arXiv subjects

Zili Wang

Publications and source records attributed to Zili Wang.

At least 73 records · Page 4Linked to original sources

arXiv2024

AIM: Let Any Multi-modal Large Language Models Embrace Efficient In-Context Learning

In-context learning (ICL) facilitates Large Language Models (LLMs) exhibiting emergent ability on downstream tasks without updating billions of parameters. However, in the area of multi-modal Large Language Models (MLLMs), two problems hinder the application of multi-modal ICL: (1) Most primary MLLMs are only trained on single-image datasets, making them unable to read multi-modal demonstrations. (2) With the demonstrations increasing, thousands of visual tokens highly challenge hardware and degrade ICL performance. During preliminary explorations, we discovered that the inner LLM tends to focus more on the linguistic modality within multi-modal demonstrations to generate responses. Therefore, we propose a general and light-weighted framework \textbf{AIM} to tackle the mentioned problems through \textbf{A}ggregating \textbf{I}mage information of \textbf{M}ultimodal demonstrations to the dense latent space of the corresponding linguistic part. Specifically, AIM first uses the frozen backbone MLLM to read each image-text demonstration and extracts the vector representations on top of the text. These vectors naturally fuse the information of the image-text pair, and AIM transforms them into fused virtual tokens acceptable for the inner LLM via a trainable projection layer. Ultimately, these fused tokens function as variants of multi-modal demonstrations, fed into the MLLM to direct its response to the current query as usual. Because these fused tokens stem from the textual component of the image-text pair, a multi-modal demonstration is nearly reduced to a pure textual demonstration, thus seamlessly applying to any MLLMs. With its de facto MLLM frozen, AIM is parameter-efficient and we train it on public multi-modal web corpora which have nothing to do with downstream test tasks.

cs.MM↗

arXiv2024

GW-MoE: Resolving Uncertainty in MoE Router with Global Workspace Theory

Mixture-of-Experts (MoE) has been demonstrated as an efficient method to scale up models. By dynamically and sparsely selecting activated experts, MoE can effectively reduce computational costs. Despite the success, we observe that many tokens in the MoE models have uncertain routing results. These tokens have nearly equal scores for choosing each expert, and we demonstrate that this uncertainty can lead to incorrect selections. Inspired by the Global Workspace Theory (GWT), we propose a new fine-tuning method, GW-MoE, to address this issue. The core idea is to broadcast the uncertain tokens across experts during fine-tuning. Therefore, these tokens can acquire the necessary knowledge from any expert during inference and become less sensitive to the choice. GW-MoE does not introduce additional inference overhead. We validate that GW can mitigate the uncertain problem and consistently improve in different tasks (text classification, question answering, summarization, code generation, and mathematical problem solving) and model sizes (650M and 8B parameters).

cs.LG↗

arXiv2024

A Novel Garment Transfer Method Supervised by Distilled Knowledge of Virtual Try-on Model

This paper proposes a novel garment transfer method supervised with knowledge distillation from virtual try-on. Our method first reasons the transfer parsing to provide shape prior to downstream tasks. We employ a multi-phase teaching strategy to supervise the training of the transfer parsing reasoning model, learning the response and feature knowledge from the try-on parsing reasoning model. To correct the teaching error, it transfers the garment back to its owner to absorb the hard knowledge in the self-study phase. Guided by the transfer parsing, we adjust the position of the transferred garment via STN to prevent distortion. Afterward, we estimate a progressive flow to precisely warp the garment with shape and content correspondences. To ensure warping rationality, we supervise the training of the garment warping model using target shape and warping knowledge from virtual try-on. To better preserve body features in the transfer result, we propose a well-designed training strategy for the arm regrowth task to infer new exposure skin. Experiments demonstrate that our method has state-of-the-art performance compared with other virtual try-on and garment transfer methods in garment transfer, especially for preserving garment texture and body features.

