arXiv subjects

Chenhui Zhang

Publications and source records attributed to Chenhui Zhang.

29 records · Page 2Linked to original sources

arXiv2024

ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools

We introduce ChatGLM, an evolving family of large language models that we have been developing over time. This report primarily focuses on the GLM-4 language series, which includes GLM-4, GLM-4-Air, and GLM-4-9B. They represent our most capable models that are trained with all the insights and lessons gained from the preceding three generations of ChatGLM. To date, the GLM-4 models are pre-trained on ten trillions of tokens mostly in Chinese and English, along with a small set of corpus from 24 languages, and aligned primarily for Chinese and English usage. The high-quality alignment is achieved via a multi-stage post-training process, which involves supervised fine-tuning and learning from human feedback. Evaluations show that GLM-4 1) closely rivals or outperforms GPT-4 in terms of general metrics such as MMLU, GSM8K, MATH, BBH, GPQA, and HumanEval, 2) gets close to GPT-4-Turbo in instruction following as measured by IFEval, 3) matches GPT-4 Turbo (128K) and Claude 3 for long context tasks, and 4) outperforms GPT-4 in Chinese alignments as measured by AlignBench. The GLM-4 All Tools model is further aligned to understand user intent and autonomously decide when and which tool(s) touse -- including web browser, Python interpreter, text-to-image model, and user-defined functions -- to effectively complete complex tasks. In practical applications, it matches and even surpasses GPT-4 All Tools in tasks like accessing online information via web browsing and solving math problems using Python interpreter. Over the course, we have open-sourced a series of models, including ChatGLM-6B (three generations), GLM-4-9B (128K, 1M), GLM-4V-9B, WebGLM, and CodeGeeX, attracting over 10 million downloads on Hugging face in the year 2023 alone. The open models can be accessed through https://github.com/THUDM and https://huggingface.co/THUDM.

cs.CL↗

arXiv2024

Pre-Training and Prompting for Few-Shot Node Classification on Text-Attributed Graphs

The text-attributed graph (TAG) is one kind of important real-world graph-structured data with each node associated with raw texts. For TAGs, traditional few-shot node classification methods directly conduct training on the pre-processed node features and do not consider the raw texts. The performance is highly dependent on the choice of the feature pre-processing method. In this paper, we propose P2TAG, a framework designed for few-shot node classification on TAGs with graph pre-training and prompting. P2TAG first pre-trains the language model (LM) and graph neural network (GNN) on TAGs with self-supervised loss. To fully utilize the ability of language models, we adapt the masked language modeling objective for our framework. The pre-trained model is then used for the few-shot node classification with a mixed prompt method, which simultaneously considers both text and graph information. We conduct experiments on six real-world TAGs, including paper citation networks and product co-purchasing networks. Experimental results demonstrate that our proposed framework outperforms existing graph few-shot learning methods on these datasets with +18.98% ~ +35.98% improvements.

cs.SI↗

arXiv2024

Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compression

Compressing high-capability Large Language Models (LLMs) has emerged as a favored strategy for resource-efficient inferences. While state-of-the-art (SoTA) compression methods boast impressive advancements in preserving benign task performance, the potential risks of compression in terms of safety and trustworthiness have been largely neglected. This study conducts the first, thorough evaluation of three (3) leading LLMs using five (5) SoTA compression techniques across eight (8) trustworthiness dimensions. Our experiments highlight the intricate interplay between compression and trustworthiness, revealing some interesting patterns. We find that quantization is currently a more effective approach than pruning in achieving efficiency and trustworthiness simultaneously. For instance, a 4-bit quantized model retains the trustworthiness of its original counterpart, but model pruning significantly degrades trustworthiness, even at 50% sparsity. Moreover, employing quantization within a moderate bit range could unexpectedly improve certain trustworthiness dimensions such as ethics and fairness. Conversely, extreme quantization to very low bit levels (3 bits) tends to reduce trustworthiness significantly. This increased risk cannot be uncovered by looking at benign performance alone, in turn, mandating comprehensive trustworthiness evaluation in practice. These findings culminate in practical recommendations for simultaneously achieving high utility, efficiency, and trustworthiness in LLMs. Code and models are available at https://decoding-comp-trust.github.io.

