Searcharxiv⌕ Search

arXiv subjects

Shuo Zhang

Publications and source records attributed to Shuo Zhang.

At least 181 records · Page 10Linked to original sources
arXiv2024

InternLM-Math: Open Math Large Language Models Toward Verifiable Reasoning

The math abilities of large language models can represent their abstract reasoning ability. In this paper, we introduce and open-source our math reasoning LLMs InternLM-Math which is continue pre-trained from InternLM2. We unify chain-of-thought reasoning, reward modeling, formal reasoning, data augmentation, and code interpreter in a unified seq2seq format and supervise our model to be a versatile math reasoner, verifier, prover, and augmenter. These abilities can be used to develop the next math LLMs or self-iteration. InternLM-Math obtains open-sourced state-of-the-art performance under the setting of in-context learning, supervised fine-tuning, and code-assisted reasoning in various informal and formal benchmarks including GSM8K, MATH, Hungary math exam, MathBench-ZH, and MiniF2F. Our pre-trained model achieves 30.3 on the MiniF2F test set without fine-tuning. We further explore how to use LEAN to solve math problems and study its performance under the setting of multi-task learning which shows the possibility of using LEAN as a unified platform for solving and proving in math. Our models, codes, and data are released at \url{https://github.com/InternLM/InternLM-Math}.

cs.CL↗
arXiv2024

Exploring the Capabilities of Large Multimodal Models on Dense Text

While large multi-modal models (LMM) have shown notable progress in multi-modal tasks, their capabilities in tasks involving dense textual content remains to be fully explored. Dense text, which carries important information, is often found in documents, tables, and product descriptions. Understanding dense text enables us to obtain more accurate information, assisting in making better decisions. To further explore the capabilities of LMM in complex text tasks, we propose the DT-VQA dataset, with 170k question-answer pairs. In this paper, we conduct a comprehensive evaluation of GPT4V, Gemini, and various open-source LMMs on our dataset, revealing their strengths and weaknesses. Furthermore, we evaluate the effectiveness of two strategies for LMM: prompt engineering and downstream fine-tuning. We find that even with automatically labeled training datasets, significant improvements in model performance can be achieved. We hope that this research will promote the study of LMM in dense text tasks. Code will be released at https://github.com/Yuliang-Liu/MultimodalOCR.

cs.CL↗
arXiv2024

BCFPL: Binary classification ConvNet based Fast Parking space recognition with Low resolution image

The automobile plays an important role in the economic activities of mankind, especially in the metropolis. Under the circumstances, the demand of quick search for available parking spaces has become a major concern for the automobile drivers. Meanwhile, the public sense of privacy is also awaking, the image-based parking space recognition methods lack the attention of privacy protection. In this paper, we proposed a binary convolutional neural network with lightweight design structure named BCFPL, which can be used to train with low-resolution parking space images and offer a reasonable recognition result. The images of parking space were collected from various complex environments, including different weather, occlusion conditions, and various camera angles. We conducted the training and testing progresses among different datasets and partial subsets. The experimental results show that the accuracy of BCFPL does not decrease compared with the original resolution image directly, and can reach the average level of the existing mainstream method. BCFPL also has low hardware requirements and fast recognition speed while meeting the privacy requirements, so it has application potential in intelligent city construction and automatic driving field.

cs.CV↗
arXiv2024

Lowest-degree robust finite element schemes for inhomogeneous bi-Laplace problems

In this paper, we study the numerical method for the bi-Laplace problems with inhomogeneous coefficients; particularly, we propose finite element schemes on rectangular grids respectively for an inhomogeneous fourth-order elliptic singular perturbation problem and for the Helmholtz transmission eigenvalue problem. The new methods use the reduced rectangle Morley (RRM for short) element space with piecewise quadratic polynomials, which are of the lowest degree possible. For the finite element space, a discrete analogue of an equality by Grisvard is proved for the stability issue and a locally-averaged interpolation operator is constructed for the approximation issue. Optimal convergence rates of the schemes are proved, and numerical experiments are given to verify the theoretical analysis.

