Search

Aligning Large Language Models (LLMs) with human intentions and values is crucial yet challenging. Current methods primarily rely on human preferences, which are costly and insufficient in capturing nuanced feedback expressed in natural language. In this paper, we present Self-Refinement Tuning (SRT), a method that leverages model feedback for alignment, thereby reducing reliance on human annotations. SRT uses a base language model (e.g., Tulu2) to generate initial responses, which are critiqued and refined by a more advanced model (e.g., GPT-4-Turbo). This process enables the base model to self-evaluate and improve its outputs, facilitating continuous learning. SRT further optimizes the model by learning from its self-generated feedback and refinements, creating a feedback loop that promotes model improvement. Our empirical evaluations demonstrate that SRT significantly outperforms strong baselines across diverse tasks and model sizes. When applied to a 70B parameter model, SRT increases the win rate from 9.6\% to 25.8\% on the AlpacaEval 2.0 benchmark, surpassing well-established systems such as GPT-4-0314, Claude 2, and Gemini. Our analysis highlights the crucial role of language feedback in the success of SRT, suggesting potential for further exploration in this direction., Comment: Findings of ACL 2024

Large Language Models (LLMs) have achieved impressive performance across various reasoning tasks. However, even state-of-the-art LLMs such as ChatGPT are prone to logical errors during their reasoning processes. Existing solutions, such as deploying task-specific verifiers or voting over multiple reasoning paths, either require extensive human annotations or fail in scenarios with inconsistent responses. To address these challenges, we introduce RankPrompt, a new prompting method that enables LLMs to self-rank their responses without additional resources. RankPrompt breaks down the ranking problem into a series of comparisons among diverse responses, leveraging the inherent capabilities of LLMs to generate chains of comparison as contextual exemplars. Our experiments across 11 arithmetic and commonsense reasoning tasks show that RankPrompt significantly enhances the reasoning performance of ChatGPT and GPT-4, with improvements of up to 13%. Moreover, RankPrompt excels in LLM-based automatic evaluations for open-ended tasks, aligning with human judgments 74% of the time in the AlpacaEval dataset. It also exhibits robustness to variations in response order and consistency. Collectively, our results validate RankPrompt as an effective method for eliciting high-quality feedback from language models., Comment: LREC-Coling 2024 Long Paper

Deep neural network-based Synthetic Aperture Radar (SAR) target recognition models are susceptible to adversarial examples. Current adversarial example generation methods for SAR imagery primarily operate in the 2D digital domain, known as image adversarial examples. Recent work, while considering SAR imaging scatter mechanisms, fails to account for the actual imaging process, rendering attacks in the three-dimensional physical domain infeasible, termed pseudo physics adversarial examples. To address these challenges, this paper proposes SAR-AE-SFP-Attack, a method to generate real physics adversarial examples by altering the scattering feature parameters of target objects. Specifically, we iteratively optimize the coherent energy accumulation of the target echo by perturbing the reflection coefficient and scattering coefficient in the scattering feature parameters of the three-dimensional target object, and obtain the adversarial example after echo signal processing and imaging processing in the RaySAR simulator. Experimental results show that compared to digital adversarial attack methods, SAR-AE-SFP Attack significantly improves attack efficiency on CNN-based models (over 30\%) and Transformer-based models (over 13\%), demonstrating significant transferability of attack effects across different models and perspectives., Comment: 10 pages, 9 figures, 2 tables

GEMM with the small size of input matrices is becoming widely used in many fields like HPC and machine learning. Although many famous BLAS libraries already supported small GEMM, they cannot achieve near-optimal performance. This is because the costs of pack operations are high and frequent boundary processing cannot be neglected. This paper proposes an input-aware adaptive tuning framework(IAAT) for small GEMM to overcome the performance bottlenecks in state-of-the-art implementations. IAAT consists of two stages, the install-time stage and the run-time stage. In the run-time stage, IAAT tiles matrices into blocks to alleviate boundary processing. This stage utilizes an input-aware adaptive tile algorithm and plays the role of runtime tuning. In the install-time stage, IAAT auto-generates hundreds of kernels of different sizes to remove pack operations. Finally, IAAT finishes the computation of small GEMM by invoking different kernels, which corresponds to the size of blocks. The experimental results show that IAAT gains better performance than other BLAS libraries on ARMv8 platform.

In recent years, general matrix-matrix multiplication with non-regular-shaped input matrices has been widely used in many applications like deep learning and has drawn more and more attention. However, conventional implementations are not suited for non-regular-shaped matrix-matrix multiplications, and few works focus on optimizing tall-and-skinny matrix-matrix multiplication on CPUs. This paper proposes an auto-tuning framework, AutoTSMM, to build high-performance tall-and-skinny matrix-matrix multiplication. AutoTSMM selects the optimal inner kernels in the install-time stage and generates an execution plan for the pre-pack tall-and-skinny matrix-matrix multiplication in the runtime stage. Experiments demonstrate that AutoTSMM achieves competitive performance comparing to state-of-the-art tall-and-skinny matrix-matrix multiplication. And, it outperforms all conventional matrix-matrix multiplication implementations., Comment: 8 pages, 12 figures, published in IEEE ISPA 2021