cs.CV↗

arXiv2024

Lyapunov Neural Network with Region of Attraction Search

Deep learning methods have been widely used in robotic applications, making learning-enabled control design for complex nonlinear systems a promising direction. Although deep reinforcement learning methods have demonstrated impressive empirical performance, they lack the stability guarantees that are important in safety-critical situations. One way to provide these guarantees is to learn Lyapunov certificates alongside control policies. There are three related problems: 1) verify that a given Lyapunov function candidate satisfies the conditions for a given controller on a region, 2) find a valid Lyapunov function and controller on a given region, and 3) find a valid Lyapunov function and a controller such that the region of attraction is as large as possible. Previous work has shown that if the dynamics are piecewise linear, it is possible to solve problems 1) and 2) by solving a Mixed-Integer Linear Program (MILP). In this work, we build upon this method by proposing a Lyapunov neural network that considers monotonicity over half spaces in different directions. We 1) propose a specific choice of Lyapunov function architecture that ensures non-negativity and a unique global minimum by construction, and 2) show that this can be leveraged to find the controller and Lyapunov certificates faster and with a larger valid region by maximizing the size of a square inscribed in a given level set. We apply our method to a 2D inverted pendulum, unicycle path following, a 3-D feedback system, and a 4-D cart pole system, and demonstrate it can shorten the training time by half compared to the baseline, as well as find a larger ROA.

eess.SY↗

arXiv2024

Xiezhi: An Ever-Updating Benchmark for Holistic Domain Knowledge Evaluation

New Natural Langauge Process~(NLP) benchmarks are urgently needed to align with the rapid development of large language models (LLMs). We present Xiezhi, the most comprehensive evaluation suite designed to assess holistic domain knowledge. Xiezhi comprises multiple-choice questions across 516 diverse disciplines ranging from 13 different subjects with 249,587 questions and accompanied by Xiezhi-Specialty and Xiezhi-Interdiscipline, both with 15k questions. We conduct evaluation of the 47 cutting-edge LLMs on Xiezhi. Results indicate that LLMs exceed average performance of humans in science, engineering, agronomy, medicine, and art, but fall short in economics, jurisprudence, pedagogy, literature, history, and management. We anticipate Xiezhi will help analyze important strengths and shortcomings of LLMs, and the benchmark is released in~\url{https://github.com/MikeGu721/XiezhiBenchmark}.

cs.CL↗

arXiv2024

Beyond Language Models: Byte Models are Digital World Simulators

Traditional deep learning often overlooks bytes, the basic units of the digital world, where all forms of information and operations are encoded and manipulated in binary format. Inspired by the success of next token prediction in natural language processing, we introduce bGPT, a model with next byte prediction to simulate the digital world. bGPT matches specialized models in performance across various modalities, including text, audio, and images, and offers new possibilities for predicting, simulating, and diagnosing algorithm or hardware behaviour. It has almost flawlessly replicated the process of converting symbolic music data, achieving a low error rate of 0.0011 bits per byte in converting ABC notation to MIDI format. In addition, bGPT demonstrates exceptional capabilities in simulating CPU behaviour, with an accuracy exceeding 99.99% in executing various operations. Leveraging next byte prediction, models like bGPT can directly learn from vast binary data, effectively simulating the intricate patterns of the digital world.

cs.LG↗

arXiv2024

ChatMusician: Understanding and Generating Music Intrinsically with LLM

While Large Language Models (LLMs) demonstrate impressive capabilities in text generation, we find that their ability has yet to be generalized to music, humanity's creative language. We introduce ChatMusician, an open-source LLM that integrates intrinsic musical abilities. It is based on continual pre-training and finetuning LLaMA2 on a text-compatible music representation, ABC notation, and the music is treated as a second language. ChatMusician can understand and generate music with a pure text tokenizer without any external multi-modal neural structures or tokenizers. Interestingly, endowing musical abilities does not harm language abilities, even achieving a slightly higher MMLU score. Our model is capable of composing well-structured, full-length music, conditioned on texts, chords, melodies, motifs, musical forms, etc, surpassing GPT-4 baseline. On our meticulously curated college-level music understanding benchmark, MusicTheoryBench, ChatMusician surpasses LLaMA2 and GPT-3.5 on zero-shot setting by a noticeable margin. Our work reveals that LLMs can be an excellent compressor for music, but there remains significant territory to be conquered. We release our 4B token music-language corpora MusicPile, the collected MusicTheoryBench, code, model and demo in GitHub.

