arXiv subjects

Tian Jin

Publications and source records attributed to Tian Jin.

36 records · Page 2Linked to original sources

arXiv2025

Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding

Decoding with autoregressive large language models (LLMs) traditionally occurs sequentially, generating one token after another. An emerging line of work explored parallel decoding by identifying and simultaneously generating semantically independent chunks of LLM responses. However, these techniques rely on hand-crafted heuristics tied to syntactic structures like lists and paragraphs, making them rigid and imprecise. We present PASTA, a learning-based system that teaches LLMs to identify semantic independence and express parallel decoding opportunities in their own responses. At its core are PASTA-LANG and its interpreter: PASTA-LANG is an annotation language that enables LLMs to express semantic independence in their own responses; the language interpreter acts on these annotations to orchestrate parallel decoding on-the-fly at inference time. Through a two-stage finetuning process, we train LLMs to generate PASTA-LANG annotations that optimize both response quality and decoding speed. Evaluation on AlpacaEval, an instruction following benchmark, shows that our approach Pareto-dominates existing methods in terms of decoding speed and response quality; our results demonstrate geometric mean speedups ranging from 1.21x to 1.93x with corresponding quality changes of +2.2% to -7.1%, measured by length-controlled win rates against sequential decoding baseline.

cs.CL↗

arXiv2025

Synthesizing Post-Training Data for LLMs through Multi-Agent Simulation

Post-training is essential for enabling large language models (LLMs) to follow human instructions. However, its effectiveness depends on high-quality instruction data, which is challenging to obtain in the real world due to privacy concerns, data scarcity, and high annotation costs. To fill this gap, inspired by the recent success of using LLMs to simulate human society, we propose MATRIX, a multi-agent simulator that automatically generates diverse text-based scenarios, capturing a wide range of real-world human needs in a realistic and scalable manner. Leveraging these outputs, we introduce a novel scenario-driven instruction generator MATRIX-Gen for controllable and highly realistic data synthesis. Extensive experiments demonstrate that our framework effectively generates both general and domain-specific data. On AlpacaEval 2 and Arena-Hard benchmarks, Llama-3-8B-Base, post-trained on datasets synthesized by MATRIX-Gen with just 20K instruction-response pairs, outperforms Meta's Llama-3-8B-Instruct model, which was trained on over 10M pairs.

cs.AI↗

arXiv2024

The relative James construction and its application to homotopy groups

In this paper, we develop the new method to compute the homotopy groups of the mapping cone $C_f=Y\cup_{f}CX$ beyond the metastable range by analysing the homotopy of the $n$-th filtration of the relative James construction $J(X,A)$ for CW-pair $A\stackrel{i}\hookrightarrow X$, defined by B. Gray, which is homotopy equivalent to the homotopy fiber of the pinch map $X\cup_{i}CA\rightarrow ΣA$. As an application, we compute the 5 and 6-dim unstable homotopy groups of 3-dimensional mod $2^r$ Moore spaces for all positive integers $r$.

math.AT↗

arXiv2024

Using Geometrical information to Measure the Vibration of A Swaying Millimeter-wave Radar

This paper presents two new, simple yet effective approaches to measure the vibration of a swaying millimeter-wave radar (mmRadar) utilizing geometrical information. Specifically, for the planar vibrations, we firstly establish an equation based on the area difference between the swaying mmRadar and the reference objects at different moments, which enables the quantification of planar displacement. Secondly, volume differences are also utilized with the same idea, achieving the self-vibration measurement of a swaying mmRadar for spatial vibrations. Experimental results confirm the effectiveness of our methods, demonstrating its capability to estimate both the amplitude and a crude direction of the mmRadar's self-vibration.

eess.SP↗

arXiv2024

BERT: Accelerating Vital Signs Measurement for Bioradar with An Efficient Recursive Technique

Recent years have witnessed the great advance of bioradar system in smart sensing of vital signs (VS) for human healthcare monitoring. As an important part of VS sensing process, VS measurement aims to capture the chest wall micromotion induced by the human respiratory and cardiac activities. Unfortunately, the existing VS measurement methods using bioradar have encountered bottlenecks in making a trade-off between time cost and measurement accuracy. To break this bottleneck, this letter proposes an efficient recursive technique (BERT) heuristically, based on the observation that the features of bioradar VS meet the conditions of Markov model. Extensive experimental results validate that BERT measurement yields lower time costs, competitive estimates of heart rate, breathing rate, and heart rate variability. Our BERT method is promising us a new and superior option to measure VS for bioradar. This work seeks not only to solve the current issue of how to accelerate VS measurement with an acceptable accuracy, but also to inspire creative new ideas that spur further advances in this promising field in the future.

