv | [
"Yizhe Zhang",
"Richard He Bai",
"Ruixiang ZHANG",
"Jiatao Gu",
"Shuangfei Zhai",
"Joshua M. Susskind",
"Navdeep Jaitly"
] | null | null | Vision-Language Models (VLMs) have recently demonstrated incredible strides on diverse vision language tasks.
We dig into vision-based deductive reasoning, a more sophisticated but less explored realm, and find previously unexposed blindspots in the current SOTA VLMs.
Specifically, we leverage Raven’s Progressive Matri... | [
"Vision Language Model",
"Reasoning"
] | VLMs struggles with visual deductive reasoning like solving Raven’s Progressive Matrices (RPMs) due to difficulty perceiving and comprehending multiple abstract patterns solely from visual cues. | 146 | 2403.04732 | title_snapshot |
S7NVVfuRv8 | How Easily do Irrelevant Inputs Skew the Responses of Large Language Models? | https://openreview.net/forum?id=S7NVVfuRv8 | [
"Siye Wu",
"Jian Xie",
"Jiangjie Chen",
"Tinghui Zhu",
"Kai Zhang",
"Yanghua Xiao"
] | null | null | By leveraging the retrieval of information from external knowledge databases, Large Language Models (LLMs) exhibit enhanced capabilities for accomplishing many knowledge-intensive tasks.
However, due to the inherent flaws of current retrieval systems, there might exist irrelevant information within those retrieving top... | [
"Retrieval Augmented Generation",
"Irrelevant Information",
"Misleading Information"
] | In this paper, we introduce a framework for categorizing irrelevant information into three graded levels, aiming to explore LLMs' robustness when encountering graded semantic related yet irrelevant information under various conditions. | 148 | 2404.03302 | title_snapshot |
sKATR2O1Y0 | OpenAgents: An Open Platform for Language Agents in the Wild | https://openreview.net/forum?id=sKATR2O1Y0 | [
"Tianbao Xie",
"Fan Zhou",
"Zhoujun Cheng",
"Peng Shi",
"Luoxuan Weng",
"Yitao Liu",
"Toh Jing Hua",
"Junning Zhao",
"Qian Liu",
"Che Liu",
"Zeyu Liu",
"Yiheng Xu",
"Hongjin SU",
"Dongchan Shin",
"Caiming Xiong",
"Tao Yu"
] | null | null | Language agents show potential in being capable of utilizing natural language for varied and intricate tasks in diverse environments, particularly when built upon large language models (LLMs). Current language agent frameworks aim to facilitate the construction of proof-of-concept language agents while neglecting the n... | [
"Language Agents",
"Large Language Models (LLMs)",
"Data Analysis",
"Web Browsing Automation",
"API Tools"
] | OpenAgents is an open platform facilitating easy access to language agents, offering three agents for data analysis, API tools, and web browsing, aiming to foster innovation and real-world evaluations in the realm of language agents. | 159 | 2310.10634 | title_snapshot |
q36rpGlG9X | Large Language Models as Biomedical Hypothesis Generators: A Comprehensive Evaluation | https://openreview.net/forum?id=q36rpGlG9X | [
"Biqing Qi",
"Kaiyan Zhang",
"Kai Tian",
"Haoxiang Li",
"Zhang-Ren Chen",
"Sihang Zeng",
"Ermo Hua",
"Hu Jinfang",
"Bowen Zhou"
] | null | null | The rapid growth of biomedical knowledge has outpaced our ability to efficiently extract insights and generate novel hypotheses.
Large language models (LLMs) have emerged as a promising tool to revolutionize knowledge interaction and potentially accelerate biomedical discovery. In this paper, we present a comprehensive... | [
"large language models",
"biomedicine",
"scientific discovery",
"zero-shot",
"multi-agent"
] | Large language models surprisingly demonstrate the ability to generate valid, novel scientific hypotheses from existing knowledge, suggesting their potential to catalyze new discoveries when leveraged with techniques that increase uncertainty. | 161 | 2407.08940 | title_snapshot |
Xh1B90iBSR | What Are Tools Anyway? A Survey from the Language Model Perspective | https://openreview.net/forum?id=Xh1B90iBSR | [
"Zhiruo Wang",
"Zhoujun Cheng",
"Hao Zhu",
"Daniel Fried",
"Graham Neubig"
] | null | null | Language models (LMs) are powerful yet mostly for text generation tasks. Tools have substantially enhanced their performance for tasks that require complex skills. However, many works adopt the term “tool” in different ways, raising the question: What is a tool anyway? Subsequently, where and how do tools help LMs? In ... | [
"tool",
"agent",
"program"
] | In this survey, we formally define what is a tool, when and how tools help, and existing issues with tool-related works. | 162 | 2403.15452 | title_snapshot |
D06yk3DBas | Can It Edit? Evaluating the Ability of Large Language Models to Follow Code Editing Instructions | https://openreview.net/forum?id=D06yk3DBas | [
"Federico Cassano",
"Luisa Li",
"Akul Sethi",
"Noah Shinn",
"Abby Brennan-Jones",
"Jacob Ginesin",
"Edward Berman",
"George Chakhnashvili",
"Anton Lozhkov",
"Carolyn Jane Anderson",
"Arjun Guha"
] | null | null | A significant amount of research is focused on developing and evaluating
large language models for a variety of code synthesis tasks. These include
synthesizing code from natural language, synthesizing tests from
code, and synthesizing explanations of code. In contrast, the behavior of
instructional code editing with ... | [
"benchmark",
"programming",
"llm",
"code editing"
] | CanItEdit evaluates the instructional code editing capabilities of large language models, reveals a performance gap between open and closed models, and demonstrates that fine-tuning with a new dataset improves performance of open models. | 171 | 2312.12450 | title_snapshot |
nGCMLATBit | Eliciting Latent Knowledge from "Quirky" Language Models | https://openreview.net/forum?id=nGCMLATBit | [
"Alex Troy Mallen",
"Madeline Brumley",
"Julia Kharchenko",
"Nora Belrose"
] | null | null | Eliciting Latent Knowledge (ELK) aims to find patterns in a capable neural network's activations that robustly track the true state of the world, especially in hard-to-verify cases where the model's output is untrusted. To further ELK research, we introduce 12 datasets and a corresponding suite of "quirky" language mod... | [
"Eliciting Latent Knowledge",
"Scalable Oversight",
"Honesty",
"Generalization"
] | We create a benchmark for Eliciting Latent Knowledge methods and find that simple probes can report reliable knowledge when the LM output is untrusted, even for questions harder than those used for training the probe. | 173 | 2312.01037 | title_snapshot |
3HTVP34WWE | Bot or Human? Detecting ChatGPT Imposters with A Single Question | https://openreview.net/forum?id=3HTVP34WWE | [
"Hong Wang",
"Xuan Luo",
"Weizhi Wang",
"Melody Yu",
"Xifeng Yan"
] | null | null | Large language models (LLMs) like GPT-4 have recently demonstrated impressive capabilities in natural language understanding and generation. However, there is a concern that they can be misused for malicious purposes, such as fraud or denial-of-service attacks. Therefore, it is crucial to develop methods for detecting ... | [
"ChatGPT imposter detection",
"large language models"
] | In this paper, we propose a framework named FLAIR, Finding Large language model Authenticity via a single Inquiry and Response, to detect conversational bots in an online manner | 182 | 2305.06424 | title_snapshot |
dcbNzhVVQj | Learning From Correctness Without Prompting Makes LLM Efficient Reasoner | https://openreview.net/forum?id=dcbNzhVVQj | [
"Yuxuan YAO",
"Han Wu",
"Zhijiang Guo",
"Zhou Biyan",
"Jiahui Gao",
"Sichun Luo",
"Hanxu Hou",
"Xiaojin Fu",
"Linqi Song"
] | null | null | Large language models (LLMs) have demonstrated outstanding performance across various tasks, yet they still exhibit limitations such as hallucination, unfaithful reasoning, and toxic content. One potential approach to mitigate these issues is learning from human or external feedback (e.g. tools). In this paper, we intr... | [
"LLM",
"Reasoning",
"Self-refine"
] | The proposed framework prioritizes learning from correct reasoning steps and measures confidence for each reasoning step based on generation logits for better multi-step reasoning. | 204 | 2403.19094 | title_snapshot |
UyNIH6CWHH | Efficient Parallelization Layouts for Large-Scale Distributed Model Training | https://openreview.net/forum?id=UyNIH6CWHH | [
"Johannes Hagemann",
"Samuel Weinbach",
"Konstantin Dobler",
"Maximilian Schall",
"Gerard de Melo"
] | null | null | Efficiently training large language models requires parallelizing across hundreds of hardware accelerators and invoking various compute and memory optimizations.