cs.SD↗

arXiv2024

Fast Deterministic Black-box Context-free Grammar Inference

Black-box context-free grammar inference is a hard problem as in many practical settings it only has access to a limited number of example programs. The state-of-the-art approach Arvada heuristically generalizes grammar rules starting from flat parse trees and is non-deterministic to explore different generalization sequences. We observe that many of Arvada's generalization steps violate common language concept nesting rules. We thus propose to pre-structure input programs along these nesting rules, apply learnt rules recursively, and make black-box context-free grammar inference deterministic. The resulting TreeVada yielded faster runtime and higher-quality grammars in an empirical comparison. The TreeVada source code, scripts, evaluation parameters, and training data are open-source and publicly available (https://doi.org/10.6084/m9.figshare.23907738).

cs.SE↗

arXiv2023

PG-VTON: A Novel Image-Based Virtual Try-On Method via Progressive Inference Paradigm

Virtual try-on is a promising computer vision topic with a high commercial value wherein a new garment is visually worn on a person with a photo-realistic effect. Previous studies conduct their shape and content inference at one stage, employing a single-scale warping mechanism and a relatively unsophisticated content inference mechanism. These approaches have led to suboptimal results in terms of garment warping and skin reservation under challenging try-on scenarios. To address these limitations, we propose a novel virtual try-on method via progressive inference paradigm (PGVTON) that leverages a top-down inference pipeline and a general garment try-on strategy. Specifically, we propose a robust try-on parsing inference method by disentangling semantic categories and introducing consistency. Exploiting the try-on parsing as the shape guidance, we implement the garment try-on via warping-mapping-composition. To facilitate adaptation to a wide range of try-on scenarios, we adopt a covering more and selecting one warping strategy and explicitly distinguish tasks based on alignment. Additionally, we regulate StyleGAN2 to implement re-naked skin inpainting, conditioned on the target skin shape and spatial-agnostic skin features. Experiments demonstrate that our method has state-of-the-art performance under two challenging scenarios. The code will be available at https://github.com/NerdFNY/PGVTON.

cs.CV↗

arXiv2023

GSDC Transformer: An Efficient and Effective Cue Fusion for Monocular Multi-Frame Depth Estimation

Depth estimation provides an alternative approach for perceiving 3D information in autonomous driving. Monocular depth estimation, whether with single-frame or multi-frame inputs, has achieved significant success by learning various types of cues and specializing in either static or dynamic scenes. Recently, these cues fusion becomes an attractive topic, aiming to enable the combined cues to perform well in both types of scenes. However, adaptive cue fusion relies on attention mechanisms, where the quadratic complexity limits the granularity of cue representation. Additionally, explicit cue fusion depends on precise segmentation, which imposes a heavy burden on mask prediction. To address these issues, we propose the GSDC Transformer, an efficient and effective component for cue fusion in monocular multi-frame depth estimation. We utilize deformable attention to learn cue relationships at a fine scale, while sparse attention reduces computational requirements when granularity increases. To compensate for the precision drop in dynamic scenes, we represent scene attributes in the form of super tokens without relying on precise shapes. Within each super token attributed to dynamic scenes, we gather its relevant cues and learn local dense relationships to enhance cue fusion. Our method achieves state-of-the-art performance on the KITTI dataset with efficient fusion speed.

