arXiv subjects

Zihan Zhang

Publications and source records attributed to Zihan Zhang.

At least 163 records · Page 9Linked to original sources

arXiv2024

MoRA: High-Rank Updating for Parameter-Efficient Fine-Tuning

Low-rank adaptation is a popular parameter-efficient fine-tuning method for large language models. In this paper, we analyze the impact of low-rank updating, as implemented in LoRA. Our findings suggest that the low-rank updating mechanism may limit the ability of LLMs to effectively learn and memorize new knowledge. Inspired by this observation, we propose a new method called MoRA, which employs a square matrix to achieve high-rank updating while maintaining the same number of trainable parameters. To achieve it, we introduce the corresponding non-parameter operators to reduce the input dimension and increase the output dimension for the square matrix. Furthermore, these operators ensure that the weight can be merged back into LLMs, which makes our method can be deployed like LoRA. We perform a comprehensive evaluation of our method across five tasks: instruction tuning, mathematical reasoning, continual pretraining, memory and pretraining. Our method outperforms LoRA on memory-intensive tasks and achieves comparable performance on other tasks.

cs.CL↗

arXiv2024

Using a Fast Adaptive Function Approximator to calculate Protein-Filament Binding Kinetics

The cytoskeleton, consisting of biopolymer filaments, molecular motors, and passive crosslinking proteins, provides the internal structure of cells that facilitate movement, growth, and cell division. Understanding the microscopic motor-filament kinetics and dynamics is essential for comprehending macroscopic behaviors of reconstituted cytoskeletal assemblies, such as self-organized flow and active stress. In this study, we employ an adaptive fast Chebyshev approximator based on tree search and parallel computing to accurately recover the equilibrium distribution of crosslinking proteins. Therefore, it satisfies detailed balance in binding through kinetic Monte Carlo sampling while maintaining cost-effectiveness. Additionally, we offer expandable features, including segregating the simulation process via pre-building and allowing the free-loading of different closed-form formulations of the motor's potential energy. Overall, this research contributes to computational advancement in function approximation and has the potential to better describe the evolution of cytoskeletal active matter.

cond-mat.soft↗

arXiv2024

Prediction of Room-Temperature Superconductivity in Quasi-atomic H2-Type Hydrides at High Pressure

Achieving superconductivity at room temperature (RT) is a holy grail in physics. Recent discoveries on high-Tc superconductivity in binary hydrides H3S and LaH10 at high pressure have directed the search for RT superconductors to compress hydrides with conventional electron-phonon mechanisms. Here, we predict an exceptional family of superhydrides under high pressures, MH12 (M = Mg, Sc, Zr, Hf, Lu), all exhibiting RT superconductivity with calculated Tcs ranging from 313 to 398 K. In contrast to H3S and LaH10, the hydrogen sublattice in MH12 is arranged as quasi-atomic H2 units. This unique configuration is closely associated with high Tc, attributed to the high electronic density of states derived from H2 antibonding states at the Fermi level and the strong electron-phonon coupling related to the bending vibration of H2 and H-M-H. Notably, MgH12 and ScH12 remain dynamically stable even at pressure below 100 GPa. Our findings offer crucial insights into achieving RT superconductivity and pave the way for innovative directions in experimental research.

cond-mat.supr-con↗

arXiv2024

Tunable exciton valley-pseudospin orders in moiré Bose-Hubbard model

Spin and charge are the two most important degrees of freedom of electrons. Their interplay lies at the heart of numerous strongly correlated phenomena including Hubbard model physics and high temperature superconductivity. Such interplay for bosons, on the other hand, is largely unexplored in condensed matter systems. Here we demonstrate a unique realization of the spin-1/2 Bose-Hubbard model through excitons in a semiconducting moiré superlattice. We find evidence of a transient in-plane ferromagnetic (FM-$xy$) order of exciton spin - here valley pseudospin - around exciton filling $ν_{ex}$ = 1, which transitions into a FM-$z$ order both with increasing exciton filling and a small magnetic field of 10 mT. The phase diagram is different from the fermion case and is qualitatively captured by a simple phenomenological model, highlighting the unique consequence of Bose-Einstein statistics. Our study paves the way for engineering exotic phases of matter from spinor bosons, as well as for unconventional devices in optics and quantum information science.

