Title stringlengths 18 151 | Highlight stringlengths 54 701 | Authors stringlengths 9 732 |
|---|---|---|
Better & Faster Large Language Models Via Multi-token Prediction | In this work, we suggest that training language models to predict multiple future tokens at once results in higher sample efficiency. | Fabian Gloeckle; Badr Youbi Idrissi; Baptiste Roziere; David Lopez-Paz; Gabriel Synnaeve |
Transformers Are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality | While Transformers have been the main architecture behind deep learning’s success in language modeling, state-space models (SSMs) such as Mamba have recently been shown to match or outperform Transformers at small to medium scale. We show that these families of models are actually quite closely related, and develop a r... | Tri Dao; Albert Gu |
Scaling Rectified Flow Transformers for High-Resolution Image Synthesis | Despite its better theoretical properties and conceptual simplicity, it is not yet decisively established as standard practice. In this work, we improve existing noise sampling techniques for training rectified flow models by biasing them towards perceptually relevant scales. | Patrick Esser; Sumith Kulal; Andreas Blattmann; Rahim Entezari; Jonas Müller; Harry Saini; Yam Levi; Dominik Lorenz; Axel Sauer; Frederic Boesel; Dustin Podell; Tim Dockhorn; Zion English; Robin Rombach |
Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads | In this paper, we present Medusa, an efficient method that augments LLM inference by adding extra decoding heads to predict multiple subsequent tokens in parallel. | Tianle Cai; Yuhong Li; Zhengyang Geng; Hongwu Peng; Jason D. Lee; Deming Chen; Tri Dao |
Improving Factuality and Reasoning in Language Models Through Multiagent Debate | In this paper, we present a complementary approach to improve language responses where multiple language model instances propose and debate their individual responses and reasoning processes over multiple rounds to arrive at a common final answer. | Yilun Du; Shuang Li; Antonio Torralba; Joshua B. Tenenbaum; Igor Mordatch |
How Language Model Hallucinations Can Snowball | A major risk of using language models in practical applications is their tendency to hallucinate incorrect statements.To do this, we construct three question-answering datasets where LMs often state an incorrect answer which is followed by an explanation with at least one incorrect claim. | Muru Zhang; Ofir Press; William Merrill; Alisa Liu; Noah A. Smith |
Chatbot Arena: An Open Platform for Evaluating LLMs By Human Preference | This paper describes the platform, analyzes the data we have collected so far, and explains the tried-and-true statistical methods we are using for efficient and accurate evaluation and ranking of models. | Wei-Lin Chiang; Lianmin Zheng; Ying Sheng; Anastasios Nikolas Angelopoulos; Tianle Li; Dacheng Li; Banghua Zhu; Hao Zhang; Michael Jordan; Joseph E. Gonzalez; Ion Stoica |
R2E: Turning Any Github Repository Into A Programming Agent Environment | In this paper, we present Repository to Environment (R2E), a framework that can turn any GitHub repository into a test environment to evaluate the performance of code-generating systems, both static and interactive. | Naman Jain; Manish Shetty; Tianjun Zhang; King Han; Koushik Sen; Ion Stoica |
Premise Order Matters in Reasoning with Large Language Models | We first examine the effect of premise ordering on deductive reasoning on a variety of LLMs, and our evaluation shows that even if the model performance is decent on the optimal order, permuting the premise order can cause a performance drop of over 30%. In addition, we release the benchmark R-GSM, based on GSM8K, to e... | Xinyun Chen; Ryan Andrew Chi; Xuezhi Wang; Denny Zhou |
Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution | In this paper, we present Promptbreeder, a general-purpose self-referential self-improvement mechanism that evolves and adapts prompts for a given domain. | Chrisantha Fernando; Dylan Sunil Banarse; Henryk Michalewski; Simon Osindero; Tim Rocktäschel |
Stay on Topic with Classifier-Free Guidance | In this work, we demonstrate across a wide array of benchmarks that CFG can be used broadly as an inference-time technique in pure language modeling. | Guillaume Sanchez; Alexander Spangher; Honglu Fan; Elad Levi; Stella Biderman |
RLAIF Vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback | Across the tasks of summarization, helpful dialogue generation, and harmless dialogue generation, we show that RLAIF achieves comparable performance to RLHF. Furthermore, we take a step towards self-improvement by demonstrating that RLAIF can outperform a supervised fine-tuned baseline even when the AI labeler is the s... | Harrison Lee; Samrat Phatale; Hassan Mansoor; Thomas Mesnard; Johan Ferret; Kellie Ren Lu; Colton Bishop; Ethan Hall; Victor Carbune; Abhinav Rastogi; Sushant Prakash |