When combined, many of these strategies have complex interactions regarding the final training efficiency. Prior work tackling this problem did not have acc... | [
"model training efficiency",
"large models",
"parallelization",
"distributed"
] | We conduct a large sweep over LLM training setups and distill the results into recommendations for the most efficient training. | 206 | 2311.05610 | title_snapshot |
qHdSA85GyZ | Look at the Text: Instruction-Tuned Language Models are More Robust Multiple Choice Selectors than You Think | https://openreview.net/forum?id=qHdSA85GyZ | [
"Xinpeng Wang",
"Chengzhi Hu",
"Bolei Ma",
"Paul Rottger",
"Barbara Plank"
] | null | null | Multiple choice questions (MCQs) are commonly used to evaluate the capabilities of large language models (LLMs). One common way to evaluate the model response is to rank the candidate answers based on the log probability of the first token prediction. An alternative way is to examine the text output. Prior work has sho... | [
"Robustness",
"LLM Evaluation",
"Multiple-Choice Question"
] | We show text answers from instruction-tuned language models have less selection bias than first token-based answers and are more robust to various prompt perturbations. | 207 | 2404.08382 | title_snapshot |
bo4pauxnIR | Tabular Transfer Learning via Prompting LLMs | https://openreview.net/forum?id=bo4pauxnIR | [
"Jaehyun Nam",
"Woomin Song",
"Seong Hyeon Park",
"Jihoon Tack",
"Sukmin Yun",
"Jaehyung Kim",
"Kyu Hwan Oh",
"Jinwoo Shin"
] | null | null | Learning with a limited number of labeled data is a central problem in real-world applications of machine learning, as it is often expensive to obtain annotations. To deal with the scarcity of labeled data, transfer learning is a conventional approach; it suggests to learn a transferable knowledge by training a neural ... | [
"Tabular learning",
"Transfer learning",
"In-context learning"
] | We propose a novel tabular transfer learning framework that leverages LLM's in-context learning capabilities. | 210 | 2408.11063 | title_snapshot |
Lmjgl2n11u | Beyond Accuracy: Evaluating the Reasoning Behavior of Large Language Models - A Survey | https://openreview.net/forum?id=Lmjgl2n11u | [
"Philipp Mondorf",
"Barbara Plank"
] | null | null | Large language models (LLMs) have recently shown impressive performance on tasks involving reasoning, leading to a lively debate on whether these models possess reasoning capabilities similar to humans. However, despite these successes, the depth of LLMs' reasoning abilities remains uncertain. This uncertainty partly s... | [
"Reasoning in Large Language Models",
"Reasoning Behavior",
"Evaluating Reasoning",
"Logical Reasoning",
"Mathematical Reasoning",
"Causal Reasoning"
] | This survey presents the first comprehensive review of literature that evaluates LLM-based reasoning beyond task accuracy, offering deeper insights into the models' reasoning behavior. | 214 | 2404.01869 | title_snapshot |
QJvfpWSpWm | The Larger the Better? Improved LLM Code-Generation via Budget Reallocation | https://openreview.net/forum?id=QJvfpWSpWm | [
"Michael Hassid",
"Tal Remez",
"Jonas Gehring",
"Roy Schwartz",
"Yossi Adi"
] | null | null | It is a common belief that large language models (LLMs) are better than smaller-sized ones. However, larger models also require significantly more time and compute during inference. This begs the question: what happens when both models operate under the same budget? (e.g., compute, run-time). To address this question, ... | [
"code-generation",
"budget constraint",
"model size"
] | We compare between small and large LMs in code-generation tasks when both are given the same budget constraint; we show that smaller models often outperform large ones in this setting | 217 | 2404.00725 | title_snapshot |
CI7D2kiih1 | Should We Attend More or Less? Modulating Attention for Fairness | https://openreview.net/forum?id=CI7D2kiih1 | [
"Abdelrahman Zayed",
"Goncalo Mordido",
"Samira Shabanian",
"Sarath Chandar"
] | null | null | The advances in natural language processing (NLP) pose both opportunities and challenges. While recent progress enables the development of high-performing models for a variety of tasks, it also poses the risk of models learning harmful biases from the data, such as gender stereotypes. In this work, we investigate the r... | [
"attention modulation",
"bias",
"fairness",
"intra-processing",
"language models",
"attention",
"gender",
"race",
"religion",
"sexual orientation"
] | This paper presents a novel approach to improve fairness in language models using attention modulation. | 222 | 2305.13088 | title_snapshot |
DRffhKBVlE | LITE: Modeling Environmental Ecosystems with Multimodal Large Language Models | https://openreview.net/forum?id=DRffhKBVlE | [
"Haoran Li",
"Junqi Liu",
"Zexian Wang",
"Shiyuan Luo",
"Xiaowei Jia",
"Huaxiu Yao"
] | null | null | The modeling of environmental ecosystems plays a pivotal role in the sustainable management of our planet. Accurate prediction of key environmental variables over space and time can aid in informed policy and decision-making, thus improving people's livelihood. Recently, deep learning-based methods have shown promise i... | [
"Environmental Ecosystems",
"Foundation Models",
"Spatio-temporal Data",
"Multimodal Learning"
] | A novel multimodal large language model for cross-domain environmental ecosystem modeling, demonstrating SOTA performance across all domains and maintaining robustness to incomplete observations and distribution shifts in environmental data. | 226 | 2404.01165 | title_snapshot |
KidynPuLNW | On Limitations of the Transformer Architecture | https://openreview.net/forum?id=KidynPuLNW | [
"Binghui Peng",
"Srini Narayanan",
"Christos Papadimitriou"
] | null | null | What are the root causes of hallucinations in large language models (LLMs)? We use Communication Complexity to prove that the Transformer layer is incapable of composing functions (e.g., identify a grandparent of a person in a genealogy) if the domains of the functions are large enough; we show through examples that t... | [
"Transformer",
"computation complexity",
"communication complexity"
] | The power and limitation of Transformer architecture | 229 | 2402.08164 | title_snapshot |
8TdcXwfNRB | PAPERCLIP: Associating Astronomical Observations and Natural Language with Multi-Modal Models | https://openreview.net/forum?id=8TdcXwfNRB | [
"Siddharth Mishra-Sharma",
"YIDING SONG",
"Jesse Thaler"
] | null | null | We present PAPERCLIP (Proposal Abstracts Provide an Effective Representation for Contrastive Language-Image Pre-training), a method which associates astronomical observations imaged by telescopes with natural language using a neural network model. The model is fine-tuned from a pre-trained Contrastive Language-Image Pr... | [
"CLIP",
"astronomy",
"astrophysics",
"physics",
"science",
"space",
"telescopes",
"vision",
"language",
"scientific discovery",
"contrastive learning",
"guided generation",
"data curation",
"fine tuning",
"foundation models"
] | We develop a method to associate astronomical images observed by telescopes with natural language in a common, semantically informative embedding space. | 230 | 2403.08851 | title_snapshot |
nMAaCsCTCI | Impact of Preference Noise on the Alignment Performance of Generative Language Models | https://openreview.net/forum?id=nMAaCsCTCI | [
"Yang Gao",
"Dana Alon",
"Donald Metzler"