math.NA↗
arXiv2024

Mapping the path to Cryogenic Atom Probe Tomography Analysis of biomolecules

The understanding of protein structure, folding, and interaction with other proteins remains one of the grand challenges of modern biology. Tremendous progress has been made thanks to X-ray- or electron-based techniques that have provided atomic configurations of proteins, and their solvation shell. These techniques though require a large number of similar molecules to provide an average view, and lack detailed compositional information that might play a major role in the biochemical activity of these macromolecules. Based on its intrinsic performance and recent impact in materials science, atom probe tomography (APT) has been touted as a potential novel tool to analyse biological materials, including proteins. However, analysis of biomolecules in their native, hydrated state by APT have not yet been routinely achieved, and the technique's true capabilities remain to be demonstrated. Here, we present and discuss systematic analyses of individual amino-acids in frozen aqueous solutions on two different nanoporous metal supports across a wide range of analysis conditions. Using a ratio of the molecular ions of water as a descriptor for the conditions of electrostatic field, we study the fragmentation and behavior of those amino acids. We discuss the importance sample support, specimen preparation route, acquisition conditions and data analysis, to pave the way towards establishing guidelines for cryo-APT analysis of biomolecules.

physics.bio-ph↗
arXiv2024

Representation Learning of Tangled Key-Value Sequence Data for Early Classification

Key-value sequence data has become ubiquitous and naturally appears in a variety of real-world applications, ranging from the user-product purchasing sequences in e-commerce, to network packet sequences forwarded by routers in networking. Classifying these key-value sequences is important in many scenarios such as user profiling and malicious applications identification. In many time-sensitive scenarios, besides the requirement of classifying a key-value sequence accurately, it is also desired to classify a key-value sequence early, in order to respond fast. However, these two goals are conflicting in nature, and it is challenging to achieve them simultaneously. In this work, we formulate a novel tangled key-value sequence early classification problem, where a tangled key-value sequence is a mixture of several concurrent key-value sequences with different keys. The goal is to classify each individual key-value sequence sharing a same key both accurately and early. To address this problem, we propose a novel method, i.e., Key-Value sequence Early Co-classification (KVEC), which leverages both inner- and inter-correlations of items in a tangled key-value sequence through key correlation and value correlation to learn a better sequence representation. Meanwhile, a time-aware halting policy decides when to stop the ongoing key-value sequence and classify it based on current sequence representation. Experiments on both real-world and synthetic datasets demonstrate that our method outperforms the state-of-the-art baselines significantly. KVEC improves the prediction accuracy by up to $4.7 - 17.5\%$ under the same prediction earliness condition, and improves the harmonic mean of accuracy and earliness by up to $3.7 - 14.0\%$.

cs.LG↗
arXiv2024

InternLM2 Technical Report

The evolution of Large Language Models (LLMs) like ChatGPT and GPT-4 has sparked discussions on the advent of Artificial General Intelligence (AGI). However, replicating such advancements in open-source models has been challenging. This paper introduces InternLM2, an open-source LLM that outperforms its predecessors in comprehensive evaluations across 6 dimensions and 30 benchmarks, long-context modeling, and open-ended subjective evaluations through innovative pre-training and optimization techniques. The pre-training process of InternLM2 is meticulously detailed, highlighting the preparation of diverse data types including text, code, and long-context data. InternLM2 efficiently captures long-term dependencies, initially trained on 4k tokens before advancing to 32k tokens in pre-training and fine-tuning stages, exhibiting remarkable performance on the 200k ``Needle-in-a-Haystack" test. InternLM2 is further aligned using Supervised Fine-Tuning (SFT) and a novel Conditional Online Reinforcement Learning from Human Feedback (COOL RLHF) strategy that addresses conflicting human preferences and reward hacking. By releasing InternLM2 models in different training stages and model sizes, we provide the community with insights into the model's evolution.