DisCo-Diff: Enhancing Continuous Diffusion Models with Discrete Latents | However, encoding a complex, potentially multimodal data distribution into a single *continuous* Gaussian distribution arguably represents an unnecessarily challenging learning problem. We propose ***Dis**crete-**Co**ntinuous Latent Variable **Diff**usion Models (DisCo-Diff)* to simplify this task by introducing comple... | Yilun Xu; Gabriele Corso; Tommi Jaakkola; Arash Vahdat; Karsten Kreis |
ULTRAFEEDBACK: Boosting Language Models with Scaled AI Feedback | This further hinders feedback learning as well as alignment research within the open-source community. To address this issue, we explore how to go beyond human feedback and collect high-quality AI feedback automatically for a scalable alternative. | Ganqu Cui; Lifan Yuan; Ning Ding; Guanming Yao; Bingxiang He; Wei Zhu; Yuan Ni; Guotong Xie; Ruobing Xie; Yankai Lin; Zhiyuan Liu; Maosong Sun |
Grokking Group Multiplication with Cosets | Building on previous work, we completely reverse engineer fully connected one-hidden layer networks that have “grokked” the arithmetic of the permutation groups $S_5$ and $S_6$. The models discover the true subgroup structure of the full group and converge on neural circuits that decompose the group arithmetic using th... | Dashiell Stander; Qinan Yu; Honglu Fan; Stella Biderman |
Align Your Steps: Optimizing Sampling Schedules in Diffusion Models | In this work, for the first time, we propose a general and principled approach to optimizing the sampling schedules of DMs for high-quality outputs, called Align Your Steps. | Amirmojtaba Sabour; Sanja Fidler; Karsten Kreis |
Disentangled 3D Scene Generation with Layout Learning | We introduce a method to generate 3D scenes that are disentangled into their component objects. | Dave Epstein; Ben Poole; Ben Mildenhall; Alexei A Efros; Aleksander Holynski |
Stealing Part of A Production Language Model | We introduce the first model-stealing attack that extracts precise, nontrivial information from black-box production language models like OpenAI’s ChatGPT or Google’s PaLM-2. | Nicholas Carlini; Daniel Paleka; Krishnamurthy Dj Dvijotham; Thomas Steinke; Jonathan Hayase; A. Feder Cooper; Katherine Lee; Matthew Jagielski; Milad Nasr; Arthur Conmy; Eric Wallace; David Rolnick; Florian Tramèr |
Position: Considerations for Differentially Private Learning with Large-Scale Public Pretraining | We primarily question whether the use of large Web-scraped datasets *should* be viewed as differential-privacy-preserving. | Florian Tramèr; Gautam Kamath; Nicholas Carlini |
Fast Adversarial Attacks on Language Models In One GPU Minute | In this paper, we introduce a novel class of fast, beam search-based adversarial attack (BEAST) for Language Models (LMs). | Vinu Sankar Sadasivan; Shoumik Saha; Gaurang Sriramanan; Priyatham Kattakinda; Atoosa Chegini; Soheil Feizi |
Training Large Language Models for Reasoning Through Reverse Curriculum Reinforcement Learning | In this paper, we propose **R**$^3$: Learning **R**easoning through **R**everse Curriculum **R**einforcement Learning (RL), a novel method that employs only outcome supervision to achieve the benefits of process supervision for large language models. | Zhiheng Xi; Wenxiang Chen; Boyang Hong; Senjie Jin; Rui Zheng; Wei He; Yiwen Ding; Shichun Liu; Xin Guo; Junzhe Wang; Honglin Guo; Wei Shen; Xiaoran Fan; Yuhao Zhou; Shihan Dou; Xiao Wang; Xinbo Zhang; peng sun; Tao Gui; Qi Zhang; Xuanjing Huang |
Compositional Image Decomposition with Diffusion Models | In this paper, we present a method to decompose an image into such compositional components. | Jocelin Su; Nan Liu; Yanbo Wang; Joshua B. Tenenbaum; Yilun Du |
Learning Iterative Reasoning Through Energy Diffusion | We introduce iterative reasoning through energy diffusion (IRED), a novel framework for learning to reason for a variety of tasks by formulating reasoning and decision-making problems with energy-based optimization. | Yilun Du; Jiayuan Mao; Joshua B. Tenenbaum |
Position: Compositional Generative Modeling: A Single Model Is Not All You Need | In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. | Yilun Du; Leslie Pack Kaelbling |
Potential Based Diffusion Motion Planning | We propose a new approach towards learning potential based motion planning, where we train a neural network to capture and learn an easily optimizable potentials over motion planning trajectories. | Yunhao Luo; Chen Sun; Joshua B. Tenenbaum; Yilun Du |