] | null | null | A key requirement in developing Generative Language Models (GLMs) is to have their values aligned with human’s values. Preference-based alignment is a widely used paradigm for this purpose, in which preferences over generation pairs are first elicited from human annotators or AI systems, and then fed into some alignmen... | [
"LM Alignment",
"Noisy Data",
"Preference Learning"
] | We studied how noise in preference pairs affect the alignment performance, and how to mitigate its negative impact. | 233 | 2404.09824 | title_snapshot |
JXcXnJJSuL | Information-Theoretic Distillation for Reference-less Summarization | https://openreview.net/forum?id=JXcXnJJSuL | [
"Jaehun Jung",
"Ximing Lu",
"Liwei Jiang",
"Faeze Brahman",
"Peter West",
"Pang Wei Koh",
"Yejin Choi"
] | null | null | The current winning recipe for automatic summarization is using proprietary large-scale language models (LLMs) such as ChatGPT as is, or imitation learning from them as teacher models. While increasingly ubiquitous dependence on such large-scale language models is convenient, there remains an important question of whet... | [
"summarization",
"self-training",
"distillation",
"PMI maximization"
] | We present a novel framework to distill a powerful summarizer based on the information-theoretic objective for summarization, without relying on either the LLM’s capability or human-written references. | 236 | 2403.13780 | title_snapshot |
YwrNePfb3E | Prompt Exploration with Prompt Regression | https://openreview.net/forum?id=YwrNePfb3E | [
"Michael Feffer",
"Ronald Xu",
"Yuekai Sun",
"Mikhail Yurochkin"
] | null | null | In the advent of democratized usage of large language models (LLMs), there is a growing desire to systematize LLM prompt creation and selection processes beyond iterative trial-and-error. Prior works majorly focus on searching the space of prompts without accounting for relations between prompt variations. Here we prop... | [
"prompt engineering",
"prompt selection",
"large language models",
"regression"
] | We introduce a regression method to predict effects of prompt combinations as well as select effective prompts based on the regression model accordingly. | 239 | 2405.11083 | title_snapshot |
v3w2a7EInO | CATS: Context-Aware Thresholding for Sparsity in Large Language Models | https://openreview.net/forum?id=v3w2a7EInO | [
"Donghyun Lee",
"Jaeyong Lee",
"Genghan Zhang",
"Mo Tiwari",
"Azalia Mirhoseini"
] | null | null | The dramatic improvements in Large Language Models (LLMs) come at the cost of increased computational resources for inference. Recent studies ameliorate the computational costs of LLMs by increasing their activation sparsity but suffer from significant performance degradation on downstream tasks.
In this work, we intro... | [
"efficient inference",
"sparsity",
"context-aware inference"
] | We make LLMs faster and more efficient by inducing sparsity; our prescription for doing so applies to many LLMs. | 253 | null | null |
MLD1cwfjUb | Your Context Is Not an Array: Unveiling Random Access Limitations in Transformers | https://openreview.net/forum?id=MLD1cwfjUb | [
"MohammadReza Ebrahimi",
"Sunny Panchal",
"Roland Memisevic"
] | null | null | Despite their recent successes, Transformer-based large language models show surprising failure modes. A well-known example of such failure modes is their inability to length-generalize: solving problem instances at inference time that are longer than those seen during training. In this work, we further explore the roo... | [
"Transformers",
"Large Language Models",
"Length Generalization",
"Arithmetic Tasks",
"Context Addressing"
] | We provide analysis that suggests failure of Transformers in length generalization is intricately linked to the model's inability to perform random memory access within its context window. | 255 | 2408.05506 | title_snapshot |
UfWwBaLuXV | List Items One by One: A New Data Source and Learning Paradigm for Multimodal LLMs | https://openreview.net/forum?id=UfWwBaLuXV | [
"An Yan",
"Zhengyuan Yang",
"Junda Wu",
"Wanrong Zhu",
"Jianwei Yang",
"Linjie Li",
"Kevin Lin",
"Jianfeng Wang",
"Julian McAuley",
"Jianfeng Gao",
"Lijuan Wang"
] | null | null | Set-of-Mark (SoM) Prompting unleashes the visual grounding capability of GPT-4V, by enabling the model to associate visual objects with tags inserted on the image. These tags, marked with alphanumerics, can be indexed via text tokens for easy reference. Despite the extraordinary performance from GPT-4V, we observe that... | [
"multimodal LLMs",
"Synthetic Data",
"Learning Paradigm"
] | A new data source and learning paradigm for training multimodal LLMs | 257 | 2404.16375 | title_snapshot |
CrzAj0kZjR | STaR-GATE: Teaching Language Models to Ask Clarifying Questions | https://openreview.net/forum?id=CrzAj0kZjR | [
"Chinmaya Andukuri",
"Jan-Philipp Fränken",
"Tobias Gerstenberg",
"Noah Goodman"
] | null | null | When prompting language models to complete a task, users often leave important aspects unsaid. While asking questions could resolve this ambiguity (GATE; Li et al., 2023), models often struggle to ask good questions. We explore a language model's ability to self-improve (STaR; Zelikman et al., 2022) by rewarding the m... | [
"question asking",
"preference elicitation",
"expert iteration",
"self-improvement"
] | We explore a language model's ability to self-improve by rewarding the model for generating useful questions. | 260 | 2403.19154 | title_snapshot |
oSG6qGkt1I | Reasoning about concepts with LLMs: Inconsistencies abound | https://openreview.net/forum?id=oSG6qGkt1I | [
"Rosario Uceda Sosa",
"Karthikeyan Natesan Ramamurthy",
"Maria Chang",
"Moninder Singh"
] | null | null | The ability to summarize and organize knowledge into abstract concepts is key to learning and reasoning. Many industrial applications rely on the consistent and systematic use of concepts, especially when dealing with decision-critical knowledge. However, we demonstrate that, when methodically questioned, large languag... | [
"KG reasoning in LLMs",
"LLM consistency",
"Synthetic data generation for LLM evaluation",
"RAG",
"prompt engineering"
] | Through systematic testing of simple knowledge graphs, we've discovered plenty of inconsistent reasoning in LLMs. Here we discuss the test results and ways to make the LLMs more robust. | 262 | 2405.20163 | title_snapshot |
HVK6nl3i97 | TriForce: Lossless Acceleration of Long Sequence Generation with Hierarchical Speculative Decoding | https://openreview.net/forum?id=HVK6nl3i97 | [
"Hanshi Sun",
"Zhuoming Chen",
"Xinyu Yang",
"Yuandong Tian",
"Beidi Chen"
] | null | null | With large language models (LLMs) widely deployed in long content generation recently, there has emerged an increasing demand for efficient long-sequence inference support. However, key-value (KV) cache, which is stored to avoid re-computation, has emerged as a critical bottleneck by growing linearly in size with the s... | [
"Long-context model",
"Speculative decoding",
"LLM efficiency"
] | Hierarchical Speculative Decoding for Lossless Acceleration of Long Sequence Generation | 264 | 2404.11912 | title_snapshot |
1ba209BACA | Agent-DocEdit: Language-Instructed LLM Agent for Content-Rich Document Editing | https://openreview.net/forum?id=1ba209BACA | [
"Te-Lin Wu",
"Rajiv Jain",
"Yufan Zhou",
"Puneet Mathur",
"Vlad I Morariu"
] | null | null | Editing content-rich and multimodal documents, such as posters, flyers, and slides, can be tedious if the edits are complex, repetitive, or require subtle skills and deep knowledge of the editing software.