cs.CL↗
arXiv2024

Latent CLAP Loss for Better Foley Sound Synthesis

Foley sound generation, the art of creating audio for multimedia, has recently seen notable advancements through text-conditioned latent diffusion models. These systems use multimodal text-audio representation models, such as Contrastive Language-Audio Pretraining (CLAP), whose objective is to map corresponding audio and text prompts into a joint embedding space. AudioLDM, a text-to-audio model, was the winner of the DCASE2023 task 7 Foley sound synthesis challenge. The winning system fine-tuned the model for specific audio classes and applied a post-filtering method using CLAP similarity scores between output audio and input text at inference time, requiring the generation of extra samples, thus reducing data generation efficiency. We introduce a new loss term to enhance Foley sound generation in AudioLDM without post-filtering. This loss term uses a new module based on the CLAP mode-Latent CLAP encode-to align the latent diffusion output with real audio in a shared CLAP embedding space. Our experiments demonstrate that our method effectively reduces the Frechet Audio Distance (FAD) score of the generated audio and eliminates the need for post-filtering, thus enhancing generation efficiency.

eess.AS↗
arXiv2024

TextMonkey: An OCR-Free Large Multimodal Model for Understanding Document

We present TextMonkey, a large multimodal model (LMM) tailored for text-centric tasks. Our approach introduces enhancement across several dimensions: By adopting Shifted Window Attention with zero-initialization, we achieve cross-window connectivity at higher input resolutions and stabilize early training; We hypothesize that images may contain redundant tokens, and by using similarity to filter out significant tokens, we can not only streamline the token length but also enhance the model's performance. Moreover, by expanding our model's capabilities to encompass text spotting and grounding, and incorporating positional information into responses, we enhance interpretability. It also learns to perform screenshot tasks through finetuning. Evaluation on 12 benchmarks shows notable improvements: 5.2% in Scene Text-Centric tasks (including STVQA, TextVQA, and OCRVQA), 6.9% in Document-Oriented tasks (such as DocVQA, InfoVQA, ChartVQA, DeepForm, Kleister Charity, and WikiTableQuestions), and 2.8% in Key Information Extraction tasks (comprising FUNSD, SROIE, and POIE). It outperforms in scene text spotting with a 10.9\% increase and sets a new standard on OCRBench, a comprehensive benchmark consisting of 29 OCR-related assessments, with a score of 561, surpassing previous open-sourced large multimodal models for document understanding. Code will be released at https://github.com/Yuliang-Liu/Monkey.

cs.CV↗
arXiv2024

Scaling Laws of RoPE-based Extrapolation

The extrapolation capability of Large Language Models (LLMs) based on Rotary Position Embedding is currently a topic of considerable interest. The mainstream approach to addressing extrapolation with LLMs involves modifying RoPE by replacing 10000, the rotary base of $θ_n={10000}^{-2n/d}$ in the original RoPE, with a larger value and providing longer fine-tuning text. In this work, we first observe that fine-tuning a RoPE-based LLM with either a smaller or larger base in pre-training context length could significantly enhance its extrapolation performance. After that, we propose \textbf{\textit{Scaling Laws of RoPE-based Extrapolation}}, a unified framework from the periodic perspective, to describe the relationship between the extrapolation performance and base value as well as tuning context length. In this process, we also explain the origin of the RoPE-based extrapolation issue by \textbf{\textit{critical dimension for extrapolation}}. Besides these observations and analyses, we achieve extrapolation up to 1 million context length within only 16K training length on LLaMA2 7B and 13B.