This paper describes the NiuTrans system for the WMT21 translation efficiency task (http://statmt.org/wmt21/efficiency-task.html). Following last year's work, we explore various techniques to improve efficiency while maintaining translation quality. We investigate the combinations of lightweight Transformer architectures and knowledge distillation strategies. Also, we improve the translation efficiency with graph optimization, low precision, dynamic batching, and parallel pre/post-processing. Our system can translate 247,000 words per second on an NVIDIA A100, being 3$\times$ faster than last year's system. Our system is the fastest and has the lowest memory consumption on the GPU-throughput track. The code, model, and pipeline will be available at NiuTrans.NMT (https://github.com/NiuTrans/NiuTrans.NMT)., Comment: NiuTrans at the WMT21 Translation Efficiency Task

The Atmospheric General Circulation Model (AGCM) has been an important research tool in the study of climate change for decades. As the demand for high-resolution simulation is becoming urgent, the scalability and simulation efficiency is faced with great challenges, especially for the latitude-longitude mesh-based models. In this paper, we propose a highly scalable 3D atmospheric general circulation model based on leap-format, namely AGCM-3DLF. Firstly, it utilizes a 3D decomposition method allowing for parallelism release in all three physical dimensions. Then the leap-format difference computation scheme is adopted to maintain computational stability in grid updating and avoid additional filtering at the high latitudes. A novel shifting window communication algorithm is designed for parallelization of the unified model. Furthermore, a series of optimizations are conducted to improve the effectiveness of large-scale simulations. Experiment results in different platforms demonstrate good efficiency and scalability of the model. AGCM-3DLF scales up to the entire CAS-Xiandao1 supercomputer (196,608 CPU cores), attaining the speed of 11.1 simulation-year-per-day (SYPD) at a high resolution of 25KM. In addition, simulations conducted on the Sunway TaihuLight supercomputer exhibit a 1.06 million cores scalability with 36.1% parallel efficiency.

Stencil computation is one of the most important kernels in various scientific and engineering applications. A variety of work has focused on vectorization techniques, aiming at exploiting the in-core data parallelism. Briefly, they either incur data alignment conflicts or hurt the data locality when integrated with tiling. In this paper, a novel transpose layout is devised to preserve the data locality for tiling in the data space and reduce the data reorganization overhead for vectorization simultaneously. We then propose an approach of temporal computation folding designed to further reduce the redundancy of arithmetic calculations by exploiting the register reuse, alleviating the increased register pressure, and deducing generalization with a linear regression model. Experimental results on the AVX-2 and AVX-512 CPUs show that our approach obtains a competitive performance., Comment: arXiv admin note: substantial text overlap with arXiv:2103.08825

Stream applications are widely deployed on the cloud. While modern distributed streaming systems like Flink and Spark Streaming can schedule and execute them efficiently, streaming dataflows are often dynamically changing, which may cause computation imbalance and backpressure. We introduce AutoFlow, an automatic, hotspot-aware dynamic load balance system for streaming dataflows. It incorporates a centralized scheduler which monitors the load balance in the entire dataflow dynamically and implements state migrations correspondingly. The scheduler achieves these two tasks using a simple asynchronous distributed control message mechanism and a hotspot-diminishing algorithm. The timing mechanism supports implicit barriers and a highly efficient state-migration without global barriers or pauses to operators. It also supports a time-window based load-balance measurement and feeds them to the hotspot-diminishing algorithm without user interference. We implemented AutoFlow on top of Ray, an actor-based distributed execution framework. Our evaluation based on various streaming benchmark dataset shows that AutoFlow achieves good load-balance and incurs a low latency overhead in highly data-skew workload.

Stencil computation is one of the most important kernels in various scientific and engineering applications. A variety of work has focused on vectorization and tiling techniques, aiming at exploiting the in-core data parallelism and data locality respectively. In this paper, the downsides of existing vectorization schemes are analyzed. Briefly, they either incur data alignment conflicts or hurt the data locality when integrated with tiling. Then we propose a novel transpose layout to preserve the data locality for tiling and reduce the data reorganization overhead for vectorization simultaneously. To further improve the data reuse at the register level, a time loop unroll-and-jam strategy is designed to perform multistep stencil computation along the time dimension. Experimental results on the AVX-2 and AVX-512 CPUs show that our approach obtains a competitive performance.

Stencil computations represent a very common class of nested loops in scientific and engineering applications. Exploiting vector units in modern CPUs is crucial to achieving peak performance. Previous vectorization approaches often consider the data space, in particular the innermost unit-strided loop. It leads to the well-known data alignment conflict problem that vector loads are overlapped due to the data sharing between continuous stencil computations. This paper proposes a novel temporal vectorization scheme for stencils. It vectorizes the stencil computation in the iteration space and assembles points with different time coordinates in one vector. The temporal vectorization leads to a small fixed number of vector reorganizations that is irrelevant to the vector length, stencil order, and dimension. Furthermore, it is also applicable to Gauss-Seidel stencils, whose vectorization is not well-studied. The effectiveness of the temporal vectorization is demonstrated by various Jacobi and Gauss-Seidel stencils.