Motivated by recent advancements in both Large Language Model (LLM) agents and multimodal modeling, we propose a f... | [
"document editing",
"multimodal",
"visual programming"
] | We propose an LLM agent framework that can automate the editing requests for content-rich documents via visual programming, grounding, and feedback mechanisms. | 265 | null | null |
95TayIeqJ4 | TMMLU+: An Improved Traditional Chinese Evaluation Suite for Foundation Models | https://openreview.net/forum?id=95TayIeqJ4 | [
"Zhi Rui Tam",
"Ya Ting Pai",
"Yen-Wei Lee",
"Hong-Han Shuai",
"Jun-Da Chen",
"Wei Min Chu",
"Sega Cheng"
] | null | null | We present TMMLU+, a new benchmark designed for Traditional Chinese language understanding. TMMLU+ is a multi-choice question-answering dataset with 66 subjects from elementary to professional level. It is six times larger and boasts a more balanced subject distribution than its predecessor, Taiwan Massive Multitask La... | [
"Traditional Chinese evaluation",
"multi-choice question answering"
] | We introduce TMMLU+, a comprehensive Traditional Chinese dataset covering over 66 subjects, with insights into tokenization effects and the impacts of Simplified versus Traditional Chinese on LLM performance. | 275 | 2403.01858 | title_judge |
9Wmdk94oKF | CHOPS: CHat with custOmer Profile Systems for Customer Service with LLMs | https://openreview.net/forum?id=9Wmdk94oKF | [
"Jingzhe Shi",
"Jialuo Li",
"Qinwei Ma",
"Zaiwen Yang",
"Huan Ma",
"Lei Li"
] | null | null | Businesses and software platforms are increasingly utilizing Large Language Models (LLMs) like GPT-3.5, GPT-4, GLM-3, and LLaMa-2 as chat assistants with file access or as reasoning agents for custom service. Current LLM-based customer service models exhibit limited integration with customer profiles and lack operation... | [
"LLM for Customer Service",
"LLM agents",
"benchmark"
] | We introduce CHOPS, leveraging LLMs to seamlessly interface with existing customer profile systems, and validate its efficiency through experiments with our proposed dataset, CPHOS-dataset, thereby aiming to enhance customer service performance. | 276 | 2404.01343 | title_snapshot |
0UK8c2kg7c | InstructAV: Instruction Fine-tuning Large Language Models for Authorship Verification | https://openreview.net/forum?id=0UK8c2kg7c | [
"Yujia Hu",
"Zhiqiang Hu",
"Chun Wei Seah",
"Roy Ka-Wei Lee"
] | null | null | Large Language Models (LLMs) have demonstrated remarkable proficiency in a wide range of NLP tasks. However, when it comes to authorship verification (AV) tasks, which involve determining whether two given texts share the same authorship, even advanced models like ChatGPT exhibit notable limitations. This paper introdu... | [
"Authorship Verification",
"Parameter-Efficient Fine-Tuning (PEFT)",
"InstructAV"
] | We propose the InstructAV framework for Authorship Verification (AV) tasks to accurately determine whether two texts share the same author and to furnish robust linguistic explanations for the AV outcomes. | 286 | 2407.12882 | title_snapshot |
wS7PxDjy6m | Dated Data: Tracing Knowledge Cutoffs in Large Language Models | https://openreview.net/forum?id=wS7PxDjy6m | [
"Jeffrey Cheng",
"Marc Marone",
"Orion Weller",
"Dawn Lawrie",
"Daniel Khashabi",
"Benjamin Van Durme"
] | null | null | Large Language Models (LLMs) are often paired with a reported cutoff date, the time at which training data was gathered.
Such information is crucial for applications where the LLM must provide up-to-date information. However, a reported cutoff only scratches the surface. Do all sub-resources in the training data share... | [
"knowledge cutoffs",
"training data",
"temporal alignment"
] | Singular knowledge cutoff dates do not capture the entirety of LLM training corpora, so we design a simple probing method using time spanning datasets and analyze a large set of open access pretraining corpora. | 289 | 2403.12958 | title_snapshot |
yfyHxvVzZT | Does Incomplete Syntax Influence Korean Language Model? Focusing on Word Order and Case Markers | https://openreview.net/forum?id=yfyHxvVzZT | [
"Jong Myoung Kim",
"Young-Jun Lee",
"Yong-Jin Han",
"Ho-Jin Choi",
"Sangkeun Jung"
] | null | null | Syntactic elements, such as word order and case markers, are fundamental in natural language processing. Recent studies show that syntactic information boosts language model performance and offers clues for people to understand their learning mechanisms. Unlike languages with a fixed word order such as English, Korean ... | [
"syntax",
"word order",
"postposition",
"case marker",
"syntactically-incomplete data"
] | We confirmed the syntactic flexibility of Korean, examined if LLMs capture this feature, and applied it as a data augmentation method to enhance performance. | 298 | 2407.09184 | title_snapshot |
taThoOlDNQ | Exploring the Mystery of Influential Data for Mathematical Reasoning | https://openreview.net/forum?id=taThoOlDNQ | [
"Xinzhe Ni",
"Yeyun Gong",
"Zhibin Gou",
"yelong shen",
"Yujiu Yang",
"Nan Duan",
"Weizhu Chen"
] | null | null | Selecting influential data for fine-tuning on downstream tasks is a key factor for both performance and computation efficiency. Recent works have shown that training with only limited data can show a superior performance on general tasks. However, the feasibility on mathematical reasoning tasks has not been validated. ... | [
"Influential Data Composition",
"Mathematical Reasoning"
] | For mathematical reasoning, we first propose a Quality-aware Diverse Selection (QaDS) strategy to select influential data, and then construct OpenMathMix, an influential data mixture with open-source data selected by QaDS. | 310 | 2404.01067 | title_snapshot |
TZ0CCGDcuT | Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanisms | https://openreview.net/forum?id=TZ0CCGDcuT | [
"Michael Hanna",
"Sandro Pezzelle",
"Yonatan Belinkov"
] | null | null | Many recent language model (LM) interpretability studies have adopted the circuits framework, which aims to find the minimal computational subgraph, or circuit, that explains LM behavior on a given task. Most studies determine which edges belong in a LM's circuit for a task by performing causal interventions on each ed... | [
"interpretability",
"mechanistic interpretability",
"circuits"
] | We introduce a new automated circuit-finding method, and show that it is more faithful than its predecessor. | 317 | 2403.17806 | title_snapshot |
YDZ7GeFLxq | Scattered Mixture-of-Experts Implementation | https://openreview.net/forum?id=YDZ7GeFLxq | [
"Shawn Tan",
"Yikang Shen",
"Rameswar Panda",
"Aaron Courville"
] | null | null | ScatterMoE is an implementation of Sparse Mixture-of-Experts (SMoE) on GPUs. ScatterMoE builds upon techniques in existing implementations, and overcoming some of the current limitations to improve batched inference, training speed, and memory footprint. This implementation achieves this by avoiding padding and making ... | [
"triton",
"gpu",
"moe",
"mixture of experts",
"sparse mixture of experts"
] | Triton-based implementation of Sparse Mixture of Experts without padded copies. | 322 | 2403.08245 | title_snapshot |
kzzwTrt04Z | AI-generated text boundary detection with RoFT | https://openreview.net/forum?id=kzzwTrt04Z | [
"Laida Kushnareva",
"Tatiana Gaintseva",
"Dmitry Abulkhanov",
"Kristian Kuznetsov",
"German Magai",
"Eduard Tulchinskii",
"Serguei Barannikov",
"Sergey Nikolenko",
"Irina Piontkovskaya"
] | null | null | Due to the rapid development of large language models, people increasingly often encounter texts that may start as written by a human but continue as machine-generated. Detecting the boundary between human-written and machine-generated parts of such texts is a challenging problem that has not received much attention in... | [
"artificial text detection",
"cross-model detection",
"cross-domain detection",
"boundary detection",
"interpretability",
"analysis"
] | We examine a number of ways to adapt up-to-date artificial text detection classifiers for the understudied task of detecting the boundary between human text and LLM generation in mixed human-AI content in cross-domain and cross-model settings. | 324 | 2311.08349 | title_snapshot |
dkpeWQRmlc | HDT: Hierarchical Document Transformer | https://openreview.net/forum?id=dkpeWQRmlc | [
"Haoyu He",
"Markus Flicke",