cs.CL↗
arXiv2024

On Truthful Item-Acquiring Mechanisms for Reward Maximization

In this research, we study the problem that a collector acquires items from the owner based on the item qualities the owner declares and an independent appraiser's assessments. The owner is interested in maximizing the probability that the collector acquires the items and is the only one who knows the items' factual quality. The appraiser performs her duties with impartiality, but her assessment may be subject to random noises, so it may not accurately reflect the factual quality of the items. The main challenge lies in devising mechanisms that prompt the owner to reveal accurate information, thereby optimizing the collector's expected reward. We consider the menu size of mechanisms as a measure of their practicability and study its impact on the attainable expected reward. For the single-item setting, we design optimal mechanisms with a monotone increasing menu size. Although the reward gap between the simplest and optimal mechanisms is bounded, we show that simple mechanisms with a small menu size cannot ensure any positive fraction of the optimal reward of mechanisms with a larger menu size. For the multi-item setting, we show that an ordinal mechanism that only takes the owner's ordering of the items as input is not incentive-compatible. We then propose a set of Union mechanisms that combine single-item mechanisms. Moreover, we run experiments to examine these mechanisms' robustness against the independent appraiser's assessment accuracy and the items' acquiring rate.

cs.GT↗
arXiv2024

AFN: Adaptive Fusion Normalization via an Encoder-Decoder Framework

The success of deep learning is inseparable from normalization layers. Researchers have proposed various normalization functions, and each of them has both advantages and disadvantages. In response, efforts have been made to design a unified normalization function that combines all normalization procedures and mitigates their weaknesses. We also proposed a new normalization function called Adaptive Fusion Normalization. Through experiments, we demonstrate AFN outperforms the previous normalization techniques in domain generalization and image classification tasks.

cs.LG↗
arXiv2024

CatNorth: An Improved Gaia DR3 Quasar Candidate Catalog with Pan-STARRS1 and CatWISE

A complete and pure sample of quasars with accurate redshifts is crucial for quasar studies and cosmology. In this paper, we present CatNorth, an improved Gaia DR3 quasar candidate catalog with more than 1.5 million sources in the 3$π$ sky built with data from Gaia, Pan-STARRS1, and CatWISE2020. The XGBoost algorithm is used to reclassify the original Gaia DR3 quasar candidates as stars, galaxies, and quasars. To construct training/validation datasets for the classification, we carefully built two different master stellar samples in addition to the spectroscopic galaxy and quasar samples. An ensemble classification model is obtained by averaging two XGBoost classifiers trained with different master stellar samples. Using a probability threshold of $p_{\mathrm{QSO\_mean}}>0.95$ in our ensemble classification model and an additional cut on the logarithmic probability density of zero proper motion, we retrieved 1,545,514 reliable quasar candidates from the parent Gaia DR3 quasar candidate catalog. We provide photometric redshifts for all candidates with an ensemble regression model. For a subset of 89,100 candidates, accurate spectroscopic redshifts are estimated with the Convolutional Neural Network from the Gaia BP/RP spectra. The CatNorth catalog has a high purity of ~ 90% while maintaining high completeness, which is an ideal sample to understand the quasar population and its statistical properties. The CatNorth catalog is used as the main source of input catalog for the LAMOST phase III quasar survey, which is expected to build a highly complete sample of bright quasars with $i < 19.5$.

astro-ph.GA↗
arXiv2024

Enhancing Adversarial Attacks: The Similar Target Method

Deep neural networks are vulnerable to adversarial examples, posing a threat to the models' applications and raising security concerns. An intriguing property of adversarial examples is their strong transferability. Several methods have been proposed to enhance transferability, including ensemble attacks which have demonstrated their efficacy. However, prior approaches simply average logits, probabilities, or losses for model ensembling, lacking a comprehensive analysis of how and why model ensembling significantly improves transferability. In this paper, we propose a similar targeted attack method named Similar Target~(ST). By promoting cosine similarity between the gradients of each model, our method regularizes the optimization direction to simultaneously attack all surrogate models. This strategy has been proven to enhance generalization ability. Experimental results on ImageNet validate the effectiveness of our approach in improving adversarial transferability. Our method outperforms state-of-the-art attackers on 18 discriminative classifiers and adversarially trained models.