cond-mat.mes-hall↗

arXiv2024

Classification with Deep Neural Networks and Logistic Loss

Deep neural networks (DNNs) trained with the logistic loss (i.e., the cross entropy loss) have made impressive advancements in various binary classification tasks. However, generalization analysis for binary classification with DNNs and logistic loss remains scarce. The unboundedness of the target function for the logistic loss is the main obstacle to deriving satisfactory generalization bounds. In this paper, we aim to fill this gap by establishing a novel and elegant oracle-type inequality, which enables us to deal with the boundedness restriction of the target function, and using it to derive sharp convergence rates for fully connected ReLU DNN classifiers trained with logistic loss. In particular, we obtain optimal convergence rates (up to log factors) only requiring the Hölder smoothness of the conditional class probability $η$ of data. Moreover, we consider a compositional assumption that requires $η$ to be the composition of several vector-valued functions of which each component function is either a maximum value function or a Hölder smooth function only depending on a small number of its input variables. Under this assumption, we derive optimal convergence rates (up to log factors) which are independent of the input dimension of data. This result explains why DNN classifiers can perform well in practical high-dimensional classification problems. Besides the novel oracle-type inequality, the sharp convergence rates given in our paper also owe to a tight error bound for approximating the natural logarithm function near zero (where it is unbounded) by ReLU DNNs. In addition, we justify our claims for the optimality of rates by proving corresponding minimax lower bounds. All these results are new in the literature and will deepen our theoretical understanding of classification with DNNs.

stat.ML↗

arXiv2024

AI WALKUP: A Computer-Vision Approach to Quantifying MDS-UPDRS in Parkinson's Disease

Parkinson's Disease (PD) is the second most common neurodegenerative disorder. The existing assessment method for PD is usually the Movement Disorder Society - Unified Parkinson's Disease Rating Scale (MDS-UPDRS) to assess the severity of various types of motor symptoms and disease progression. However, manual assessment suffers from high subjectivity, lack of consistency, and high cost and low efficiency of manual communication. We want to use a computer vision based solution to capture human pose images based on a camera, reconstruct and perform motion analysis using algorithms, and extract the features of the amount of motion through feature engineering. The proposed approach can be deployed on different smartphones, and the video recording and artificial intelligence analysis can be done quickly and easily through our APP.

cs.CV↗

arXiv2024

Enhancing the General Agent Capabilities of Low-Parameter LLMs through Tuning and Multi-Branch Reasoning

Open-source pre-trained Large Language Models (LLMs) exhibit strong language understanding and generation capabilities, making them highly successful in a variety of tasks. However, when used as agents for dealing with complex problems in the real world, their performance is far inferior to large commercial models such as ChatGPT and GPT-4. As intelligent agents, LLMs need to have the capabilities of task planning, long-term memory, and the ability to leverage external tools to achieve satisfactory performance. Various methods have been proposed to enhance the agent capabilities of LLMs. On the one hand, methods involve constructing agent-specific data and fine-tuning the models. On the other hand, some methods focus on designing prompts that effectively activate the reasoning abilities of the LLMs. We explore both strategies on the 7B and 13B models. We propose a comprehensive method for constructing agent-specific data using GPT-4. Through supervised fine-tuning with constructed data, we find that for these models with a relatively small number of parameters, supervised fine-tuning can significantly reduce hallucination outputs and formatting errors in agent tasks. Furthermore, techniques such as multi-path reasoning and task decomposition can effectively decrease problem complexity and enhance the performance of LLMs as agents. We evaluate our method on five agent tasks of AgentBench and achieve satisfactory results.

cs.CL↗

arXiv2024

Sampling Proper Colorings on Line Graphs Using $(1+o(1))Δ$ Colors

We prove that the single-site Glauber dynamics for sampling proper $q$-colorings mixes in $O_Δ(n\log n)$ time on line graphs with $n$ vertices and maximum degree $Δ$ when $q>(1+o(1))Δ$. The main tool in our proof is the matrix trickle-down theorem developed by Abdolazimi, Liu and Oveis Gharan (FOCS, 2021).

cs.DS↗

arXiv2024

Horizon-Free Regret for Linear Markov Decision Processes

A recent line of works showed regret bounds in reinforcement learning (RL) can be (nearly) independent of planning horizon, a.k.a.~the horizon-free bounds. However, these regret bounds only apply to settings where a polynomial dependency on the size of transition model is allowed, such as tabular Markov Decision Process (MDP) and linear mixture MDP. We give the first horizon-free bound for the popular linear MDP setting where the size of the transition model can be exponentially large or even uncountable. In contrast to prior works which explicitly estimate the transition model and compute the inhomogeneous value functions at different time steps, we directly estimate the value functions and confidence sets. We obtain the horizon-free bound by: (1) maintaining multiple weighted least square estimators for the value functions; and (2) a structural lemma which shows the maximal total variation of the inhomogeneous value functions is bounded by a polynomial factor of the feature dimension.