cs.CL↗

arXiv2024

DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineer

Large Language Models (LLMs) have emerged as dominant tools for various tasks, particularly when tailored for a specific target by prompt tuning. Nevertheless, concerns surrounding data privacy present obstacles due to the tuned prompts' dependency on sensitive private information. A practical solution is to host a local LLM and optimize a soft prompt privately using data. Yet, hosting a local model becomes problematic when model ownership is protected. Alternative methods, like sending data to the model's provider for training, intensify these privacy issues facing an untrusted provider. In this paper, we present a novel solution called Differentially-Private Offsite Prompt Tuning (DP-OPT) to address this challenge. Our approach involves tuning a discrete prompt on the client side and then applying it to the desired cloud models. We demonstrate that prompts suggested by LLMs themselves can be transferred without compromising performance significantly. To ensure that the prompts do not leak private information, we introduce the first private prompt generation mechanism, by a differentially-private (DP) ensemble of in-context learning with private demonstrations. With DP-OPT, generating privacy-preserving prompts by Vicuna-7b can yield competitive performance compared to non-private in-context learning on GPT3.5 or local private prompt tuning. Codes are available at https://github.com/VITA-Group/DP-OPT .

cs.CL↗

arXiv2024

DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models

Generative Pre-trained Transformer (GPT) models have exhibited exciting progress in their capabilities, capturing the interest of practitioners and the public alike. Yet, while the literature on the trustworthiness of GPT models remains limited, practitioners have proposed employing capable GPT models for sensitive applications such as healthcare and finance -- where mistakes can be costly. To this end, this work proposes a comprehensive trustworthiness evaluation for large language models with a focus on GPT-4 and GPT-3.5, considering diverse perspectives -- including toxicity, stereotype bias, adversarial robustness, out-of-distribution robustness, robustness on adversarial demonstrations, privacy, machine ethics, and fairness. Based on our evaluations, we discover previously unpublished vulnerabilities to trustworthiness threats. For instance, we find that GPT models can be easily misled to generate toxic and biased outputs and leak private information in both training data and conversation history. We also find that although GPT-4 is usually more trustworthy than GPT-3.5 on standard benchmarks, GPT-4 is more vulnerable given jailbreaking system or user prompts, potentially because GPT-4 follows (misleading) instructions more precisely. Our work illustrates a comprehensive trustworthiness evaluation of GPT models and sheds light on the trustworthiness gaps. Our benchmark is publicly available at https://decodingtrust.github.io/ ; our dataset can be previewed at https://huggingface.co/datasets/AI-Secure/DecodingTrust ; a concise version of this work is at https://openreview.net/pdf?id=kaHpo8OZw2 .

cs.CL↗

arXiv2024

Good at captioning, bad at counting: Benchmarking GPT-4V on Earth observation data

Large Vision-Language Models (VLMs) have demonstrated impressive performance on complex tasks involving visual input with natural language instructions. However, it remains unclear to what extent capabilities on natural images transfer to Earth observation (EO) data, which are predominantly satellite and aerial images less common in VLM training data. In this work, we propose a comprehensive benchmark to gauge the progress of VLMs toward being useful tools for EO data by assessing their abilities on scene understanding, localization and counting, and change detection tasks. Motivated by real-world applications, our benchmark includes scenarios like urban monitoring, disaster relief, land use, and conservation. We discover that, although state-of-the-art VLMs like GPT-4V possess extensive world knowledge that leads to strong performance on open-ended tasks like location understanding and image captioning, their poor spatial reasoning limits usefulness on object localization and counting tasks. Our benchmark will be made publicly available at https://vleo.danielz.ch/ and on Hugging Face at https://huggingface.co/collections/mit-ei/vleo-benchmark-datasets-65b789b0466555489cce0d70 for easy model evaluation.

cs.CL↗

arXiv2023

Proton-mediated reversible switching of metastable ferroelectric phases with low operation voltages

The exploration of ferroelectric phase transitions enables an in-depth understanding of ferroelectric switching and promising applications in information storage. However, controllably tuning the dynamics of ferroelectric phase transitions remains challenging owing to inaccessible hidden phases. Here, using protonic gating technology, we create a series of metastable ferroelectric phases and demonstrate their reversible transitions in van der Waals ferroelectric α-In2Se3 transistors. By varying the gate bias, protons can be incrementally injected or extracted, achieving controllable tuning of the ferroelectric α-In2Se3 protonic dynamics across the channel and obtaining numerous intermediate phases. Combining piezoelectric force microscopy and electrical hysteresis measurements, we surprisingly discover that the gate tuning of α-In2Se3 protonation is volatile, unlike the conventional gating effect, and the created new phases remain polar. Their origin, revealed by first-principles calculations, is related to the formation of metastable hydrogen-stabilized α-In2Se3 phases with switchable dipoles. Furthermore, our approach enables ultralow gate-voltage switching of different phases (below 0.4 V), and easy integration with transistor components. This work provides fundamental insights into the interactions between protons and ferroelectric phase transitions, providing a possible avenue for accessing hidden phases in ferroelectric switching and for developing energy-efficient, multilevel-switching memory devices.

