Search Results for author:
Found 75 papers, 59 papers with code
FlowReasoner: Reinforcing Query-Level Meta-Agents
This paper proposes a query-level meta-agent named FlowReasoner to automate the design of query-level multi-agent systems, i. e., one system per user query.
Understanding R1-Zero-Like Training: A Critical Perspective
DeepSeek-R1-Zero has shown that reinforcement learning (RL) at scale can directly enhance the reasoning capabilities of LLMs without supervised fine-tuning.
PipeOffload: Improving Scalability of Pipeline Parallelism with Memory Optimization
In this paper, we focus on addressing this challenge by leveraging the under-explored memory offload strategy in PP.
Improving Your Model Ranking on Chatbot Arena by Vote Rigging
We first introduce a straightforward target-only rigging strategy that focuses on new battles involving $m_{t}$, identifying it via watermarking or a binary classifier, and exclusively voting for $m_{t}$ wins.
StereoGen: High-quality Stereo Image Generation from a Single Image
This pipeline utilizes arbitrary single images as left images and pseudo disparities generated by a monocular depth estimation model to synthesize high-quality corresponding right images.
FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation
In this paper, we propose a novel global-aware scene flow estimation network with global motion propagation, named FlowMamba.
Stochastic Taylor Derivative Estimator: Efficient amortization for arbitrary differential operators
Separately, the exponential scaling in $k$ for univariate functions ($d=1$) was addressed with high-order auto-differentiation (AD).
Diagonalization without Diagonalization: A Direct Optimization Approach for Solid-State Density Functional Theory
We present a novel approach to address the challenges of variable occupation numbers in direct optimization of density functional theory (DFT).
Sample-Efficient Alignment for LLMs
The results demonstrate that SEA achieves highly sample-efficient alignment with oracle's preferences, outperforming recent active exploration methods for LLMs.
Scaling up Masked Diffusion Models on Text
Masked diffusion models (MDMs) have shown promise in language modeling, yet their scalability and effectiveness in core language tasks, such as text generation and language understanding, remain underexplored.
SimLayerKV: A Simple Framework for Layer-Level KV Cache Reduction
To mitigate this issue, we present SimLayerKV, a simple yet effective method that reduces inter-layer KV cache redundancies by selectively dropping cache in identified lazy layers.
Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts
With the rapid progress of diffusion-based content generation, significant efforts are being made to unlearn harmful or copyrighted concepts from pretrained diffusion models (DMs) to prevent potential model misuse.
When Attention Sink Emerges in Language Models: An Empirical View
In this work, we first demonstrate that attention sinks exist universally in LMs with various inputs, even in small models.
Denial-of-Service Poisoning Attacks against Large Language Models
To overcome this limitation, we propose poisoning-based DoS (P-DoS) attacks for LLMs, demonstrating that injecting a single poisoned sample designed for DoS purposes can break the output length limit.
A Closer Look at Machine Unlearning for Large Language Models
Specifically, the behavior that untargeted unlearning attempts to approximate is unpredictable and may involve hallucinations, and existing regularization is insufficient for targeted unlearning.
Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates
Achieving high win rates on these benchmarks can significantly boost the promotional impact of newly released language models.
Scaling Laws with Vocabulary: Larger Models Deserve Larger Vocabularies
We investigate how vocabulary size impacts LLM scaling laws by training models ranging from 33M to 3B parameters on up to 500B characters with various vocabulary configurations.
RegMix: Data Mixture as Regression for Language Model Pre-training
RegMix trains many small models on diverse data mixtures, uses regression to predict performance of unseen mixtures, and applies the best predicted mixture to train a large-scale model with orders of magnitude more compute.
Bootstrapping Language Models with DPO Implicit Rewards
In this work, we make a novel observation that this implicit reward model can by itself be used in a bootstrapping fashion to further align the LLM.
Chain of Preference Optimization: Improving Chain-of-Thought Reasoning in LLMs
The recent development of chain-of-thought (CoT) decoding has enabled large language models (LLMs) to generate explicit logical reasoning paths for complex problem-solving.
Improved Few-Shot Jailbreaking Can Circumvent Aligned Language Models and Their Defenses
In addition, we conduct comprehensive and elaborate (e. g., making sure to use correct system prompts) evaluations against other aligned LLMs and advanced defenses, where our method consistently achieves nearly 100% ASRs.
Improved Techniques for Optimization-Based Jailbreaking on Large Language Models
Many red-teaming efforts aim to jailbreak LLMs, where among these efforts, the Greedy Coordinate Gradient (GCG) attack's success has led to a growing interest in the study of optimization-based jailbreaking techniques.
Pipeline Parallelism with Controllable Memory
To address this, we introduce a family of memory efficient building blocks with controllable activation memory, which can reduce the peak activation memory to 1/2 of 1F1B without sacrificing efficiency, and even to 1/3 with comparable throughput.
Sailor: Open Language Models for South-East Asia
We present Sailor, a family of open language models ranging from 0. 5B to 7B parameters, tailored for South-East Asian (SEA) languages.
