Search

Extracting implicit knowledge and logical reasoning abilities from large language models (LLMs) has consistently been a significant challenge. The advancement of multi-agent systems has further en-hanced the capabilities of LLMs. Inspired by the structure of multi-polar neurons (MNs), we propose the XAgents framework, an in-terpretable multi-agent cooperative framework based on the IF-THEN rule-based system. The IF-Parts of the rules are responsible for logical reasoning and domain membership calculation, while the THEN-Parts are comprised of domain expert agents that generate domain-specific contents. Following the calculation of the member-ship, XAgetns transmits the task to the disparate domain rules, which subsequently generate the various responses. These re-sponses are analogous to the answers provided by different experts to the same question. The final response is reached at by eliminat-ing the hallucinations and erroneous knowledge of the LLM through membership computation and semantic adversarial genera-tion of the various domain rules. The incorporation of rule-based interpretability serves to bolster user confidence in the XAgents framework. We evaluate the efficacy of XAgents through a com-parative analysis with the latest AutoAgents, in which XAgents demonstrated superior performance across three distinct datasets. We perform post-hoc interpretable studies with SHAP algorithm and case studies, proving the interpretability of XAgent in terms of input-output feature correlation and rule-based semantics.

Generative Models (GMs), particularly Large Language Models (LLMs), have garnered significant attention in machine learning and artificial intelligence for their ability to generate new data by learning the statistical properties of training data and creating data that resemble the original. This capability offers a wide range of applications across various domains. However, the complex structures and numerous model parameters of GMs make the input-output processes opaque, complicating the understanding and control of outputs. Moreover, the purely data-driven learning mechanism limits GM's ability to acquire broader knowledge. There remains substantial potential for enhancing the robustness and generalization capabilities of GMs. In this work, we introduce the fuzzy system, a classical modeling method that combines data and knowledge-driven mechanisms, to generative tasks. We propose a novel Generative Fuzzy System framework, named GenFS, which integrates the deep learning capabilities of GM with the interpretability and dual-driven mechanisms of fuzzy systems. Specifically, we propose an end-to-end GenFS-based model for sequence generation, called FuzzyS2S. A series of experimental studies were conducted on 12 datasets, covering three distinct categories of generative tasks: machine translation, code generation, and summary generation. The results demonstrate that FuzzyS2S outperforms the Transformer in terms of accuracy and fluency. Furthermore, it exhibits better performance on some datasets compared to state-of-the-art models T5 and CodeT5., Comment: 12 pages, 5 figures

Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks but their performance in complex logical reasoning tasks remains unsatisfactory. Although some prompting methods, such as Chain-of-Thought, can improve the reasoning ability of LLMs to some extent, they suffer from an unfaithful issue where derived conclusions may not align with the generated reasoning chain. To address this issue, some studies employ the approach of propositional logic to further enhance logical reasoning abilities of LLMs. However, the potential omissions in the extraction of logical expressions in these methods can cause information loss in the logical reasoning process, thereby generating incorrect results. To this end, we propose Logic-of-Thought (LoT) prompting which employs propositional logic to generate expanded logical information from input context, and utilizes the generated logical information as an additional augmentation to the input prompts, thereby enhancing the capability of logical reasoning. The LoT is orthogonal to existing prompting methods and can be seamlessly integrated with them. Extensive experiments demonstrate that LoT boosts the performance of various prompting methods with a striking margin across five logical reasoning tasks. In particular, the LoT enhances Chain-of-Thought's performance on the ReClor dataset by +4.35%; moreover, it improves Chain-of-Thought with Self-Consistency's performance on LogiQA by +5%; additionally, it boosts performance of Tree-of-Thoughts on ProofWriter dataset by +8%., Comment: 20 pages

In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse NLP tasks. Extensive research has explored how to enhance the logical reasoning abilities such as Chain-of-Thought, Chain-of-Thought with Self-Consistency, Tree-Of-Thoughts, and multi-agent debates. In the context of multi-agent debates, significant performance improvements can be achieved with an increasing number of agents and debate rounds. However, the escalation in the number of agents and debate rounds can drastically raise the tokens cost of debates, thereby limiting the scalability of the multi-agent debate technique. To better harness the advantages of multi-agent debates in logical reasoning tasks, this paper proposes a method to significantly reduce token cost in multi-agent debates. This approach involves dividing all agents into multiple debate groups, with agents engaging in debates within their respective groups and sharing interim debate results between groups. Comparative experiments across multiple datasets have demonstrated that this method can reduce the total tokens by up to 51.7% during debates and while potentially enhancing accuracy by as much as 25%. Our method significantly enhances the performance and efficiency of interactions in the multi-agent debate., Comment: 18 pages