"Jan Buchmann",
"Iryna Gurevych",
"Andreas Geiger"
] | null | null | In this paper, we propose the Hierarchical Document Transformer (HDT), a novel sparse Transformer architecture tailored for structured hierarchical documents. Such documents are extremely important in numerous domains, including science, law or medicine. However, most existing solutions are inefficient and fail to make... | [
"compute & memory efficient Transformer",
"sparse attention",
"encoder-only",
"encoder-decoder",
"long-text Transformer"
] | A compute efficient Transformer using hierarchical sparse attention. | 326 | 2407.08330 | title_snapshot |
xm8zYRfrqE | Studying Large Language Model Behaviors Under Context-Memory Conflicts With Real Documents | https://openreview.net/forum?id=xm8zYRfrqE | [
"Evgenii Kortukov",
"Alexander Rubinstein",
"Elisa Nguyen",
"Seong Joon Oh"
] | null | null | Retrieval-augmented generation (RAG) mitigates many problems of fully parametric language models, such as temporal degradation, hallucinations, and lack of grounding. In RAG, the model’s knowledge can be updated from documents provided in context. This leads to cases of conflict between the model’s parametric knowledge... | [
"large language model",
"retrieval augmentation",
"knowledge conflict"
] | Studying context-memory knowledge conflicts as they appear in practice: how does factual knowledge influence LLM reading behaviors? | 328 | 2404.16032 | title_snapshot |
dj9x6JuiD5 | With Greater Text Comes Greater Necessity: Inference-Time Training Helps Long Text Generation | https://openreview.net/forum?id=dj9x6JuiD5 | [
"Yan Wang",
"Dongyang Ma",
"Deng Cai"
] | null | null | Long text generation, such as novel writing and discourse-level translation with extremely long contexts, presents significant challenges to current language models. Existing methods mainly focus on extending the model's context window through strategies like length extrapolation. However, these approaches demand subst... | [
"Inference-Time Training",
"Long Context"
] | Temp-Lora embeds the infinite-long-context information directly into the model's parameters via a temporary Lora module, thus substantially enhances generation quality for long texts | 329 | 2401.11504 | title_snapshot |
7jSMMvXLri | Measuring Taiwanese Mandarin Language Understanding | https://openreview.net/forum?id=7jSMMvXLri | [
"Po-Heng Chen",
"Sijia Cheng",
"Wei-Lin Chen",
"Yen-Ting Lin",
"Yun-Nung Chen"
] | null | null | The evaluation of large language models (LLMs) has drawn substantial attention in the field recently.
This work focuses on evaluating LLMs in a Chinese context, specifically, for Traditional Chinese which has been largely underrepresented in existing benchmarks.
We present TMLU, a comprehensive evaluation suit tailored... | [
"Large Language Models",
"Taiwanese Mandarin",
"Traditional Chinese",
"Benchmark"
] | A comprehensive benchmark for evaluating advanced knowledge and reasoning capability in LLMs under the context of Taiwanese Mandarin. | 331 | 2403.20180 | title_snapshot |
Nd950RAcCW | Multi-hop Question Answering under Temporal Knowledge Editing | https://openreview.net/forum?id=Nd950RAcCW | [
"Keyuan Cheng",
"Gang Lin",
"Haoyang Fei",
"Yuxuan Zhai",
"Lu Yu",
"Muhammad Asif Ali",
"Lijie Hu",
"Di Wang"
] | null | null | Multi-hop question answering (MQA) under knowledge editing (KE) has garnered significant attention in the era of large language models. However, existing models for MQA under KE exhibit poor performance when dealing with questions containing explicit temporal contexts. To address this limitation, we propose a novel fra... | [
"Knowledge Editing",
"Multi-hop QA"
] | Multi-hop Question Answering under Temporal Knowledge Editing | 336 | 2404.00492 | title_snapshot |
KqK5XcgEhR | Empowering Large Language Model Agents through Action Learning | https://openreview.net/forum?id=KqK5XcgEhR | [
"Haiteng Zhao",
"Chang Ma",
"Guoyin Wang",
"Jing Su",
"Lingpeng Kong",
"Jingjing Xu",
"Zhi-Hong Deng",
"Hongxia Yang"
] | null | null | Large Language Model (LLM) Agents have recently garnered increasing interest yet they are limited in their ability to learn from trial and error, a key element of intelligent behavior. In this work, we argue that the capacity to learn new actions from experience is fundamental to the advancement of learning in LLM agen... | [
"Large Language Model Agent",
"Agent",
"Large Language Model"
] | We introduce a framework LearnAct with an iterative learning strategy to improve learnable action space. | 338 | 2402.15809 | title_snapshot |
DMUGTMWrKZ | Enhancing Adversarial Robustness of LLMs with Analytic Hierarchy Process | https://openreview.net/forum?id=DMUGTMWrKZ | [
"Jiahao Zhao",
"Minzheng Wang",
"Nan Xu",
"Yin Luo",
"Wenji Mao"
] | null | null | With the increasing impact of large language models (LLMs) across diverse applications, ensuring the robustness of LLMs has become a pressing concern. Existing defense strategies are tailored to specific attack scenarios, which typically require high-cost model training and cannot rapidly respond to new threats. To tac... | [
"Adversarial Robustness",
"Analytic Hierarchy Process",
"AI Feedback"
] | Drawing inspiration from cognitive theory, we introduce an innovative Analytic Hierarchy Process (AHP) inference framework to enhance the adversarial robustness of LLMs. | 343 | null | null |
Dt6qXZsgaU | Self-Guide: Better Task-Specific Instruction Following via Self-Synthetic Finetuning | https://openreview.net/forum?id=Dt6qXZsgaU | [
"Chenyang Zhao",
"Xueying Jia",
"Vijay Viswanathan",
"Graham Neubig",
"Tongshuang Wu"
] | null | null | Large language models (LLMs) hold the promise of solving diverse tasks when provided with appropriate natural language prompts. However, prompting often leads models to make predictions with lower accuracy compared to finetuning a model with ample training data. On the other hand, while finetuning LLMs on task-specific... | [
"instruction following",
"synthetic data",
"self-training"
] | We improve instruction-following ability in language models via self-distillation (finetuning on data generated by themselves). | 345 | 2407.12874 | title_snapshot |
eJ3cHNu7ss | HuatuoGPT-II, One-stage Training for Medical Adaption of LLMs | https://openreview.net/forum?id=eJ3cHNu7ss | [
"Junying Chen",
"Xidong Wang",
"Ke Ji",
"Anningzhe Gao",
"Feng Jiang",
"Shunian Chen",
"Hongbo Zhang",
"Song Dingjie",
"Wenya Xie",
"Chuyi Kong",
"Jianquan Li",
"Xiang Wan",
"Haizhou Li",
"Benyou Wang"
] | null | null | Adapting a language model (LM) into a specific domain, *a.k.a* domain adaption, is a common practice when specialized knowledge, e.g. medicine, is not encapsulated in a general language model like Llama2. This typically involves a two-stage process including *continued pre-training* and *supervised fine-tuning*. Implem... | [
"domain adaption",
"one-stage training",
"data sampling strategy",
"specialized LLM"
] | A unified one-stage protocol for domain adaptation and a strong Chinese medical LLM. | 346 | 2311.09774 | title_snapshot |
9gdZI7c6yr | Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators | https://openreview.net/forum?id=9gdZI7c6yr | [
"Yinhong Liu",
"Han Zhou",
"Zhijiang Guo",
"Ehsan Shareghi",
"Ivan Vulić",
"Anna Korhonen",
"Nigel Collier"
] | null | null | Large Language Models (LLMs) have demonstrated promising capabilities as automatic evaluators in assessing the quality of generated natural language. However, LLMs still exhibit biases in evaluation and often struggle to generate coherent evaluations that align with human assessments. In this work, we first conduct a s... | [
"LLM evaluator",
"pairwise comparison",
"human alignment"
] | We discuss the motivation of pairwise preference, formulate the LLM evaluation as a ranking problem and introduce a search-based ranking method that achieves sota performance. | 350 | 2403.16950 | title_snapshot |
3nTbuygoop | StateFlow: Enhancing LLM Task-Solving through State-Driven Workflows | https://openreview.net/forum?id=3nTbuygoop | [
"Yiran Wu",
"Tianwei Yue",
"Shaokun Zhang",
"Chi Wang",
"Qingyun Wu"
] | null | null | It is a notable trend to use Large Language Models (LLMs) to tackle complex tasks, e.g., tasks that require a sequence of actions and dynamic interaction with tools and external environments.