LESS: Selecting Influential Data for Targeted Instruction Tuning | We propose LESS, an optimizer-aware and practically efficient algorithm to estimate data influences and perform **L**ow-rank gradi**E**nt **S**imilarity **S**earch for instruction data selection.To facilitate future work, we release code and data at [princeton-nlp/LESS](https://github.com/princeton-nlp/LESS). | Mengzhou Xia; Sadhika Malladi; Suchin Gururangan; Sanjeev Arora; Danqi Chen |
Position: Data-driven Discovery with Large Generative Models | We instead advocate for fail-proof tool integration, along with active user moderation through feedback mechanisms, to foster data-driven scientific discoveries with efficiency and reproducibility. | Bodhisattwa Prasad Majumder; Harshit Surana; Dhruv Agarwal; Sanchaita Hazra; Ashish Sabharwal; Peter Clark |
Genie: Generative Interactive Environments | We introduce Genie, the first *generative interactive environment* trained in an unsupervised manner from unlabelled Internet videos. | Jake Bruce; Michael D Dennis; Ashley Edwards; Jack Parker-Holder; Yuge Shi; Edward Hughes; Matthew Lai; Aditi Mavalankar; Richie Steigerwald; Chris Apps; Yusuf Aytar; Sarah Maria Elisabeth Bechtle; Feryal Behbahani; Stephanie C.Y. Chan; Nicolas Heess; Lucy Gonzalez; Simon Osindero; Sherjil Ozair; Scott Reed; Jingwei Zh... |
Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Models | We introduce a framework called Patchscopes and show how it can be used to answer a wide range of questions about an LLM’s computation. | Asma Ghandeharioun; Avi Caciularu; Adam Pearce; Lucas Dixon; Mor Geva |
Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision | However, future superhuman models will behave in complex ways too difficult for humans to reliably evaluate; humans will only be able to *weakly supervise* superhuman models. We study an analogy to this problem: can weak model supervision elicit the full capabilities of a much stronger model? | Collin Burns; Pavel Izmailov; Jan Hendrik Kirchner; Bowen Baker; Leo Gao; Leopold Aschenbrenner; Yining Chen; Adrien Ecoffet; Manas Joglekar; Jan Leike; Ilya Sutskever; Jeffrey Wu |
Language Models with Conformal Factuality Guarantees | In this work, we propose conformal factuality, a framework that can ensure high probability correctness guarantees for LMs by connecting language modeling and conformal prediction. | Christopher Mohri; Tatsunori Hashimoto |
Equivariant Graph Neural Operator for Modeling 3D Dynamics | In this work, we propose Equivariant Graph Neural Operator (EGNO), a novel and principled method that directly models dynamics as trajectories instead of just next-step prediction. | Minkai Xu; Jiaqi Han; Aaron Lou; Jean Kossaifi; Arvind Ramanathan; Kamyar Azizzadenesheli; Jure Leskovec; Stefano Ermon; Anima Anandkumar |
Position: Open-Endedness Is Essential for Artificial Superhuman Intelligence | Nevertheless, the creation of open-ended, ever self-improving AI remains elusive. **In this position paper, we argue that the ingredients are now in place to achieve *open-endedness* in AI systems with respect to a human observer. | Edward Hughes; Michael D Dennis; Jack Parker-Holder; Feryal Behbahani; Aditi Mavalankar; Yuge Shi; Tom Schaul; Tim Rocktäschel |
Magicoder: Empowering Code Generation with OSS-Instruct | We introduce Magicoder, a series of fully open-source (code, weights, and data) Large Language Models (LLMs) for code that significantly closes the gap with top code models while having no more than 7B parameters. | Yuxiang Wei; Zhe Wang; Jiawei Liu; Yifeng Ding; LINGMING ZHANG |
HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal | Automated red teaming holds substantial promise for uncovering and mitigating the risks associated with the malicious use of large language models (LLMs), yet the field lacks a standardized evaluation framework to rigorously assess new methods. To address this issue, we introduce HarmBench, a standardized evaluation fr... | Mantas Mazeika; Long Phan; Xuwang Yin; Andy Zou; Zifan Wang; Norman Mu; Elham Sakhaee; Nathaniel Li; Steven Basart; Bo Li; David Forsyth; Dan Hendrycks |
MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities | We propose MM-Vet, an evaluation benchmark that examines large multimodal models (LMMs) on complicated multimodal tasks. | Weihao Yu; Zhengyuan Yang; Linjie Li; Jianfeng Wang; Kevin Lin; Zicheng Liu; Xinchao Wang; Lijuan Wang |
NExT-GPT: Any-to-Any Multimodal LLM | To fill the gap, we present an end-to-end general-purpose any-to-any MM-LLM system, NExT-GPT. | Shengqiong Wu; Hao Fei; Leigang Qu; Wei Ji; Tat-Seng Chua |