Large language models (LLMs) have emerged as important components across various fields, yet their training requires substantial computation resources and abundant labeled data. It poses a challenge to robustly training LLMs for individual users (clients). To tackle this challenge, the intuitive idea is to introduce federated learning (FL), which can collaboratively train models on distributed private data. However, existing methods suffer from the challenges of data heterogeneity, system heterogeneity, and model size, resulting in suboptimal performance and high costs. In this work, we proposed a variant of personalized federated learning (PFL) framework, namely FDLoRA, which allows the client to be a single device or a cluster and adopts low-rank adaptation (LoRA) tuning. FDLoRA sets dual LoRA modules on each client to capture personalized and global knowledge, respectively, and only the global LoRA module uploads parameters to the central server to aggregate cross-client knowledge. Finally, an adaptive fusion approach is employed to combine the parameters of the dual LoRAs. This enables FDLoRA to make effective use of private data distributed across different clients, thereby improving performance on the client without incurring high communication and computing costs. We conducted extensive experiments in two practice scenarios. The results demonstrate that FDLoRA outperforms six baselines in terms of performance, stability, robustness, computation cost, and communication cost.

As modern HPC computing platforms become increasingly heterogeneous, it is challenging for programmers to fully leverage the computation power of massive parallelism offered by such heterogeneity. Consequently, task-based runtime systems have been proposed as an intermediate layer to hide the complex heterogeneity from the application programmers. The core functionality of these systems is to realize efficient task-to-resource mapping in the form of Directed Acyclic Graph (DAG) scheduling. However, existing scheduling schemes face several drawbacks to determine task priorities due to the heavy reliance on domain knowledge or failure to efficiently exploit the interaction of application and hardware characteristics. In this paper, we propose INSPIRIT, an efficient and lightweight scheduling framework with adaptive priority designed for task-based runtime systems. INSPIRIT introduces two novel task attributes \textit{inspiring ability} and \textit{inspiring efficiency} for dictating scheduling, eliminating the need for application domain knowledge. In addition, INSPIRIT jointly considers runtime information such as ready tasks in worker queues to guide task scheduling. This approach exposes more performance opportunities in heterogeneous hardware at runtime while effectively reducing the overhead for adjusting task priorities. Our evaluation results demonstrate that INSPIRIT achieves superior performance compared to cutting edge scheduling schemes on both synthesized and real-world task DAGs., Comment: 11 pages

Large language models (LLM) have recently attracted surging interest due to their outstanding capabilities across various domains. However, enabling efficient LLM inference is challenging due to its autoregressive decoding that generates tokens only one at a time. Although research works apply pruning or quantization to speed up LLM inference, they typically require fine-tuning the LLM, incurring significant time and economic costs. Meanwhile, speculative decoding has been proposed to use small speculative models (SSMs) to accelerate the inference of LLM. However, the low acceptance rate of SSM and the high verification cost of LLM prohibit further performance improvement of inference. In this paper, we propose Minions, an LLM inference system that accelerates LLM inference with a collective and adaptive speculative generation. Specifically, Minions proposes a majority-voted mechanism to leverage multiple SSMs to jointly speculate the outputs of LLM, which improves the inference performance without introducing prohibitive computation costs for LLM. To better trade off the number of tokens speculated from SSM and the verification cost of LLM, Minions proposes an adaptive mechanism to dynamically determine the optimal speculation length of SSM, which can achieve better inference performance across different models, datasets, and hyper-parameters. In addition, Minions decouples the SSM decoding and LLM verification efficiently and adopts a pipelined execution mechanism to further improve the inference performance of LLM. By comparing with the state-of-the-art LLM inference systems, we demonstrate that Minions can achieve higher inference throughput and lower inference time.

The increasing volume of log data produced by software-intensive systems makes it impractical to analyze them manually. Many deep learning-based methods have been proposed for log-based anomaly detection. These methods face several challenges such as high-dimensional and noisy log data, class imbalance, generalization, and model interpretability. Recently, ChatGPT has shown promising results in various domains. However, there is still a lack of study on the application of ChatGPT for log-based anomaly detection. In this work, we proposed LogGPT, a log-based anomaly detection framework based on ChatGPT. By leveraging the ChatGPT's language interpretation capabilities, LogGPT aims to explore the transferability of knowledge from large-scale corpora to log-based anomaly detection. We conduct experiments to evaluate the performance of LogGPT and compare it with three deep learning-based methods on BGL and Spirit datasets. LogGPT shows promising results and has good interpretability. This study provides preliminary insights into prompt-based models, such as ChatGPT, for the log-based anomaly detection task.