cs.CV↗

arXiv2023

RefGPT: Dialogue Generation of GPT, by GPT, and for GPT

Large Language Models (LLMs) have attained the impressive capability to resolve a wide range of NLP tasks by fine-tuning high-quality instruction data. However, collecting human-written data of high quality, especially multi-turn dialogues, is expensive and unattainable for most people. Though previous studies have used powerful LLMs to generate the dialogues automatically, they all suffer from generating untruthful dialogues because of the model hallucination. Therefore, we propose a method called RefGPT to generate enormous truthful and customized dialogues without worrying about factual errors caused by the model hallucination. RefGPT solves the model hallucination in dialogue generation by restricting the LLMs to leverage the given reference instead of reciting their own knowledge to generate dialogues. Additionally, RefGPT adds detailed controls on every utterance to enable high customization capability, which previous studies have ignored. On the basis of RefGPT, we also propose two high-quality dialogue datasets generated by GPT-4, namely RefGPT-Fact and RefGPT-Code. RefGPT-Fact is a dataset with 100k multi-turn dialogues based on factual knowledge and RefGPT-Code has 76k multi-turn dialogues covering a wide range of coding scenarios. Our code and datasets are released in https://github.com/mutonix/RefGPT.

cs.CL↗

arXiv2023

A Cross-Scale Hierarchical Transformer with Correspondence-Augmented Attention for inferring Bird's-Eye-View Semantic Segmentation

As bird's-eye-view (BEV) semantic segmentation is simple-to-visualize and easy-to-handle, it has been applied in autonomous driving to provide the surrounding information to downstream tasks. Inferring BEV semantic segmentation conditioned on multi-camera-view images is a popular scheme in the community as cheap devices and real-time processing. The recent work implemented this task by learning the content and position relationship via the vision Transformer (ViT). However, the quadratic complexity of ViT confines the relationship learning only in the latent layer, leaving the scale gap to impede the representation of fine-grained objects. And their plain fusion method of multi-view features does not conform to the information absorption intention in representing BEV features. To tackle these issues, we propose a novel cross-scale hierarchical Transformer with correspondence-augmented attention for semantic segmentation inferring. Specifically, we devise a hierarchical framework to refine the BEV feature representation, where the last size is only half of the final segmentation. To save the computation increase caused by this hierarchical framework, we exploit the cross-scale Transformer to learn feature relationships in a reversed-aligning way, and leverage the residual connection of BEV features to facilitate information transmission between scales. We propose correspondence-augmented attention to distinguish conducive and inconducive correspondences. It is implemented in a simple yet effective way, amplifying attention scores before the Softmax operation, so that the position-view-related and the position-view-disrelated attention scores are highlighted and suppressed. Extensive experiments demonstrate that our method has state-of-the-art performance in inferring BEV semantic segmentation conditioned on multi-camera-view images.

cs.CV↗

arXiv2022

Few-shot Query-Focused Summarization with Prefix-Merging

Query-focused summarization has been considered as an important extension for text summarization. It aims to generate a concise highlight for a given query. Different from text summarization, query-focused summarization has long been plagued by the problem of lacking high-quality large-scale datasets. In this paper, we investigate the idea that whether we can integrate and transfer the knowledge of text summarization and question answering to assist the few-shot learning in query-focused summarization. Here, we propose prefix-merging, a prefix-based pretraining strategy for few-shot learning in query-focused summarization. Drawn inspiration from prefix-tuning, we are allowed to integrate the task knowledge from text summarization and question answering into a properly designed prefix and apply the merged prefix to query-focused summarization. With only a small amount of trainable parameters, prefix-merging outperforms fine-tuning on query-focused summarization. We further discuss the influence of different prefix designs and propose a visualized explanation for how prefix-merging works.

cs.CL↗

arXiv2022

Physical Logic Enhanced Network for Small-Sample Bi-Layer Metallic Tubes Bending Springback Prediction

Bi-layer metallic tube (BMT) plays an extremely crucial role in engineering applications, with rotary draw bending (RDB) the high-precision bending processing can be achieved, however, the product will further springback. Due to the complex structure of BMT and the high cost of dataset acquisi-tion, the existing methods based on mechanism research and machine learn-ing cannot meet the engineering requirements of springback prediction. Based on the preliminary mechanism analysis, a physical logic enhanced network (PE-NET) is proposed. The architecture includes ES-NET which equivalent the BMT to the single-layer tube, and SP-NET for the final predic-tion of springback with sufficient single-layer tube samples. Specifically, in the first stage, with the theory-driven pre-exploration and the data-driven pretraining, the ES-NET and SP-NET are constructed, respectively. In the second stage, under the physical logic, the PE-NET is assembled by ES-NET and SP-NET and then fine-tuned with the small sample BMT dataset and composite loss function. The validity and stability of the proposed method are verified by the FE simulation dataset, the small-sample dataset BMT springback angle prediction is achieved, and the method potential in inter-pretability and engineering applications are demonstrated.