Linguistic Calibration of Long-Form Generations | This issue can be mitigated by having the LM verbally convey the probability that its claims are correct, but existing models cannot produce long-form text with calibrated confidence statements. Through the lens of decision-making, we define linguistic calibration for long-form generations: an LM is linguistically cali... | Neil Band; Xuechen Li; Tengyu Ma; Tatsunori Hashimoto |
Rolling Diffusion Models | This paper explores Rolling Diffusion: a new approach that uses a sliding window denoising process. | David Ruhe; Jonathan Heek; Tim Salimans; Emiel Hoogeboom |
Debating with More Persuasive LLMs Leads to More Truthful Answers | In anticipation of this, we ask: can weaker models assess the correctness of stronger models? We investigate this question in an analogous setting, where stronger models (experts) possess the necessary information to answer questions and weaker models (non-experts) lack this information. | Akbir Khan; John Hughes; Dan Valentine; Laura Ruis; Kshitij Sachan; Ansh Radhakrishnan; Edward Grefenstette; Samuel R. Bowman; Tim Rocktäschel; Ethan Perez |
IM-3D: Iterative Multiview Diffusion and Reconstruction for High-Quality 3D Generation | We significantly improve multi-view generation by considering video instead of image generators. | Luke Melas-Kyriazi; Iro Laina; Christian Rupprecht; Natalia Neverova; Andrea Vedaldi; Oran Gafni; Filippos Kokkinos |
Prismatic VLMs: Investigating The Design Space of Visually-Conditioned Language Models | Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance � a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps... | Siddharth Karamcheti; Suraj Nair; Ashwin Balakrishna; Percy Liang; Thomas Kollar; Dorsa Sadigh |
Learning to Route Among Specialized Experts for Zero-Shot Generalization | In this work, we propose $\textbf{P}$ost-$\textbf{H}$oc $\textbf{A}$daptive $\textbf{T}$okenwise $\textbf{G}$ating $\textbf{O}$ver an $\textbf{O}$cean of $\textbf{S}$pecialized $\textbf{E}$xperts (**PHATGOOSE**), which learns to route among specialized modules that were produced through parameter-efficient fine-tuning. | Mohammed Muqeeth; Haokun Liu; Yufan Liu; Colin Raffel |
Mechanistic Design and Scaling of Hybrid Architectures | The development of deep learning architectures is a resource-demanding process, due to a vast design space, long prototyping times, and high compute costs associated with at-scale model training and evaluation. We set out to simplify this process by grounding it in an end-to-end mechanistic architecture design (MAD) pi... | Michael Poli; Armin W Thomas; Eric Nguyen; Pragaash Ponnusamy; Björn Deiseroth; Kristian Kersting; Taiji Suzuki; Brian Hie; Stefano Ermon; Christopher Re; Ce Zhang; Stefano Massaroli |
Does Label Smoothing Help Deep Partial Label Learning? | In theory, we prove lower and upper bounds of the expected risk to show that label smoothing can help deep PLL. | Xiuwen Gong; Nitin Bisht; Guandong Xu |
Learning to Explore in POMDPs with Informational Rewards | In this work, we design a POMDP agent that gathers information about the hidden state, using ideas from the meta-exploration literature. | Annie Xie; Logan Mondal Bhamidipaty; Evan Zheran Liu; Joey Hong; Sergey Levine; Chelsea Finn |
Fundamental Limitations of Alignment in Large Language Models | In this paper, we propose a theoretical approach called Behavior Expectation Bounds (BEB) which allows us to formally investigate several inherent characteristics and limitations of alignment in large language models. | Yotam Wolf; Noam Wies; Oshri Avnery; Yoav Levine; Amnon Shashua |
Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model | In this paper, we show that the reliance on self-attention for visual representation learning is not necessary and propose a new generic vision backbone with bidirectional Mamba blocks (Vim), which marks the image sequences with position embeddings and compresses the visual representation with bidirectional state space... | Lianghui Zhu; Bencheng Liao; Qian Zhang; Xinlong Wang; Wenyu Liu; Xinggang Wang |
Online Conformal Prediction with Decaying Step Sizes | We introduce a method for online conformal prediction with decaying step sizes. | Anastasios Nikolas Angelopoulos; Rina Barber; Stephen Bates |
Monitoring AI-Modified Content at Scale: A Case Study on The Impact of ChatGPT on AI Conference Peer Reviews | We present an approach for estimating the fraction of text in a large corpus which is likely to be substantially modified or produced by a large language model (LLM). | Weixin Liang; Zachary Izzo; Yaohui Zhang; Haley Lepp; Hancheng Cao; Xuandong Zhao; Lingjiao Chen; Haotian Ye; Sheng Liu; Zhi Huang; Daniel McFarland; James Y. Zou |