cs.LG↗

arXiv2024

ResLoRA: Identity Residual Mapping in Low-Rank Adaption

As one of the most popular parameter-efficient fine-tuning (PEFT) methods, low-rank adaptation (LoRA) is commonly applied to fine-tune large language models (LLMs). However, updating the weights of LoRA blocks effectively and expeditiously is challenging due to the long calculation path in the original model. To address this, we propose ResLoRA, an improved framework of LoRA. By adding residual paths during training and using merging approaches to eliminate these extra paths during inference, our method can achieve better results in fewer training steps without any extra trainable parameters or inference cost compared to LoRA. The experiments on NLG, NLU, and text-to-image tasks demonstrate the effectiveness of our method. To the best of our knowledge, ResLoRA is the first work that combines the residual path with LoRA. The code of our method is available at https://github.com/microsoft/LMOps/tree/main/reslora .

cs.CL↗

arXiv2024

HD-Eval: Aligning Large Language Model Evaluators Through Hierarchical Criteria Decomposition

Large language models (LLMs) have emerged as a promising alternative to expensive human evaluations. However, the alignment and coverage of LLM-based evaluations are often limited by the scope and potential bias of the evaluation prompts and criteria. To address this challenge, we propose HD-Eval, a novel framework that iteratively aligns LLM-based evaluators with human preference via Hierarchical Criteria Decomposition. HD-Eval inherits the essence from the evaluation mindset of human experts and enhances the alignment of LLM-based evaluators by decomposing a given evaluation task into finer-grained criteria, aggregating them according to estimated human preferences, pruning insignificant criteria with attribution, and further decomposing significant criteria. By integrating these steps within an iterative alignment training process, we obtain a hierarchical decomposition of criteria that comprehensively captures aspects of natural language at multiple levels of granularity. Implemented as a white box, the human preference-guided aggregator is efficient to train and more explainable than relying solely on prompting, and its independence from model parameters makes it applicable to closed-source LLMs. Extensive experiments on three evaluation domains demonstrate the superiority of HD-Eval in further aligning state-of-the-art evaluators and providing deeper insights into the explanation of evaluation results and the task itself.

cs.CL↗

arXiv2024

We Choose to Go to Space: Agent-driven Human and Multi-Robot Collaboration in Microgravity

We present SpaceAgents-1, a system for learning human and multi-robot collaboration (HMRC) strategies under microgravity conditions. Future space exploration requires humans to work together with robots. However, acquiring proficient robot skills and adept collaboration under microgravity conditions poses significant challenges within ground laboratories. To address this issue, we develop a microgravity simulation environment and present three typical configurations of intra-cabin robots. We propose a hierarchical heterogeneous multi-agent collaboration architecture: guided by foundation models, a Decision-Making Agent serves as a task planner for human-robot collaboration, while individual Skill-Expert Agents manage the embodied control of robots. This mechanism empowers the SpaceAgents-1 system to execute a range of intricate long-horizon HMRC tasks.

cs.RO↗

arXiv2024

Text Diffusion with Reinforced Conditioning

Diffusion models have demonstrated exceptional capability in generating high-quality images, videos, and audio. Due to their adaptiveness in iterative refinement, they provide a strong potential for achieving better non-autoregressive sequence generation. However, existing text diffusion models still fall short in their performance due to a challenge in handling the discreteness of language. This paper thoroughly analyzes text diffusion models and uncovers two significant limitations: degradation of self-conditioning during training and misalignment between training and sampling. Motivated by our findings, we propose a novel Text Diffusion model called TREC, which mitigates the degradation with Reinforced Conditioning and the misalignment by Time-Aware Variance Scaling. Our extensive experiments demonstrate the competitiveness of TREC against autoregressive, non-autoregressive, and diffusion baselines. Moreover, qualitative analysis shows its advanced ability to fully utilize the diffusion process in refining samples.