cs.LG↗

arXiv2022

Digital-twin-enhanced metal tube bending forming real-time prediction method based on Multi-source-input MTL

As one of the most widely used metal tube bending methods, the rotary draw bending (RDB) process enables reliable and high-precision metal tube bending forming (MTBF). The forming accuracy is seriously affected by the springback and other potential forming defects, of which the mechanism analysis is difficult to deal with. At the same time, the existing methods are mainly conducted in offline space, ignoring the real-time information in the physical world, which is unreliable and inefficient. To address this issue, a digital-twin-enhanced (DT-enhanced) metal tube bending forming real-time prediction method based on multi-source-input multi-task learning (MTL) is proposed. The new method can achieve comprehensive MTBF real-time prediction. By sharing the common feature of the multi-close domain and adopting group regularization strategy on feature sharing and accepting layers, the accuracy and efficiency of the multi-source-input MTL can be guaranteed. Enhanced by DT, the physical real-time deformation data is aligned in the image dimension by an improved Grammy Angle Field (GAF) conversion, realizing the reflection of the actual processing. Different from the traditional offline prediction methods, the new method integrates the virtual and physical data to achieve a more efficient and accurate real-time prediction result. and the DT mapping connection between virtual and physical systems can be achieved. To exclude the effects of equipment errors, the effectiveness of the proposed method is verified on the physical experiment-verified FE simulation scenarios. At the same time, the common pre-training networks are compared with the proposed method. The results show that the proposed DT-enhanced prediction method is more accurate and efficient.

cs.LG↗

arXiv2022

Flip Distance and Triangulations of a Ball

It is known that the flip distance between two triangulations of a convex polygon is related to the minimum number of tetrahedra in the triangulation of some polyhedron. It is interesting to know whether these two numbers are the same. In this work, we find examples to show that the two numbers are different in nature, and their ratio can be arbitrarily close to $\frac{3}{2}$.

math.GT↗

arXiv2022

Bearing-Based Formation Control with Optimal Motion Trajectory

Bearing-based distributed formation control is attractive because it can be implemented using vision-based measurements to achieve a desired formation. Gradient-descent-based controllers using bearing measurements have been shown to have many beneficial characteristics, such as global convergence, applicability to different graph topologies and workspaces of arbitrary dimension, and some flexibility in the choice of the cost. In practice, however, such controllers typically yield convoluted paths from their initial location to the final position in the formation. In this paper we propose a novel procedure to optimize gradient-descent-based bearing-based formation controllers to obtain shorter paths. Our approach is based on the parameterization of the cost function and, by extension, of the controller. We form and solve a nonlinear optimization problem with the sum of path lengths of the agent trajectories as the objective and subject to the original equilibria and global convergence conditions for formation control. Our simulation shows that the parameters can be optimized from a very small number of training samples (1 to 7) to straighten the trajectory by around 16% for a large number of random initial conditions for bearing-only formation. However, in the absence of any range information, the scale of the formation is not fixed and this optimization may lead to an undesired compression of the formation size. Including range measurements avoids this issue and leads to further trajectories straightening by 66%.

eess.SY↗

arXiv2020

Fact-level Extractive Summarization with Hierarchical Graph Mask on BERT

Most current extractive summarization models generate summaries by selecting salient sentences. However, one of the problems with sentence-level extractive summarization is that there exists a gap between the human-written gold summary and the oracle sentence labels. In this paper, we propose to extract fact-level semantic units for better extractive summarization. We also introduce a hierarchical structure, which incorporates the multi-level of granularities of the textual information into the model. In addition, we incorporate our model with BERT using a hierarchical graph mask. This allows us to combine BERT's ability in natural language understanding and the structural information without increasing the scale of the model. Experiments on the CNN/DaliyMail dataset show that our model achieves state-of-the-art results.

cs.CL↗