Graph Positional and Structural Encoder | Here, we present the Graph Positional and Structural Encoder (GPSE), the first-ever graph encoder designed to capture rich PSE representations for augmenting any GNN. | Semih Cantürk; Renming Liu; Olivier Lapointe-Gagné; Vincent Létourneau; Guy Wolf; Dominique Beaini; Ladislav Rampášek |
Model Alignment As Prospect Theoretic Optimization | Using a Kahneman-Tversky model of human utility, we propose a HALO that directly maximizes the utility of generations instead of maximizing the log-likelihood of preferences, as current methods do. | Kawin Ethayarajh; Winnie Xu; Niklas Muennighoff; Dan Jurafsky; Douwe Kiela |
Neural Operators with Localized Integral and Differential Kernels | In this work, we present a principled approach to operator learning that can capture local features under two frameworks by learning differential operators and integral operators with locally supported kernels. | Miguel Liu-Schiaffini; Julius Berner; Boris Bonev; Thorsten Kurth; Kamyar Azizzadenesheli; Anima Anandkumar |
Position: The No Free Lunch Theorem, Kolmogorov Complexity, and The Role of Inductive Biases in Machine Learning | Whereas no free lunch theorems seemingly indicate that individual problems require specialized learners, we explain how tasks that often require human intervention such as picking an appropriately sized model when labeled data is scarce or plentiful can be automated into a single learning algorithm. | Micah Goldblum; Marc Anton Finzi; Keefer Rowan; Andrew Gordon Wilson |
Q-Probe: A Lightweight Approach to Reward Maximization for Language Models | We present an approach called Q-probing to adapt a pre-trained language model to maximize a task-specific reward function. | Kenneth Li; Samy Jelassi; Hugh Zhang; Sham M. Kakade; Martin Wattenberg; David Brandfonbrener |
Test-Time Model Adaptation with Only Forward Passes | In light of this, existing methods are often infeasible since they heavily depend on computation-intensive backpropagation for model updating that may be not supported. To address this, we propose a test-time Forward-Optimization Adaptation (FOA) method. | Shuaicheng Niu; Chunyan Miao; Guohao Chen; Pengcheng Wu; Peilin Zhao |
Solving Poisson Equations Using Neural Walk-on-Spheres | We propose Neural Walk-on-Spheres (NWoS), a novel neural PDE solver for the efficient solution of high-dimensional Poisson equations. | Hong Chul Nam; Julius Berner; Anima Anandkumar |
StrokeNUWA—Tokenizing Strokes for Vector Graphic Synthesis | Thus, we introduce StrokeNUWA, a pioneering work exploring a better visual representation stroke tokens on vector graphics, which is inherently visual semantics rich, naturally compatible with LLMs, and highly compressed. | Zecheng Tang; Chenfei Wu; Zekai Zhang; Minheng Ni; Shengming Yin; Yu Liu; Zhengyuan Yang; Lijuan Wang; Zicheng Liu; Juntao Li; Nan Duan |
NExT: Teaching Large Language Models to Reason About Code Execution | However, large language models (LLMs) of code are typically trained on the surface textual form of programs, thus may lack a semantic understanding of how programs execute at run-time. To address this issue, we propose NExT, a method to teach LLMs to inspect the execution traces of programs (variable states of executed... | Ansong Ni; Miltiadis Allamanis; Arman Cohan; Yinlin Deng; Kensen Shi; Charles Sutton; Pengcheng Yin |
Interpretability Illusions in The Generalization of Simplified Models | Here, we illustrate an important caveat to this assumption: even if the simplified representations can accurately approximate the full model on the training set, they may fail to accurately capture the model’s behavior out of distribution. We illustrate this by training Transformer models on controlled datasets with sy... | Dan Friedman; Andrew Kyle Lampinen; Lucas Dixon; Danqi Chen; Asma Ghandeharioun |
Language Models As Science Tutors | To address this, we introduce TutorEval and TutorChat.Our datasets build on open-source materials, and we release our models, data, and evaluations publicly. | Alexis Chevalier; Jiayi Geng; Alexander Wettig; Howard Chen; Sebastian Mizera; Toni Annala; Max Aragon; Arturo Rodriguez Fanlo; Simon Frieder; Simon Machado; Akshara Prabhakar; Ellie Thieu; Jiachen T. Wang; Zirui Wang; Xindi Wu; Mengzhou Xia; Wenhan Xia; Jiatong Yu; Junjie Zhu; Zhiyong Ren; Sanjeev Arora; Danqi Chen |
The WMDP Benchmark: Measuring and Reducing Malicious Use with Unlearning | To guide progress on unlearning, we develop RMU, a state-of-the-art unlearning method based on controlling model representations.We release our benchmark and code publicly at https://wmdp.ai. | Nathaniel Li; Alexander Pan; Anjali Gopal; Summer Yue; Daniel Berrios; Alice Gatti; Justin D. Li; Ann-Kathrin Dombrowski; Shashwat Goel; Gabriel Mukobi; Nathan Helm-Burger; Rassin Lababidi; Lennart Justen; Andrew Bo Liu; Michael Chen; Isabelle Barrass; Oliver Zhang; Xiaoyuan Zhu; Rishub Tamirisa; Bhrugu Bharathi; Ariel... |