cs.CL↗

arXiv2024

Improving Domain Adaptation through Extended-Text Reading Comprehension

To enhance the domain-specific capabilities of large language models, continued pre-training on a domain-specific corpus is a prevalent method. Recent work demonstrates that adapting models using reading comprehension data formatted by regex-based patterns can significantly improve performance on domain-specific tasks. However, regex-based patterns are incapable of parsing raw corpora using domain-specific knowledge. Furthermore, the question and answer pairs are extracted directly from the corpus in predefined formats offers limited context. To address this limitation, we improve reading comprehension via LLM and clustering. LLM focuses on leveraging domain knowledge within the corpus to refine comprehension stage, while clustering supplies relevant knowledge by extending the context to enrich reading stage. Additionally, our method incorporates parameter-efficient fine-tuning to improve the efficiency of domain adaptation. In comparison to AdaptLLM, our method achieves an improvement exceeding 5% in domain-specific tasks. Our code will available at https://github.com/microsoft/LMOps.

cs.CL↗

arXiv2024

Universal scaling relation and criticality in metabolism and growth of Escherichia coli

The metabolic network plays a crucial role in regulating bacterial metabolism and growth, but it is subject to inherent molecular stochasticity. Previous studies have utilized flux balance analysis and the maximum entropy method to predict metabolic fluxes and growth rates, while the underlying principles governing bacterial metabolism and growth, especially the criticality hypothesis, remain unclear. In this study, we employ a maximum entropy approach to investigate the universality in various constraint-based metabolic networks of Escherichia coli. Our findings reveal the existence of universal scaling relations across different nutritional environments and metabolic network models, similar to the universality observed in physics. By analyzing single-cell data, we confirm that metabolism of Escherichia coli operates close to the state with maximum Fisher information, which serves as a signature of criticality. This critical state provides functional advantages such as high sensitivity and long-range correlation. Moreover, we demonstrate that a metabolic system operating at criticality takes a compromise solution between growth and adaptation, thereby serving as a survival strategy in fluctuating environments.

q-bio.MN↗

arXiv2024

BS-PLCNet: Band-split Packet Loss Concealment Network with Multi-task Learning Framework and Multi-discriminators

Packet loss is a common and unavoidable problem in voice over internet phone (VoIP) systems. To deal with the problem, we propose a band-split packet loss concealment network (BS-PLCNet). Specifically, we split the full-band signal into wide-band (0-8kHz) and high-band (8-24kHz). The wide-band signals are processed by a gated convolutional recurrent network (GCRN), while the high-band counterpart is processed by a simple GRU network. To ensure high speech quality and automatic speech recognition (ASR) compatibility, multi-task learning (MTL) framework including fundamental frequency (f0) prediction, linguistic awareness, and multi-discriminators are used. The proposed approach tied for 1st place in the ICASSP 2024 PLC Challenge.

eess.AS↗

arXiv2024

SELM: Speech Enhancement Using Discrete Tokens and Language Models

Language models (LMs) have shown superior performances in various speech generation tasks recently, demonstrating their powerful ability for semantic context modeling. Given the intrinsic similarity between speech generation and speech enhancement, harnessing semantic information holds potential advantages for speech enhancement tasks. In light of this, we propose SELM, a novel paradigm for speech enhancement, which integrates discrete tokens and leverages language models. SELM comprises three stages: encoding, modeling, and decoding. We transform continuous waveform signals into discrete tokens using pre-trained self-supervised learning (SSL) models and a k-means tokenizer. Language models then capture comprehensive contextual information within these tokens. Finally, a detokenizer and HiFi-GAN restore them into enhanced speech. Experimental results demonstrate that SELM achieves comparable performance in objective metrics alongside superior results in subjective perception. Our demos are available https://honee-w.github.io/SELM/.

eess.AS↗

arXiv2023

OSM vs HD Maps: Map Representations for Trajectory Prediction

While High Definition (HD) Maps have long been favored for their precise depictions of static road elements, their accessibility constraints and susceptibility to rapid environmental changes impede the widespread deployment of autonomous driving, especially in the motion forecasting task. In this context, we propose to leverage OpenStreetMap (OSM) as a promising alternative to HD Maps for long-term motion forecasting. The contributions of this work are threefold: firstly, we extend the application of OSM to long-horizon forecasting, doubling the forecasting horizon compared to previous studies. Secondly, through an expanded receptive field and the integration of intersection priors, our OSM-based approach exhibits competitive performance, narrowing the gap with HD Map-based models. Lastly, we conduct an exhaustive context-aware analysis, providing deeper insights in motion forecasting across diverse scenarios as well as conducting class-aware comparisons. This research not only advances long-term motion forecasting with coarse map representations but additionally offers a potential scalable solution within the domain of autonomous driving.

cs.CV↗