Beyond Memorization: The Challenge of Random Memory Access in Language Models
Through carefully-designed synthetic tasks, covering the scenarios of full recitation, selective recitation and grounded question answering, we reveal that LMs manage to sequentially access their memory while encountering challenges in randomly accessing memorized content.
Graph Diffusion Policy Optimization
Recent research has made significant progress in optimizing diffusion models for downstream objectives, which is an important pursuit in fields such as graph generation for drug design.
Purifying Large Language Models by Ensembling a Small Language Model
The emerging success of large language models (LLMs) heavily relies on collecting abundant training data from external (untrusted) sources.
Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast
A multimodal large language model (MLLM) agent can receive instructions, capture images, retrieve histories from memory, and decide which tools to use.
Test-Time Backdoor Attacks on Multimodal Large Language Models
Backdoor attacks are commonly executed by contaminating training data, such that a trigger can activate predetermined harmful effects during the test phase.
Locality Sensitive Sparse Encoding for Learning World Models Online
Unfortunately, NN-based models need re-training on all accumulated data at every interaction step to achieve FTL, which is computationally expensive for lifelong agents.
Benchmarking Large Multimodal Models against Common Corruptions
This technical report aims to fill a deficiency in the assessment of large multimodal models (LMMs) by specifically examining the self-consistency of their outputs when subjected to common corruptions.
Automatic Functional Differentiation in JAX
We present a set of primitive operators that serve as foundational building blocks for constructing several key types of functionals.
Zero Bubble Pipeline Parallelism
Pipeline parallelism is one of the key components for large-scale distributed training, yet its efficiency suffers from pipeline bubbles which were deemed inevitable.
Instant3D: Instant Text-to-3D Generation
We achieve this remarkable speed by devising a new network that directly constructs a 3D triplane from a text prompt.
Finetuning Text-to-Image Diffusion Models for Fairness
The rapid adoption of text-to-image diffusion models in society underscores an urgent need to address their biases.
Intriguing Properties of Data Attribution on Diffusion Models
Data attribution seeks to trace model outputs back to training data.
On Memorization in Diffusion Models
Looking into this, we first observe that memorization behaviors tend to occur on smaller-sized datasets, which motivates our definition of effective model memorization (EMM), a metric measuring the maximum size of training data at which a learned diffusion model approximates its theoretical optimum.
Cleanba: A Reproducible and Efficient Distributed Reinforcement Learning Platform
Distributed Deep Reinforcement Learning (DRL) aims to leverage more computational resources to train autonomous agents with less training time.
LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition
This paper investigates LoRA composability for cross-task generalization and introduces LoraHub, a simple framework devised for the purposive assembly of LoRA modules trained on diverse given tasks, with the objective of achieving adaptable performance on unseen tasks.
NU-MCC: Multiview Compressive Coding with Neighborhood Decoder and Repulsive UDF
Second, our Repulsive UDF is a novel alternative to the occupancy field used in MCC, significantly improving the quality of 3D object reconstruction.
On Evaluating Adversarial Robustness of Large Vision-Language Models
Large vision-language models (VLMs) such as GPT-4 have achieved unprecedented performance in response generation, especially with visual inputs, enabling more creative and adaptable interaction than large language models such as ChatGPT.
Nonparametric Generative Modeling with Conditional Sliced-Wasserstein Flows
Sliced-Wasserstein Flow (SWF) is a promising approach to nonparametric generative modeling but has not been widely adopted due to its suboptimal generative quality and lack of conditional modeling capabilities.
From Zero to Hero: Examining the Power of Symbolic Tasks in Instruction Tuning
Empirical results on various benchmarks validate that the integration of SQL execution leads to significant improvements in zero-shot scenarios, particularly in table reasoning.
Exploring Incompatible Knowledge Transfer in Few-shot Image Generation
To this end, we propose knowledge truncation to mitigate this issue in FSIG, which is a complementary operation to knowledge preservation and is implemented by a lightweight pruning-based method.
A Recipe for Watermarking Diffusion Models
Diffusion models (DMs) have demonstrated advantageous potential on generative tasks.
D4FT: A Deep Learning Approach to Kohn-Sham Density Functional Theory
Kohn-Sham Density Functional Theory (KS-DFT) has been traditionally solved by the Self-Consistent Field (SCF) method.
On Calibrating Diffusion Probabilistic Models
In this work, we observe that the stochastic reverse process of data scores is a martingale, from which concentration bounds and the optional stopping theorem for data scores can be derived.
Bag of Tricks for Training Data Extraction from Language Models
With the advance of language models, privacy protection is receiving more attention.
Better Diffusion Models Further Improve Adversarial Training
Under the $\ell_\infty$-norm threat model with $\epsilon=8/255$, our models achieve $70. 69\%$ and $42. 67\%$ robust accuracy on CIFAR-10 and CIFAR-100, respectively, i. e. improving upon previous state-of-the-art models by $+4. 58\%$ and $+8. 03\%$.