How Learning By Reconstruction Produces Uninformative Features For Perception | Despite interpretability benefit of reconstruction and generation, we identify a misalignment between learning to reconstruct, and learning for perception. | Randall Balestriero; Yann LeCun |
Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformers | We present the Hourglass Diffusion Transformer (HDiT), an image-generative model that exhibits linear scaling with pixel count, supporting training at high resolution (e.g. $1024 \times 1024$) directly in pixel-space. | Katherine Crowson; Stefan Andreas Baumann; Alex Birch; Tanishq Mathew Abraham; Daniel Z Kaplan; Enrico Shippole |
RLVF: Learning from Verbal Feedback Without Overgeneralization | We propose a new method Contextualized Critiques with Constrained Preference Optimization (C3PO) to learn from high-level verbal feedback while reducing overgeneralization compared to current work. | Moritz Pascal Stephan; Alexander Khazatsky; Eric Mitchell; Annie S Chen; Sheryl Hsu; Archit Sharma; Chelsea Finn |
Distinguishing The Knowable from The Unknowable with Language Models | We study the feasibility of identifying *epistemic* uncertainty (reflecting a lack of knowledge), as opposed to *aleatoric* uncertainty (reflecting entropy in the underlying distribution), in the outputs of large language models (LLMs) over free-form text. | Gustaf Ahdritz; Tian Qin; Nikhil Vyas; Boaz Barak; Benjamin L. Edelman |
MathScale: Scaling Instruction Tuning for Mathematical Reasoning | We propose MathScale, a simple and scalable method to create high-quality mathematical reasoning data using frontier LLMs (e.g., GPT-3.5).As a result, we create a mathematical reasoning dataset (MathScaleQA) containing two million math question-answer pairs. | Zhengyang Tang; Xingxing Zhang; Benyou Wang; Furu Wei |
InstructSpeech: Following Speech Editing Instructions Via Large Language Models | In this work, we construct triplet paired data (instruction, input speech, output speech) to alleviate data scarcity and train a multi-task large language model named InstructSpeech. | Rongjie Huang; Ruofan Hu; Yongqi Wang; Zehan Wang; Xize Cheng; Ziyue Jiang; Zhenhui Ye; Dongchao Yang; Luping Liu; Peng Gao; Zhou Zhao |
Position: Video As The New Language for Real-World Decision Making | Yet video data captures important information about the physical world that is difficult to express in language. To address this gap, we discuss an under-appreciated opportunity to extend video generation to solve tasks in the real world. | Sherry Yang; Jacob C Walker; Jack Parker-Holder; Yilun Du; Jake Bruce; Andre Barreto; Pieter Abbeel; Dale Schuurmans |
Retrieval-Augmented Score Distillation for Text-to-3D Generation | To this end, we introduce novel framework for retrieval-based quality enhancement in text-to-3D generation. | Junyoung Seo; Susung Hong; Wooseok Jang; Inès Hyeonsu Kim; Min-Seop Kwak; Doyup Lee; Seungryong Kim |
PIVOT: Iterative Visual Prompting Elicits Actionable Knowledge for VLMs | In this paper, we propose a novel visual prompting approach for VLMs that we call Prompting with Iterative Visual Optimization (PIVOT), which casts tasks as iterative visual question answering. | Soroush Nasiriany; Fei Xia; Wenhao Yu; Ted Xiao; Jacky Liang; Ishita Dasgupta; Annie Xie; Danny Driess; Ayzaan Wahid; Zhuo Xu; Quan Vuong; Tingnan Zhang; Tsang-Wei Edward Lee; Kuang-Huei Lee; Peng Xu; Sean Kirmani; Yuke Zhu; Andy Zeng; Karol Hausman; Nicolas Heess; Chelsea Finn; Sergey Levine; brian ichter |
Connect Later: Improving Fine-tuning for Robustness with Targeted Augmentations | To better leverage pretraining for distribution shifts, we propose the Connect Later framework, which fine-tunes the model with targeted augmentations designed with knowledge of the shift. | Helen Qu; Sang Michael Xie |
Modeling Caption Diversity in Contrastive Vision-Language Pretraining | In this work, we introduce Llip, Latent Language Image Pretraining, which models the diversity of captions that could match an image. | Samuel Lavoie; Polina Kirichenko; Mark Ibrahim; Mido Assran; Andrew Gordon Wilson; Aaron Courville; Nicolas Ballas |
Position: Levels of AGI for Operationalizing Progress on The Path to AGI | We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. | Meredith Ringel Morris; Jascha Sohl-Dickstein; Noah Fiedel; Tris Warkentin; Allan Dafoe; Aleksandra Faust; Clement Farabet; Shane Legg |