eess.SP↗

arXiv2024

Self-Selected Attention Span for Accelerating Large Language Model Inference

Large language models (LLMs) can solve challenging tasks. However, their inference computation on modern GPUs is highly inefficient due to the increasing number of tokens they must attend to as they generate new ones. To address this inefficiency, we capitalize on LLMs' problem-solving capabilities to optimize their own inference-time efficiency. We demonstrate with two specific tasks: (a) evaluating complex arithmetic expressions and (b) summarizing news articles. For both tasks, we create custom datasets to fine-tune an LLM. The goal of fine-tuning is twofold: first, to make the LLM learn to solve the evaluation or summarization task, and second, to train it to identify the minimal attention spans required for each step of the task. As a result, the fine-tuned model is able to convert these self-identified minimal attention spans into sparse attention masks on-the-fly during inference. We develop a custom CUDA kernel to take advantage of the reduced context to attend to. We demonstrate that using this custom CUDA kernel improves the throughput of LLM inference by 28%. Our work presents an end-to-end demonstration showing that training LLMs to self-select their attention spans speeds up autoregressive inference in solving real-world tasks.

cs.CL↗

arXiv2024

The $n+3$, $n+4$ dimensional homotopy groups of $\mathbf{A}_n^2$-complexes

In this paper, we calculate the $n+3$, $n+4$ dimensional homotopy groups of indecomposable $\mathbf{A}_n^2$-complexes after localization at 2. This job is seen as a sequel to P.J. Hilton's work on the $n+1,n+2$ dimensional homotopy groups of $\mathbf{A}_n^2$-complexes (1950-1951). The main technique used is analysing the homotopy property of $J(X,A)$, defined by B. Gray for a CW-pair $(X,A)$, which is homotopy equivalent to the homotopy fibre of the pinch map $X\cup CA\rightarrow ΣA$. By the way, the results of these homotopy groups have been used to make progress on recent popular topic about the homotopy decomposition of the (multiple) suspension of oriented closed manifolds.

math.AT↗

arXiv2023

Striped Attention: Faster Ring Attention for Causal Transformers

To help address the growing demand for ever-longer sequence lengths in transformer models, Liu et al. recently proposed Ring Attention, an exact attention algorithm capable of overcoming per-device memory bottle- necks by distributing self-attention across multiple devices. In this paper, we study the performance characteristics of Ring Attention in the important special case of causal transformer models, and identify a key workload imbal- ance due to triangular structure of causal attention computations. We propose a simple extension to Ring Attention, which we call Striped Attention to fix this imbalance. Instead of devices having contiguous subsequences, each device has a subset of tokens distributed uniformly throughout the sequence, which we demonstrate leads to more even workloads. In experiments running Striped Attention on A100 GPUs and TPUv4s, we are able to achieve up to 1.45x end-to-end throughput improvements over the original Ring Attention algorithm on causal transformer training at a sequence length of 256k. Furthermore, on 16 TPUv4 chips, we were able to achieve 1.65x speedups at sequence lengths of 786k. We release the code for our experiments as open source

cs.LG↗

arXiv2023

The Cost of Down-Scaling Language Models: Fact Recall Deteriorates before In-Context Learning

How does scaling the number of parameters in large language models (LLMs) affect their core capabilities? We study two natural scaling techniques -- weight pruning and simply training a smaller or larger model, which we refer to as dense scaling -- and their effects on two core capabilities of LLMs: (a) recalling facts presented during pre-training and (b) processing information presented in-context during inference. By curating a suite of tasks that help disentangle these two capabilities, we find a striking difference in how these two abilities evolve due to scaling. Reducing the model size by more than 30\% (via either scaling approach) significantly decreases the ability to recall facts seen in pre-training. Yet, a 60--70\% reduction largely preserves the various ways the model can process in-context information, ranging from retrieving answers from a long context to learning parameterized functions from in-context exemplars. The fact that both dense scaling and weight pruning exhibit this behavior suggests that scaling model size has an inherently disparate effect on fact recall and in-context learning.