cs.CV↗
arXiv2023

Quantum State Compression Shadow

Quantum state readout serves as the cornerstone of quantum information processing, exerting profound influence on quantum communication, computation, and metrology. In this study, we introduce an innovative readout architecture called Compression Shadow (CompShadow), which transforms the conventional readout paradigm by compressing multi-qubit states into single-qubit shadows before measurement. Compared to direct measurements of the initial quantum states, CompShadow achieves comparable accuracy in amplitude and observable expectation estimation while consuming similar measurement resources. Furthermore, its implementation on near-term quantum hardware with nearest-neighbor coupling architectures is straightforward. Significantly, CompShadow brings forth novel features, including the complete suppression of correlated readout noise, fundamentally reducing the quantum hardware demands for readout. It also facilitates the exploration of multi-body system properties through single-qubit probes and opens the door to designing quantum communication protocols with exponential loss suppression. Our findings mark the emergence of a new era in quantum state readout, setting the stage for a revolutionary leap in quantum information processing capabilities.

quant-ph↗
arXiv2023

MERGE: Fast Private Text Generation

The drastic increase in language models' parameters has led to a new trend of deploying models in cloud servers, raising growing concerns about private inference for Transformer-based models. Existing two-party privacy-preserving techniques, however, only take into account natural language understanding (NLU) scenarios. Private inference in natural language generation (NLG), crucial for applications like translation and code completion, remains underexplored.In addition, previous privacy-preserving techniques suffer from convergence issues during model training and exhibit poor inference speed when used with NLG models due to the neglect of time-consuming operations in auto-regressive generations. To address these issues, we propose a fast private text generation framework for Transformer-based language models, namely MERGE.MERGE reuses the output hidden state as the word embedding to bypass the embedding computation and reorganize the linear operations in the Transformer module to accelerate the forward procedure. Extensive experiments show that MERGE achieves a 26.5x speedup to the vanilla encrypted model under the sequence length 512, and reduces 80\% communication cost, with an up to 10x speedup to state-of-the-art approximated models.

cs.CL↗
arXiv2023

Protein Language Model-Powered 3D Ligand Binding Site Prediction from Protein Sequence

Prediction of ligand binding sites of proteins is a fundamental and important task for understanding the function of proteins and screening potential drugs. Most existing methods require experimentally determined protein holo-structures as input. However, such structures can be unavailable on novel or less-studied proteins. To tackle this limitation, we propose LaMPSite, which only takes protein sequences and ligand molecular graphs as input for ligand binding site predictions. The protein sequences are used to retrieve residue-level embeddings and contact maps from the pre-trained ESM-2 protein language model. The ligand molecular graphs are fed into a graph neural network to compute atom-level embeddings. Then we compute and update the protein-ligand interaction embedding based on the protein residue-level embeddings and ligand atom-level embeddings, and the geometric constraints in the inferred protein contact map and ligand distance map. A final pooling on protein-ligand interaction embedding would indicate which residues belong to the binding sites. Without any 3D coordinate information of proteins, our proposed model achieves competitive performance compared to baseline methods that require 3D protein structures when predicting binding sites. Given that less than 50% of proteins have reliable structure information in the current stage, LaMPSite will provide new opportunities for drug discovery.

q-bio.QM↗
arXiv2023

CoLLiE: Collaborative Training of Large Language Models in an Efficient Way

Large language models (LLMs) are increasingly pivotal in a wide range of natural language processing tasks. Access to pre-trained models, courtesy of the open-source community, has made it possible to adapt these models to specific applications for enhanced performance. However, the substantial resources required for training these models necessitate efficient solutions. This paper introduces CoLLiE, an efficient library that facilitates collaborative training of large language models using 3D parallelism, parameter-efficient fine-tuning (PEFT) methods, and optimizers such as Lion, Adan, Sophia, LOMO and AdaLomo. With its modular design and comprehensive functionality, CoLLiE offers a balanced blend of efficiency, ease of use, and customization. CoLLiE has proven superior training efficiency in comparison with prevalent solutions in pre-training and fine-tuning scenarios. Furthermore, we provide an empirical evaluation of the correlation between model size and GPU memory consumption under different optimization methods, as well as an analysis of the throughput. Lastly, we carry out a comprehensive comparison of various optimizers and PEFT methods within the instruction-tuning context. CoLLiE is available at https://github.com/OpenLMLab/collie.

cs.CL↗