UniAudio: Towards Universal Audio Generation with Large Language Models | As recent research on large language models (LLMs) has demonstrated their strong ability to handle multiple tasks, this work presents UniAudio, an LLM-based audio generation model that supports a wide range of audio generation tasks. | Dongchao Yang; Jinchuan Tian; Xu Tan; Rongjie Huang; Songxiang Liu; Haohan Guo; Xuankai Chang; Jiatong Shi; sheng zhao; Jiang Bian; Zhou Zhao; Xixin Wu; Helen M. Meng |
Unified Generation, Reconstruction, and Representation: Generalized Diffusion with Adaptive Latent Encoding-Decoding | We introduce *Generalized* ***E****ncoding*-***D****ecoding ****D****iffusion ****P****robabilistic ****M****odels* (EDDPMs) which integrate the core capabilities for broad applicability and enhanced performance. | Guangyi Liu; Yu Wang; Zeyu Feng; Qiyu Wu; Liping Tang; Yuan Gao; Zhen Li; Shuguang Cui; Julian McAuley; Zichao Yang; Eric P. Xing; Zhiting Hu |
Revisiting The Role of Language Priors in Vision-Language Models | We study $\textit{generative VLMs}$ that are trained for next-word generation given an image. | Zhiqiu Lin; Xinyue Chen; Deepak Pathak; Pengchuan Zhang; Deva Ramanan |
MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases | This paper addresses the growing need for efficient large language models (LLMs) on mobile devices, driven by increasing cloud costs and latency concerns. | Zechun Liu; Changsheng Zhao; Forrest Iandola; Chen Lai; Yuandong Tian; Igor Fedorov; Yunyang Xiong; Ernie Chang; Yangyang Shi; Raghuraman Krishnamoorthi; Liangzhen Lai; Vikas Chandra |
InstructRetro: Instruction Tuning Post Retrieval-Augmented Pretraining | In this work, we introduce Retro 48B, the largest LLM pretrained with retrieval. | Boxin Wang; Wei Ping; Lawrence McAfee; Peng Xu; Bo Li; Mohammad Shoeybi; Bryan Catanzaro |
Characterizing Large Language Model Geometry Helps Solve Toxicity Detection and Generation | In this work, we propose to shed the light on LLMs inner mechanisms through the lens of geometry. | Randall Balestriero; Romain Cosentino; Sarath Shekkizhar |
Compositional Text-to-Image Generation with Dense Blob Representations | To leverage the compositionality of large language models (LLMs), we introduce a new in-context learning approach to generate blob representations from text prompts. | Weili Nie; Sifei Liu; Morteza Mardani; Chao Liu; Benjamin Eckart; Arash Vahdat |
Position Paper: On The Societal Impact of Open Foundation Models | Foundation models are powerful technologies: how they are released publicly directly shapes their societal impact. In this position paper, we focus on *open* foundation models, defined here as those with broadly available model weights (e.g., Llama 3, Stable Diffusion XL). | Sayash Kapoor; Rishi Bommasani; Kevin Klyman; Shayne Longpre; Ashwin Ramaswami; Peter Cihon; Aspen K Hopkins; Kevin Bankston; Stella Biderman; Miranda Bogen; Rumman Chowdhury; Alex Engler; Peter Henderson; Yacine Jernite; Seth Lazar; Stefano Maffulli; Alondra Nelson; Joelle Pineau; Aviya Skowron; Dawn Song; Victor Stor... |
Scalable Pre-training of Large Autoregressive Image Models | This paper introduces AIM, a collection of vision models pre-trained with an autoregressive objective. | Alaaeldin El-Nouby; Michal Klein; Shuangfei Zhai; Miguel Ángel Bautista; Vaishaal Shankar; Alexander T Toshev; Joshua M. Susskind; Armand Joulin |
Unlocking The Power of Spatial and Temporal Information in Medical Multimodal Pre-training | In this paper, we introduce the Med-ST framework for fine-grained spatial and temporal modeling to exploit information from multiple spatial views of chest radiographs and temporal historical records. | Jinxia Yang; Bing Su; Xin Zhao; Ji-Rong Wen |
QuRating: Selecting High-Quality Data for Training Language Models | We introduce QuRating, a method for selecting pre-training data that can capture human intuitions about data quality. | Alexander Wettig; Aatmik Gupta; Saumya Malik; Danqi Chen |
MC-GTA: Metric-Constrained Model-Based Clustering Using Goodness-of-fit Tests with Autocorrelations | The model-based variations of these clustering algorithms (e.g. TICC and STICC) achieve SOTA performance, yet suffer from computational instability and complexity by using a metric-constrained Expectation-Maximization procedure. In order to address these two problems, we propose a novel clustering algorithm, MC-GTA (**... | Zhangyu Wang; Gengchen Mai; Krzysztof Janowicz; Ni Lao |
Assessing The Brittleness of Safety Alignment Via Pruning and Low-Rank Modifications | We develop methods to identify critical regions that are vital for safety guardrails, and that are disentangled from utility-relevant regions at both the neuron and rank levels. | Boyi Wei; Kaixuan Huang; Yangsibo Huang; Tinghao Xie; Xiangyu Qi; Mengzhou Xia; Prateek Mittal; Mengdi Wang; Peter Henderson |