cs.CL↗

arXiv2023

Efficient Automatic Scheduling of Imaging and Vision Pipelines for the GPU

We present a new algorithm to quickly generate high-performance GPU implementations of complex imaging and vision pipelines, directly from high-level Halide algorithm code. It is fully automatic, requiring no schedule templates or hand-optimized kernels. We address the scalability challenge of extending search-based automatic scheduling to map large real-world programs to the deep hierarchies of memory and parallelism on GPU architectures in reasonable compile time. We achieve this using (1) a two-phase search algorithm that first 'freezes' decisions for the lowest cost sections of a program, allowing relatively more time to be spent on the important stages, (2) a hierarchical sampling strategy that groups schedules based on their structural similarity, then samples representatives to be evaluated, allowing us to explore a large space with few samples, and (3) memoization of repeated partial schedules, amortizing their cost over all their occurrences. We guide the process with an efficient cost model combining machine learning, program analysis, and GPU architecture knowledge. We evaluate our method's performance on a diverse suite of real-world imaging and vision pipelines. Our scalability optimizations lead to average compile time speedups of 49x (up to 530x). We find schedules that are on average 1.7x faster than existing automatic solutions (up to 5x), and competitive with what the best human experts were able to achieve in an active effort to beat our automatic results.

cs.PL↗

arXiv2022

The Effect of Data Dimensionality on Neural Network Prunability

Practitioners prune neural networks for efficiency gains and generalization improvements, but few scrutinize the factors determining the prunability of a neural network the maximum fraction of weights that pruning can remove without compromising the model's test accuracy. In this work, we study the properties of input data that may contribute to the prunability of a neural network. For high dimensional input data such as images, text, and audio, the manifold hypothesis suggests that these high dimensional inputs approximately lie on or near a significantly lower dimensional manifold. Prior work demonstrates that the underlying low dimensional structure of the input data may affect the sample efficiency of learning. In this paper, we investigate whether the low dimensional structure of the input data affects the prunability of a neural network.

cs.LG↗

arXiv2022

Pruning's Effect on Generalization Through the Lens of Training and Regularization

Practitioners frequently observe that pruning improves model generalization. A long-standing hypothesis based on bias-variance trade-off attributes this generalization improvement to model size reduction. However, recent studies on over-parameterization characterize a new model size regime, in which larger models achieve better generalization. Pruning models in this over-parameterized regime leads to a contradiction -- while theory predicts that reducing model size harms generalization, pruning to a range of sparsities nonetheless improves it. Motivated by this contradiction, we re-examine pruning's effect on generalization empirically. We show that size reduction cannot fully account for the generalization-improving effect of standard pruning algorithms. Instead, we find that pruning leads to better training at specific sparsities, improving the training loss over the dense model. We find that pruning also leads to additional regularization at other sparsities, reducing the accuracy degradation due to noisy examples over the dense model. Pruning extends model training time and reduces model size. These two factors improve training and add regularization respectively. We empirically demonstrate that both factors are essential to fully explaining pruning's impact on generalization.

cs.LG↗

arXiv2020

Compiling ONNX Neural Network Models Using MLIR

Deep neural network models are becoming increasingly popular and have been used in various tasks such as computer vision, speech recognition, and natural language processing. Machine learning models are commonly trained in a resource-rich environment and then deployed in a distinct environment such as high availability machines or edge devices. To assist the portability of models, the open-source community has proposed the Open Neural Network Exchange (ONNX) standard. In this paper, we present a high-level, preliminary report on our onnx-mlir compiler, which generates code for the inference of deep neural network models described in the ONNX format. Onnx-mlir is an open-source compiler implemented using the Multi-Level Intermediate Representation (MLIR) infrastructure recently integrated in the LLVM project. Onnx-mlir relies on the MLIR concept of dialects to implement its functionality. We propose here two new dialects: (1) an ONNX specific dialect that encodes the ONNX standard semantics, and (2) a loop-based dialect to provide for a common lowering point for all ONNX dialect operations. Each intermediate representation facilitates its own characteristic set of graph-level and loop-based optimizations respectively. We illustrate our approach by following several models through the proposed representations and we include some early optimization work and performance results.