BAFFLE: A Baseline of Backpropagation-Free Federated Learning
BAFFLE is 1) memory-efficient and easily fits uploading bandwidth; 2) compatible with inference-only hardware optimization and model quantization or pruning; and 3) well-suited to trusted execution environments, because the clients in BAFFLE only execute forward propagation and return a set of scalars to the server.
IHNet: Iterative Hierarchical Network Guided by High-Resolution Estimated Information for Scene Flow Estimation
This approach circulates high-resolution estimated information (scene flow and feature) from the preceding iteration back to the low-resolution layer of the current iteration.
Mutual Information Regularized Offline Reinforcement Learning
In this work, we propose a novel MISA framework to approach offline RL from the perspective of Mutual Information between States and Actions in the dataset by directly constraining the policy improvement direction.
Optical Neural Ordinary Differential Equations
Increasing the layer number of on-chip photonic neural networks (PNNs) is essential to improve its model performance.
EnvPool: A Highly Parallel Reinforcement Learning Environment Execution Engine
EnvPool is open-sourced at https://github. com/sail-sg/envpool.
$O(N^2)$ Universal Antisymmetry in Fermionic Neural Networks
In this paper, we substitute the Slater determinant with a pairwise antisymmetry construction, which is easy to implement and can reduce the computational cost to $O(N^2)$.
CINO: A Chinese Minority Pre-trained Language Model
It covers Standard Chinese, Yue Chinese, and six other ethnic minority languages.
Robustness and Accuracy Could Be Reconcilable by (Proper) Definition
The trade-off between robustness and accuracy has been widely studied in the adversarial literature.
Causal Attention for Interpretable and Generalizable Graph Classification
To endow the classifier with better interpretation and generalization, we propose the Causal Attention Learning (CAL) strategy, which discovers the causal patterns and mitigates the confounding effect of shortcuts.
How Should Pre-Trained Language Models Be Fine-Tuned Towards Adversarial Robustness?
The fine-tuning of pre-trained language models has a great success in many NLP fields.
LSTM-RPA: A Simple but Effective Long Sequence Prediction Algorithm for Music Popularity Prediction
The traditional trend prediction models can better predict the short trend than the long trend.
Outage Constrained Robust Secure Beamforming in Cognitive Satellite-Aerial Networks
This paper proposes a robust beamforming scheme to enhance the physical layer security (PLS) of multicast transmission in a cognitive satellite and aerial network (CSAN) operating in the millimeter wave frequency band.
Online Fast Adaptation and Knowledge Accumulation (OSAKA): a New Approach to Continual Learning
The main challenge is that the agent must not forget previous tasks and also adapt to novel tasks in the stream.
Continual Learning from the Perspective of Compression
We compare these approaches in terms of both compression and forgetting and empirically study the reasons that limit the performance of continual learning methods based on variational posterior approximation.
Online Fast Adaptation and Knowledge Accumulation: a New Approach to Continual Learning
We propose Continual-MAML, an online extension of the popular MAML algorithm as a strong baseline for this scenario.
Online Continual Learning with Maximal Interfered Retrieval
Methods based on replay, either generative or from a stored memory, have been shown to be effective approaches for continual learning, matching or exceeding the state of the art in a number of standard benchmarks.
Online Continual Learning with Maximally Interfered Retrieval
Methods based on replay, either generative or from a stored memory, have been shown to be effective approaches for continual learning, matching or exceeding the state of the art in a number of standard benchmarks.
Conditional Computation for Continual Learning
In this study, we analyze parameter sharing under the conditional computation framework where the parameters of a neural network are conditioned on each input example.
Gradient based sample selection for online continual learning
To prevent forgetting, a replay buffer is usually employed to store the previous data for the purpose of rehearsal.
On the Spectral Bias of Neural Networks
Neural networks are known to be a class of highly expressive functions able to fit even random input-output mappings with $100\%$ accuracy.
A Machine Learning Framework for Resource Allocation Assisted by Cloud Computing
Moreover, optimal or near-optimal solutions of historical scenarios can be searched offline and stored in advance.
Softmax GAN
In the generator training phase, the target is to assign equal probability to all data points in the batch, each with probability $\frac{1}{M+N}$.
MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems
This paper describes both the API design and the system implementation of MXNet, and explains how embedding of both symbolic expression and tensor operation is handled in a unified fashion.
Correntropy Induced L2 Graph for Robust Subspace Clustering
In this paper, we study the robust subspace clustering problem, which aims to cluster the given possibly noisy data points into their underlying subspaces.
Purine: A bi-graph based deep learning framework
In this paper, we introduce a novel deep learning framework, termed Purine.
Network In Network
With enhanced local modeling via the micro network, we are able to utilize global average pooling over feature maps in the classification layer, which is easier to interpret and less prone to overfitting than traditional fully connected layers.
Ranked #4 on
Face Identification
on DroneSURF