In this paper, we propose StateFlow, a novel LLM-based task-solving paradigm that conceptualizes complex task-solving processes ... | [
"workflows",
"state machines",
"LLM",
"prompt engineering"
] | We propose StateFlow, a novel LLM-based task-solving paradigm that conceptualizes complex task-solving processes backed by LLMs as state machines. | 354 | 2403.11322 | title_snapshot |
wi9IffRhVM | Guiding Language Model Reasoning with Planning Tokens | https://openreview.net/forum?id=wi9IffRhVM | [
"Xinyi Wang",
"Lucas Caccia",
"Oleksiy Ostapenko",
"Xingdi Yuan",
"William Yang Wang",
"Alessandro Sordoni"
] | null | null | Large language models (LLMs) have recently attracted considerable interest for their ability to perform complex reasoning tasks, such as chain-of-thought (CoT) reasoning. However, most of the existing approaches to enhance this ability rely heavily on data-driven methods, while neglecting the structural aspects of the ... | [
"chain-of-thought reasoning",
"fine-tuning"
] | We propose to add specialized planning tokens in front of each chain-of-thought step to guide and improve language models' math reasoning ability. | 356 | 2310.05707 | title_snapshot |
HLoWN6m4fS | Elephants Never Forget: Memorization and Learning of Tabular Data in Large Language Models | https://openreview.net/forum?id=HLoWN6m4fS | [
"Sebastian Bordt",
"Harsha Nori",
"Vanessa Cristiny Rodrigues Vasconcelos",
"Besmira Nushi",
"Rich Caruana"
] | null | null | While many have shown how Large Language Models (LLMs) can be applied to a diverse set of tasks, the critical issues of data contamination and memorization are often glossed over. In this work, we address this concern for tabular data. Specifically, we introduce a variety of different techniques to assess whether a lan... | [
"Memorization",
"Few-Shot Learning",
"Tabular Data"
] | We compare the few-shot learning abilities of LLMs on tabular datasets that they have seen during training to datasets that were released after the cutoff date of the training data. | 365 | 2404.06209 | title_snapshot |
7BCmIWVT0V | Corex: Pushing the Boundaries of Complex Reasoning through Multi-Model Collaboration | https://openreview.net/forum?id=7BCmIWVT0V | [
"Qiushi Sun",
"Zhangyue Yin",
"Xiang Li",
"Zhiyong Wu",
"Xipeng Qiu",
"Lingpeng Kong"
] | null | null | Large Language Models (LLMs) are evolving at an unprecedented pace and have exhibited considerable capability in the realm of natural language processing (NLP) with world knowledge. Benefiting from ultra-large-scale training corpora, a single LLM can manage typical NLP tasks competently. However, its performance in exe... | [
"Model Collaborations",
"Complex Reasoning",
"Large Language Models"
] | We introduce Corex, a suite of strategies designed to enhance the capabilities of LLMs in complex task-solving, with a pivotal focus on advancing multi-model collaboration. | 366 | 2310.00280 | title_snapshot |
lY6XTF9tPv | LlaSMol: Advancing Large Language Models for Chemistry with a Large-Scale, Comprehensive, High-Quality Instruction Tuning Dataset | https://openreview.net/forum?id=lY6XTF9tPv | [
"Botao Yu",
"Frazier N. Baker",
"Ziqi Chen",
"Xia Ning",
"Huan Sun"
] | null | null | Chemistry plays a crucial role in many domains, such as drug discovery and material science. While large language models (LLMs) such as GPT-4 exhibit remarkable capabilities on natural language processing tasks, existing research indicates that their performance on chemistry tasks is discouragingly low. In this paper, ... | [
"chemistry",
"LLMs",
"fine-tuning",
"instruction tuning",
"dataset",
"molecule"
] | Our LLMs that can achieve very strong results on chemistry tasks, outperforming GPT-4 and Claude 3 Opus by a substantial margin and approaching the SoTA task-specific models. | 368 | 2402.09391 | title_snapshot |
Zq9Dfj4nBo | Redesigning Information Markets in the Era of Language Models | https://openreview.net/forum?id=Zq9Dfj4nBo | [
"Martin Weiss",
"Nasim Rahaman",
"Manuel Wuthrich",
"Yoshua Bengio",
"Li Erran Li",
"Bernhard Schölkopf",
"Christopher Pal"
] | null | null | Information markets face many challenges leading to instability, inefficiency, and failure, ultimately reducing incentives for the creation and distribution of high-quality information. A long-standing issue for information markets is the Buyer's Inspection Paradox: buyers need to inspect information to assess its valu... | [
"Language Model Agents",
"Information Economics"
] | This work proposes an information market design using language models to balance buyer inspection and seller protection. | 370 | null | null |
EHPns3hVkj | Tower: An Open Multilingual Large Language Model for Translation-Related Tasks | https://openreview.net/forum?id=EHPns3hVkj | [
"Duarte Miguel Alves",
"José Pombal",
"Nuno M Guerreiro",
"Pedro Henrique Martins",
"João Alves",
"Amin Farajian",
"Ben Peters",
"Ricardo Rei",
"Patrick Fernandes",
"Sweta Agrawal",
"Pierre Colombo",
"José G. C. de Souza",
"Andre Martins"
] | null | null | While general-purpose large language models (LLMs) demonstrate proficiency on multiple tasks within the domain of translation, approaches based on open LLMs are competitive only when specializing on a single task. In this paper, we propose a recipe for tailoring LLMs to multiple tasks present in translation workflows. ... | [
"Machine Translation",
"Multilinguality",
"Adaptation",
"Continual Pretraining"
] | We propose a recipe to tailor LLMs for translation workflows. | 372 | 2402.17733 | title_snapshot |
5Evv4tIjUI | Exploiting the Potential of Seq2Seq Models as Robust Few-Shot Learners | https://openreview.net/forum?id=5Evv4tIjUI | [
"Jihyeon Lee",
"Dain Kim",
"Doohae Jung",
"Boseop Kim",
"Kyoung-Woon On"
] | null | null | In-context learning, which offers substantial advantages over fine-tuning, is predominantly observed in decoder-only models, while encoder-decoder (i.e., seq2seq) models excel in methods that rely on weight updates. Recently, a few studies have demonstrated the feasibility of few-shot learning with seq2seq models; howe... | [
"Encoder-Decoder Model",
"In-context Learning",
"Few-shot Learning"
] | Seq2seq models, with objective-aligned prompting and fusion-based methods, exhibit promising few-shot learning capabilities across diverse tasks, surpassing larger decoder-only models. | 376 | 2307.14856 | title_snapshot |
tRxIB7y3wF | LalaEval: A Holistic Human Evaluation Framework for Domain-Specific Large Language Models | https://openreview.net/forum?id=tRxIB7y3wF | [
"Chongyan Sun",
"Ken Lin",
"Shiwei Wang",
"Hulong Wu",
"Chengfei Fu",
"Zhen Wang"
] | null | null | This paper introduces LalaEval, a holistic framework designed for the human evaluation of domain-specific large language models (LLMs). LalaEval proposes a comprehensive suite of end-to-end protocols that cover five main components including domain specification, criteria establishment, benchmark dataset creation, cons... | [
"Domain-Specific Large Language Model",
"Human Evaluation",
"Holistic Framework",
"End-to-end Protocol",
"Logistic Industry"