In-Context Unlearning: Language Models As Few-Shot Unlearners | In this work, we propose a new class of unlearning methods for LLMs called “In-Context Unlearning.” | Martin Pawelczyk; Seth Neel; Himabindu Lakkaraju |
Position Paper: Scaling Simulation Is Neither Necessary Nor Sufficient for In-the-Wild Robot Manipulation | In this paper, we develop a structured critique of robotic simulations for real-world manipulation, by arguing that scaling simulators is neither necessary nor sufficient for making progress in general-purpose real-world robotic manipulation agents that are compliant with human preferences. | Homanga Bharadhwaj |
In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steering | However, in-context learning has seen limited effectiveness in many settings, is difficult to quantitatively control and takes up context window space. To overcome these limitations, we propose an alternative approach that recasts in-context learning as in-context vectors (ICV). | Sheng Liu; Haotian Ye; Lei Xing; James Y. Zou |
Neural Networks Learn Statistics of Increasing Complexity | The _distributional simplicity bias_ (DSB) posits that neural networks learn low-order moments of the data distribution first, before moving on to higher-order correlations. In this work, we present compelling new evidence for the DSB by showing that networks automatically learn to perform well on maximum-entropy distr... | Nora Belrose; Quintin Pope; Lucia Quirke; Alex Troy Mallen; Xiaoli Fern |
Offline Training of Language Model Agents with Functions As Learnable Weights | To facilitate the development of LLM agents, we present a novel paradigm of training LLM agents without modifying the LLM weights, which is particularly useful when the LLMs are difficult or inaccessible for modifications. | Shaokun Zhang; Jieyu Zhang; Jiale Liu; Linxin Song; Chi Wang; Ranjay Krishna; Qingyun Wu |
Chain of Code: Reasoning with A Language Model-Augmented Code Emulator | In this work, we propose Chain of Code (CoC), a simple yet surprisingly effective extension that improves LM code-driven reasoning. | Chengshu Li; Jacky Liang; Andy Zeng; Xinyun Chen; Karol Hausman; Dorsa Sadigh; Sergey Levine; Li Fei-Fei; Fei Xia; brian ichter |
Prodigy: An Expeditiously Adaptive Parameter-Free Learner | We propose Prodigy, an algorithm that provably estimates the distance to the solution $D$, which is needed to set the learning rate optimally. | Konstantin Mishchenko; Aaron Defazio |
Projecting Molecules Into Synthesizable Chemical Spaces | In this work, we introduce a novel framework that is capable of generating new chemical structures while ensuring synthetic accessibility. | Shitong Luo; Wenhao Gao; Zuofan Wu; Jian Peng; Connor W. Coley; Jianzhu Ma |
Learning Divergence Fields for Shift-Robust Graph Representations | In this work, we propose a geometric diffusion model with learnable divergence fields for the challenging generalization problem with interdependent data. | Qitian Wu; Fan Nie; Chenxiao Yang; Junchi Yan |
Graph Neural Networks Use Graphs When They Shouldn’t | While GNNs have the ability to ignore the graph-structure in such cases, it is not clear that they will. In this work, we show that GNNs actually tend to overfit the given graph-structure in the sense that they use it even when a better solution can be obtained by ignoring it. | Maya Bechler-Speicher; Ido Amos; Ran Gilad-Bachrach; Amir Globerson |
Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text | However, we find that a score based on contrasting two closely related language models is highly accurate at separating human-generated and machine-generated text. Based on this mechanism, we propose a novel LLM detector that only requires simple calculations using a pair of pre-trained LLMs. | Abhimanyu Hans; Avi Schwarzschild; Valeriia Cherepanova; Hamid Kazemi; Aniruddha Saha; Micah Goldblum; Jonas Geiping; Tom Goldstein |
Explorations of Self-Repair in Language Models | We highlight two different mechanisms that contribute to self-repair, including changes in the final LayerNorm scaling factor and sparse sets of neurons implementing Anti-Erasure. | Cody Rushing; Neel Nanda |
Position: A Roadmap to Pluralistic Alignment | However, aligning models to serve *pluralistic* human values remains an open research question. In this piece, we propose a roadmap to pluralistic alignment, specifically using large language models as a test bed. | Taylor Sorensen; Jared Moore; Jillian Fisher; Mitchell L Gordon; Niloofar Mireshghallah; Christopher Michael Rytting; Andre Ye; Liwei Jiang; Ximing Lu; Nouha Dziri; Tim Althoff; Yejin Choi |
End of preview. Expand in Data Studio
No dataset card yet
- Downloads last month
- 25