Modern systems produce a large volume of logs to record run-time status and events. System operators use these raw logs to track a system in order to obtain some useful information to diagnose system anomalies. One of the most important problems in this area is to help operators find the answers to log-based questions efficiently and user-friendly. In this work, we propose LogQA, which aims at answering log-based questions in the form of natural language based on large-scale unstructured log corpora. Our system presents the answer to a question directly instead of returning a list of relevant snippets, thus offering better user-friendliness and efficiency. LogQA represents the first approach to solve question answering in lod domain. LogQA has two key components: Log Retriever and Log Reader. Log Retriever aims at retrieving relevant logs w.r.t. a given question, while Log Reader is responsible for inferring the final answer. Given the lack of a public dataset for log questing answering, we manually labelled a QA dataset of three open-source log corpus and will make them publicly available. We evaluated our proposed model on these datasets by comparing its performance with 6 other baseline methods. Our experimental results demonstrate that LogQA has outperformed other baseline methods.

Larger deep learning models usually lead to higher model quality with an ever-increasing GPU memory footprint. Although tensor checkpointing techniques have been proposed to enable training under a restricted GPU memory budget, the input tensor dynamics have been unexploited for optimizing performance while reducing GPU memory footprint. Specifically, due to the diverse datasets and subsequent data argumentation, the input tensor size per mini-batch is dynamic during the training process, leading to a changing GPU memory footprint. However, to leverage such input tensor dynamics in checkpointing, there are two challenges to be solved. First, the checkpointing plan needs to be determined during runtime due to the dynamics of input tensors. Second, the checkpointing plan needs to be applied on the fly without significantly deteriorating the performance. In this paper, we propose Mimose, an input-aware tensor checkpointing planner respecting the memory budget while enabling efficient model training on GPU. Mimose builds a lightweight but accurate prediction model of GPU memory usage online, without pre-analyzing the model. It generates a tensor checkpointing plan based on per-layer memory prediction and applies it to training progress on the fly. It also adopts a caching strategy to avoid having to regenerate the plan for repeated input size. Our experiments show that Mimose achieves superior training throughput compared to state-of-the-art memory planners under the same GPU memory budgets.

Distributed synchronized GPU training is commonly used for deep learning. The resource constraint of using a fixed number of GPUs makes large-scale training jobs suffer from long queuing time for resource allocation, and lowers the cluster utilization. Adapting to resource elasticity can alleviate this but often introduces inconsistent model accuracy, due to lacking of capability to decouple model training procedure from resource allocation. We propose EasyScale, an elastic training system that achieves consistent model accuracy under resource elasticity for both homogeneous and heterogeneous GPUs. EasyScale preserves the data-parallel training behaviors strictly, traces the consistency-relevant factors carefully, utilizes the deep learning characteristics for EasyScaleThread abstraction and fast context-switching. To utilize heterogeneous cluster, EasyScale dynamically assigns workers based on the intra-/inter-job schedulers, minimizing load imbalance and maximizing aggregated job throughput. Deployed in an online serving cluster, EasyScale powers the training jobs to utilize idle GPUs opportunistically, improving overall cluster utilization by 62.1%., Comment: To be appeared at SC'23. Link: https://sc23.supercomputing.org/presentation/?id=pap262&sess=sess168

Deploying various deep learning (DL) models efficiently has boosted the research on DL compilers. The difficulty of generating optimized tensor codes drives DL compiler to ask for the auto-tuning approaches, and the increasing demands require increasing auto-tuning efficiency and quality. Currently, the DL compilers partition the input DL models into several subgraphs and leverage the auto-tuning to find the optimal tensor codes of these subgraphs. However, existing auto-tuning approaches usually regard subgraphs as individual ones and overlook the similarities across them, and thus fail to exploit better tensor codes under limited time budgets. We propose FamilySeer, an auto-tuning framework for DL compilers that can generate better tensor codes even with limited time budgets. FamilySeer exploits the similarities and differences among subgraphs can organize them into subgraph families, where the tuning of one subgraph can also improve other subgraphs within the same family. The cost model of each family gets more purified training samples generated by the family and becomes more accurate so that the costly measurements on real hardware can be replaced with the lightweight estimation through cost model. Our experiments show that FamilySeer can generate model codes with the same code performance more efficiently than state-of-the-art auto-tuning frameworks.

Although the matrix multiplication plays a vital role in computational linear algebra, there are few efficient solutions for matrix multiplication of the near-sparse matrices. The Sparse Approximate Matrix Multiply (SpAMM) is one of the algorithms to fill the performance gap neglected by traditional optimizations for dense/sparse matrix multiplication. However, existing SpAMM algorithms fail to exploit the performance potential of GPUs for acceleration. In this paper, we present cuSpAMM, the first parallel SpAMM algorithm optimized for multiple GPUs. Several performance optimizations have been proposed, including algorithm re-design to adapt to the thread parallelism, blocking strategies for memory access optimization, and the acceleration with the tensor core. In addition, we scale cuSpAMM to run on multiple GPUs with an effective load balance scheme. We evaluate cuSpAMM on both synthesized and real-world datasets on multiple GPUs. The experiment results show that cuSpAMM achieves significant performance speedup compared to vendor optimized cuBLAS and cuSPARSE libraries.