] | This paper introduces LalaEval, a holistic framework designed for the human evaluation of domain-specific large language models and presents its deployment with real-world outcomes in the logistics industry. | 377 | 2408.13338 | title_snapshot |
dJfBejh478 | Scalable Model Editing via Customized Expert Networks | https://openreview.net/forum?id=dJfBejh478 | [
"Zihan Yao",
"Yu He",
"Tianyu Qi",
"Ming Li"
] | null | null | Addressing the issues of hallucinations and outdated knowledge in large language models is critical for their reliable application. Model Editing presents a promising avenue for mitigating these challenges in a costeffective manner. However, existing methods often suffer from unsatisfactory generalization and uninten... | [
"Large Language Model",
"Model Editing",
"Continual Learning"
] | We propose a two-stage continual training paradigm for Model Editing, which can achieve state-of-the-art performance through the addition of extra expert networks and neurons in sequence. | 378 | 2404.02699 | title_snapshot |
RCdoMrg4I0 | Chinese Tiny LLM: Pretraining a Chinese-Centered Large Language Model | https://openreview.net/forum?id=RCdoMrg4I0 | [
"Xeron Du",
"Zhouliang Yu",
"Songyang Gao",
"Ding Pan",
"Cheng Yuyang",
"Ziyang Ma",
"Ruibin Yuan",
"Xingwei Qu",
"Jiaheng Liu",
"Tianyu Zheng",
"Xinchen Luo",
"Guorui Zhou",
"Wenhu Chen",
"Ge Zhang"
] | null | null | In this study, we introduce $\textbf{CT-LLM}$, a groundbreaking 2B large language model (LLM) that illustrates a pivotal shift towards prioritizing the Chinese language in the development of LLMs. Uniquely initiated from scratch, CT-LLM diverges from the conventional methodology by primarily incorporating Chinese textu... | [
"Chinese LLM",
"pretrain",
"alignment"
] | CT-LLM, a groundbreaking 2B large language model (LLM) that illustrates a pivotal shift towards prioritizing the Chinese language in the development of LLMs. | 385 | 2404.04167 | title_judge |
oqYiYG8PtY | Stop Reasoning! When Multimodal LLM with Chain-of-Thought Reasoning Meets Adversarial Image | https://openreview.net/forum?id=oqYiYG8PtY | [
"Zefeng Wang",
"Zhen Han",
"Shuo Chen",
"Fan Xue",
"Zifeng Ding",
"Xun Xiao",
"Volker Tresp",
"Philip Torr",
"Jindong Gu"
] | null | null | Multimodal LLMs (MLLMs) with a great ability of text and image under- standing have received great attention. To achieve better reasoning with MLLMs, Chain-of-Thought (CoT) reasoning has been widely explored, which further promotes MLLMs’ explainability by giving intermediate reasoning steps. Despite the strong power d... | [
"Multimodal LLMs",
"Adversarial Robustness",
"Vision Language Models",
"Chain-of-thought"
] | We explored the influence of Chain-of-Thought on Multimodal-LLM's robustness, and proposed a novel effective attack method "stop-reasoning attack". And, the generated rationales open a window for revealing the reason the model has wrong predictions. | 386 | 2402.14899 | title_snapshot |
Qmq4zqdnWh | Using Natural Language Explanations to Rescale Human Judgments | https://openreview.net/forum?id=Qmq4zqdnWh | [
"Manya Wadhwa",
"Jifan Chen",
"Junyi Jessy Li",
"Greg Durrett"
] | null | null | The rise of large language models (LLMs) has brought a critical need for high-quality human-labeled data, particularly for processes like human feedback and evaluation. A common practice is to label data via consensus annotation over human judgments. However, annotators' judgments for subjective tasks can differ in man... | [
"human evaluation",
"natural language explanation",
"likert ratings",
"question answering",
"LLM"
] | In the era of subjective annotations, eliciting natural language explanations from human judges can surface nuances in evaluation. We propose a method that operationalizes these explanations in a principled manner. | 393 | 2305.14770 | title_snapshot |
PPTrmvEnpW | Emergent World Models and Latent Variable Estimation in Chess-Playing Language Models | https://openreview.net/forum?id=PPTrmvEnpW | [
"Adam Karvonen"
] | null | null | Language models have shown unprecedented capabilities, sparking debate over the source of their performance. Is it merely the outcome of learning syntactic patterns and surface level statistics, or do they extract semantics and a world model from the text? Prior work by Li et al. investigated this by training a GPT mod... | [
"GPT",
"large language model",
"interpretability",
"world model"
] | We train a GPT model from scratch to play chess and find that it learns to compute board state and estimate player Elo. We use these representations to edit the GPT's internal board state and increase or decrease its chess-playing ability. | 397 | 2403.15498 | title_snapshot |
S1XnUsqwr7 | Deductive Beam Search: Decoding Deducible Rationale for Chain-of-Thought Reasoning | https://openreview.net/forum?id=S1XnUsqwr7 | [
"Tinghui Zhu",
"Kai Zhang",
"Jian Xie",
"Yu Su"
] | null | null | Recent advancements have significantly augmented the reasoning capabilities of Large Language Models (LLMs) through various methodologies, especially chain-of-thought (CoT) reasoning. However, previous methods often struggle to address reasoning errors in intermediate steps, which can lead to accumulative errors. In th... | [
"Deductive Reasoning",
"Large Language Models",
"Decoding Algorithm"
] | Deductive Beam Search, constrained by the principle of deductive reasoning, integrates CoT with step-wise beam search to guide reasoning towards a deducible path. | 399 | 2401.17686 | title_snapshot |
xI8C7sfN1H | Factual and Tailored Recommendation Endorsements using Language Models and Reinforcement Learning | https://openreview.net/forum?id=xI8C7sfN1H | [
"Jihwan Jeong",
"Yinlam Chow",
"Guy Tennenholtz",
"ChihWei Hsu",
"Mohammad Ghavamzadeh",
"Craig Boutilier"
] | null | null | Recommender systems (RSs) play a central role in matching candidate items to users based on their preferences. While traditional RSs rely on user feed-back signals, conversational RSs interact with users in natural language. In this work, we develop P4LM, an _aPpealing, Precise, Preference-comprehensive and Prioritized... | [
"Large language model",
"reinforcement learning",
"conversational recommender systems",
"recommender systems"
] | Develop a conversational recommender system that uses natural language and reinforcement learning to make precise, appealing, and personalized item recommendations based on user preferences, showing promising results on major datasets. | 405 | null | null |
xMt9kCv5YR | Helmsman of the Masses? Evaluate the Opinion Leadership of Large Language Models in the Werewolf Game | https://openreview.net/forum?id=xMt9kCv5YR | [
"Silin Du",
"Xiaowei Zhang"
] | null | null | Large language models (LLMs) have exhibited memorable strategic behaviors in social deductive games. However, the significance of opinion leadership exhibited by LLM-based agents has been largely overlooked, which is crucial for practical applications in multi-agent and human-AI interaction settings. Opinion leaders ar... | [
"Opinion Leadership",
"Large Language Models",
"Werewolf Game",