cs.PL↗

arXiv2019

Novel Co-variant Feature Point Matching Based on Gaussian Mixture Model

The feature frame is a key idea of feature matching problem between two images. However, most of the traditional matching methods only simply employ the spatial location information (the coordinates), which ignores the shape and orientation information of the local feature. Such additional information can be obtained along with coordinates using general co-variant detectors such as DOG, Hessian, Harris-Affine and MSER. In this paper, we develop a novel method considering all the feature center position coordinates, the local feature shape and orientation information based on Gaussian Mixture Model for co-variant feature matching. We proposed three sub-versions in our method for solving the matching problem in different conditions: rigid, affine and non-rigid, respectively, which all optimized by expectation maximization algorithm. Due to the effective utilization of the additional shape and orientation information, the proposed model can significantly improve the performance in terms of convergence speed and recall. Besides, it is more robust to the outliers.

cs.CV↗

arXiv2019

Cross-correlated Contrast Source Inversion

In this paper, we improved the performance of the contrast source inversion (CSI) method by incorporating a so-called cross-correlated cost functional, which interrelates the state error and the data error in the measurement domain. The proposed method is referred to as the cross-correlated CSI. It enables better robustness and higher inversion accuracy than both the classical CSI and multiplicative regularized CSI (MR-CSI). In addition, we show how the gradient of the modified cost functional can be calculated without significantly increasing the computational burden. The advantages of the proposed algorithms are demonstrated using a 2-D benchmark problem excited by a transverse magnetic wave as well as a transverse electric wave, respectively, in comparison to classical CSI and MR-CSI.

eess.SP↗

arXiv2019

A Linear Method for Shape Reconstruction based on the Generalized Multiple Measurement Vectors Model

In this paper, a novel linear method for shape reconstruction is proposed based on the generalized multiple measurement vectors (GMMV) model. Finite difference frequency domain (FDFD) is applied to discretized Maxwell's equations, and the contrast sources are solved iteratively by exploiting the joint sparsity as a regularized constraint. Cross validation (CV) technique is used to terminate the iterations, such that the required estimation of the noise level is circumvented. The validity is demonstrated with an excitation of transverse magnetic (TM) experimental data, and it is observed that, in the aspect of focusing performance, the GMMV-based linear method outperforms the extensively used linear sampling method (LSM).

eess.SP↗

arXiv2019

Clustering Bioactive Molecules in 3D Chemical Space with Unsupervised Deep Learning

Unsupervised clustering has broad applications in data stratification, pattern investigation and new discovery beyond existing knowledge. In particular, clustering of bioactive molecules facilitates chemical space mapping, structure-activity studies, and drug discovery. These tasks, conventionally conducted by similarity-based methods, are complicated by data complexity and diversity. We ex-plored the superior learning capability of deep autoencoders for unsupervised clustering of 1.39 mil-lion bioactive molecules into band-clusters in a 3-dimensional latent chemical space. These band-clusters, displayed by a space-navigation simulation software, band molecules of selected bioactivity classes into individual band-clusters possessing unique sets of common sub-structural features beyond structural similarity. These sub-structural features form the frameworks of the literature-reported pharmacophores and privileged fragments. Within each band-cluster, molecules are further banded into selected sub-regions with respect to their bioactivity target, sub-structural features and molecular scaffolds. Our method is potentially applicable for big data clustering tasks of different fields.

q-bio.BM↗

arXiv2016

Hybrid Static/Dynamic Schedules for Tiled Polyhedral Programs

Polyhedral compilers perform optimizations such as tiling and parallelization; when doing both, they usually generate code that executes "barrier-synchronized wavefronts" of tiles. We present a system to express and generate code for hybrid schedules, where some constraints are automatically satisfied through the structure of the code, and the remainder are dynamically enforced at run-time with data flow mechanisms. We prove bounds on the added overheads that are better, by at least one polynomial degree, than those of previous techniques. We propose a generic mechanism to implement the needed synchronization, and show it can be easily realized for a variety of targets: OpenMP, Pthreads, GPU (CUDA or OpenCL) code, languages like X10, Habanero, Cilk, as well as data flow platforms like DAGuE, and OpenStream and MPI. We also provide a simple concrete implementation that works without the need of any sophisticated run-time mechanism. Our experiments show our simple implementation to be competitive or better than the wavefront-synchronized code generated by other systems. We also show how the proposed mechanism can achieve 24% to 70% reduction in energy.

cs.PL↗