cond-mat.mtrl-sci↗

arXiv2022

SCR: Training Graph Neural Networks with Consistency Regularization

We present the SCR framework for enhancing the training of graph neural networks (GNNs) with consistency regularization. Regularization is a set of strategies used in Machine Learning to reduce overfitting and improve the generalization ability. However, it is unclear how to best design the generalization strategies in GNNs, as it works in a semi-supervised setting for graph data. The major challenge lies in how to efficiently balance the trade-off between the error from the labeled data and that from the unlabeled data. SCR is a simple yet general framework in which we introduce two strategies of consistency regularization to address the challenge above. One is to minimize the disagreements among the perturbed predictions by different versions of a GNN model. The other is to leverage the Mean Teacher paradigm to estimate a consistency loss between teacher and student models instead of the disagreement of the predictions. We conducted experiments on three large-scale node classification datasets in the Open Graph Benchmark (OGB). Experimental results demonstrate that the proposed SCR framework is a general one that can enhance various GNNs to achieve better performance. Finally, SCR has been the top-1 entry on all three OGB leaderboards as of this submission.

cs.SI↗

arXiv2020

Unveiling Defect-Mediated Carrier Dynamics in Monolayer Semiconductors by Spatiotemporal Microwave Imaging

The optoelectronic properties of atomically thin transition-metal dichalcogenides are strongly correlated with the presence of defects in the materials, which are not necessarily detrimental for certain applications. For instance, defects can lead to an enhanced photoconduction, a complicated process involving charge generation and recombination in the time domain and carrier transport in the spatial domain. Here, we report the simultaneous spatial and temporal photoconductivity imaging in two types of WS2 monolayers by laser-illuminated microwave impedance microscopy. The diffusion length and carrier lifetime were directly extracted from the spatial profile and temporal relaxation of microwave signals respectively. Time-resolved experiments indicate that the critical process for photo-excited carriers is the escape of holes from trap states, which prolongs the apparent lifetime of mobile electrons in the conduction band. As a result, counterintuitively, the photoconductivity is stronger in CVD samples than exfoliated monolayers with a lower defect density. Our work reveals the intrinsic time and length scales of electrical response to photo-excitation in van der Waals materials, which is essential for their applications in novel optoelectronic devices.

cond-mat.mes-hall↗

arXiv2019

p-type codoping effect in (Ga,Mn)As: Mn lattice location versus magnetic properties

In the present work, we perform a systematic investigation on p-type codoping in (Ga,Mn)As. Through gradually increasing Zn doping concentration, the hole concentration increases, which should theoretically lead to an increase of the Curie temperature (TC) according to the p-d Zener model. Unexpectedly, although the film keeps its epitaxial structure, both TC and the magnetization decrease. The samples present a phase transition from ferromagnetism to paramagnetism upon increasing hole concentration. In the intermediate regime, we observe a signature of antiferromagnetism. By using channeling Rutherford backscattering spectrometry and particle-induced x-ray emission, the substitutional Mn atoms are observed to shift to interstitial sites, while more Zn atoms occupy Ga sites, which explains the observed behavior. This is also consistent with first-principles calculations, showing that the complex of substitutional Zn and interstitial Mn has the lowest formation energy.

cond-mat.mtrl-sci↗

arXiv2018

Giant Planar Hall Effect in the Dirac Semimetal ZrTe5

Recently, giant planar Hall effect originating from chiral anomaly has been predicted in nonmagnetic Dirac/Weyl semimetals. ZrTe5 is considered to be an intriguing Dirac semimetal at the boundary of weak topological insulators and strong topological insulators, though this claim still remains controversial. Here, we report the observation in ZrTe5 of the giant planar Hall resistivity that shows two different magnetic-field dependences as predicted by theory and a maximum at the Lifshitz transition temperature. We found that the giant planar Hall resistivity fades out with decreasing the thickness of ZrTe5 nanoplates, which may be ascribed to the vanishing of the 3D nature of the samples. In addition, we have observed a nontrivial Berry phase, chiral-anomaly-induced negative longitudinal magnetoresistance, and a giant in-plane anisotropic magnetoresistance in these ZrTe5 nanoplates. All the experimental observations demonstrated coherently that ZrTe5 is a Dirac semimetal.

cond-mat.mes-hall↗