"Simulation",
"Human-AI Interaction"
] | We construct a simulation platform for the Werewolf game integrating the Sheriff role, then propose two metrics to evaluate the opinion leadership of LLMs in the Werewolf game and find that few LLMs possess the capacity for opinion leadership. | 406 | 2404.01602 | title_snapshot |
wLQ3I0F1oj | Large Language Model is not a (Multilingual) Compositional Relation Reasoner | https://openreview.net/forum?id=wLQ3I0F1oj | [
"Jinman Zhao",
"Xueyan Zhang"
] | null | null | We present a comprehensive evaluation of large language models'
capability to reason compositional relations through
a benchmark encompassing 1,800 test cases in both English and Chinese,
covering six distinct categories of composition relations:
Positional, Comparative, Personal, Mathematical, Identity, and Other.... | [
"Benchmarks",
"Composition Relation"
] | We estabilish a benchmark for evaluating LLMs' composition relation reasoning | 407 | null | null |
MXLBXjQkmb | Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning | https://openreview.net/forum?id=MXLBXjQkmb | [
"Ruiqi Zhang",
"Licong Lin",
"Yu Bai",
"Song Mei"
] | null | null | Large Language Models (LLMs) often memorize sensitive, private, or copyrighted data during pre-training. LLM unlearning aims to eliminate the influence of undesirable data from the pre-trained model while preserving the model's utilities on other tasks. Several practical methods have recently been proposed for LLM unle... | [
"Unlearning",
"RLHF",
"preference optimization"
] | We propose a simple alignment-inspired objective function for machine unlearning, achieving state-of-art performance in TOFU dataset. | 412 | 2404.05868 | title_snapshot |
eDWcNqiQWW | The Curious Case of Nonverbal Abstract Reasoning with Multi-Modal Large Language Models | https://openreview.net/forum?id=eDWcNqiQWW | [
"Kian Ahrabian",
"Zhivar Sourati",
"Kexuan Sun",
"Jiarui Zhang",
"Yifan Jiang",
"Fred Morstatter",
"Jay Pujara"
] | null | null | While large language models (LLMs) are still being adopted to new domains and utilized in novel applications, we are experiencing an influx of the new generation of foundation models, namely multi-modal large language models (MLLMs). These models integrate verbal and visual information, opening new possibilities to dem... | [
"multi-modal large language models",
"nonverbal abstract reasoning",
"in-context learning",
"raven's progressive matrices"
] | A study of nonverbal reasoning abilities of multi-modal large language models using variations of Raven's Progressive Matrices. | 414 | 2401.12117 | title_snapshot |
3ypWPhMGhV | Cohesive Conversations: Enhancing Authenticity in Multi-Agent Simulated Dialogues | https://openreview.net/forum?id=3ypWPhMGhV | [
"KuanChao Chu",
"Yi-Pei Chen",
"Hideki Nakayama"
] | null | null | This paper investigates the quality of multi-agent dialogues in simulations powered by Large Language Models (LLMs). Analyzing dialogues and memory over multiple sessions revealed significant issues such as repetition, inconsistency, and hallucination, exacerbated by the propagation of erroneous information. To combat ... | [
"multi-agent dialogues",
"multi-agent communication",
"multi-session dialogues",
"machine-generated dialogues",
"inconsistencies and repetition in dialogues",
"discourse-level error detection",
"longitudinal dialogue analysis",
"generative agents",
"llm agents"
] | We investigate problems in multi-agent simulated dialogues over a span of time and propose a Screening, Diagnosis, Re-generation framework to instantly correct inconsistencies and hallucinations while bolstering multi-dialogue diversity. | 416 | 2407.09897 | title_snapshot |
gUNeyiLNxr | Uncovering Intermediate Variables in Transformers using Circuit Probing | https://openreview.net/forum?id=gUNeyiLNxr | [
"Michael A. Lepori",
"Thomas Serre",
"Ellie Pavlick"
] | null | null | Neural network models have achieved high performance on a wide variety
of complex tasks, but the algorithms that they implement are notoriously
difficult to interpret. It is often necessary to hypothesize intermediate variables involved in a network’s computation in order to understand these
algorithms. For example, do... | [
"Mechanistic Interpretability",
"Deep Learning"
] | We introduce circuit probing, a novel technique that automatically uncovers low-level circuits that compute hypothesized intermediate variables. | 418 | 2311.04354 | title_snapshot |
soz1SEiPeq | Eagle and Finch: RWKV with Matrix-Valued States and Dynamic Recurrence | https://openreview.net/forum?id=soz1SEiPeq | [
"Bo Peng",
"Daniel Goldstein",
"Quentin Gregory Anthony",
"Alon Albalak",
"Eric Alcaide",
"Stella Biderman",
"Eugene Cheah",
"Teddy Ferdinan",
"Kranthi Kiran GV",
"Haowen Hou",
"Satyapriya Krishna",
"Ronald McClelland Jr.",
"Niklas Muennighoff",
"Fares Obeid",
"Atsushi Saito",
"Guangyu... | null | null | We present Eagle (RWKV-5) and Finch (RWKV-6), sequence models improving upon the RWKV architecture. Our architectural design
advancements include multi-headed matrix-valued states and a dynamic recurrence mechanism that improve expressivity while maintaining the inference efficiency characteristics of RNNs. We introduc... | [
"large language model",
"scaling laws",
"open source",
"pretraining",
"RNN"
] | We improve upon the design of RWKV models, an RNN-based language model with computational benefits compared to transformers | 422 | 2404.05892 | title_snapshot |
Hi8jKh4HE9 | What is in Your Safe Data? Identifying Benign Data that Breaks Safety | https://openreview.net/forum?id=Hi8jKh4HE9 | [
"Luxi He",
"Mengzhou Xia",
"Peter Henderson"
] | null | null | Current Large Language Models (LLMs), even those tuned for safety and alignment, are susceptible to jailbreaking. Some have found that just further fine-tuning an aligned model with benign data (i.e., data without harmful content) surprisingly leads to substantial degradation in safety. We delve into the data-centric a... | [
"AI Safety",
"AI Alignment",
"Data Selection",
"Data Problems",
"Fine-tuning Vulnerabilities"
] | Our work seeks to understand which benign data is more likely to degrade safety after fine-tuning. We introduce representation and gradient-based methods that effectively select a small subset of benign data that jailbreaks models after fine-tuning. | 423 | 2404.01099 | title_snapshot |
B41hNBoWLo | TOFU: A Task of Fictitious Unlearning for LLMs | https://openreview.net/forum?id=B41hNBoWLo | [
"Pratyush Maini",
"Zhili Feng",
"Avi Schwarzschild",
"Zachary Chase Lipton",
"J Zico Kolter"
] | null | null | Large language models trained on massive corpora of data from the web can memorize and reproduce sensitive or private data
raising both legal and ethical concerns. Unlearning, or tuning models to forget information present in their training data, provides us with a way to protect private data after training. Although s... | [
"Machine unlearning"
] | Synthetic benchmark for machine unlearning for LLMs | 424 | 2401.06121 | title_snapshot |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- -