Natural language to code generation is an important application area of LLMs and has received wide attention from the community. The majority of relevant studies have exclusively concentrated on increasing the quantity and functional correctness of training sets while disregarding other stylistic elements of programs. More recently, data quality has garnered a lot of interest and multiple works have showcased its importance for improving performance. In this work, we investigate data quality for code and find that making the code more structured and readable leads to improved code generation performance of the system. We build a novel data-cleaning pipeline that uses these principles to transform existing programs by 1.) renaming variables, 2.) modularizing and decomposing complex code into smaller helper sub-functions, and 3.) inserting natural-language based plans via LLM based transformations. We evaluate our approach on two challenging algorithmic code generation benchmarks and find that fine-tuning CodeLLaMa-7B on our transformed modularized programs improves the performance by up to 30% compared to fine-tuning on the original dataset. Additionally, we demonstrate improved performance from using a smaller amount of higher-quality data, finding that a model fine-tuned on the entire original dataset is outperformed by a model trained on 15% of our cleaned dataset. Even in comparison to closed-source models, our models outperform the much larger AlphaCoder models.
相關內容
We introduce a new task -- language-driven video inpainting, which uses natural language instructions to guide the inpainting process. This approach overcomes the limitations of traditional video inpainting methods that depend on manually labeled binary masks, a process often tedious and labor-intensive. We present the Remove Objects from Videos by Instructions (ROVI) dataset, containing 5,650 videos and 9,091 inpainting results, to support training and evaluation for this task. We also propose a novel diffusion-based language-driven video inpainting framework, the first end-to-end baseline for this task, integrating Multimodal Large Language Models to understand and execute complex language-based inpainting requests effectively. Our comprehensive results showcase the dataset's versatility and the model's effectiveness in various language-instructed inpainting scenarios. We will make datasets, code, and models publicly available.
To enhance the domain-specific capabilities of large language models, continued pre-training on a domain-specific corpus is a prevalent method. Recent work demonstrates that adapting models using reading comprehension data formatted by regex-based patterns can significantly improve performance on domain-specific tasks. However, regex-based patterns are incapable of parsing raw corpora using domain-specific knowledge. Furthermore, the question and answer pairs are extracted directly from the corpus in predefined formats offers limited context. To address this limitation, we improve reading comprehension via LLM and clustering. LLM focuses on leveraging domain knowledge within the corpus to refine comprehension stage, while clustering supplies relevant knowledge by extending the context to enrich reading stage. Additionally, our method incorporates parameter-efficient fine-tuning to improve the efficiency of domain adaptation. In comparison to AdaptLLM, our method achieves an improvement exceeding 5% in domain-specific tasks. Our code will available at //github.com/microsoft/LMOps.
We consider nonlinear eigenvalue problems to compute all eigenvalues in a bounded region on the complex plane. Based on domain decomposition and contour integrals, two robust and scalable parallel multi-step methods are proposed. The first method 1) uses the spectral indicator method to find eigenvalues and 2) calls a linear eigensolver to compute the associated eigenvectors. The second method 1) divides the region into subregions and uses the spectral indicator method to decide candidate regions that contain eigenvalues, 2) computes eigenvalues in each candidate subregion using Beyn's method; and 3) verifies each eigenvalue by substituting it back to the system and computes the smallest eigenvalue. Each step of the two methods is carried out in parallel. Both methods are robust, accurate, and does not require prior knowledge of the number and distribution of the eigenvalues in the region. Examples are presented to show the performance of the two methods.
One of the most promising applications of quantum computers is to simulate quantum mechanical systems and deliver an advantage to classical computation by leveraging their inherent quantum behaviour. In this work, we present a new approach to achieve a noise tolerant Hamiltonian simulation algorithm for ground state energy estimation which also surmounts stochastic limitations most of its counterparts face. This algorithm is based on an adaptive set of fuzzy bisection searches to estimate the ground state energy digit by digit that can get to any arbitrary target precision. It builds upon the Quantum Eigenvalue Transformation of Unitary Matrices (QETU) algorithm and it delivers good approximations in simulations with quantum depolarizing probability up to 1e-3, particularly for the Transverse-Field Ising Model (TFIM). We ran simulations with different system Hamiltonians, system sizes and time evolution encoding methods on IBM Qiskit and we demonstrate the key results in this work, as well as compare the performance with other existing methods.
Illegitimate intelligent reflective surfaces (IRSs) can pose significant physical layer security risks on multi-user multiple-input single-output (MU-MISO) systems. Recently, a DISCO approach has been proposed an illegitimate IRS with random and time-varying reflection coefficients, referred to as a "disco" IRS (DIRS). Such DIRS can attack MU-MISO systems without relying on either jamming power or channel state information (CSI), and classical anti-jamming techniques are ineffective for the DIRS-based fully-passive jammers (DIRS-based FPJs). In this paper, we propose an IRS-enhanced anti-jamming precoder against DIRS-based FPJs that requires only statistical rather than instantaneous CSI of the DIRS-jammed channels. Specifically, a legitimate IRS is introduced to reduce the strength of the DIRS-based jamming relative to the transmit signals at a legitimate user (LU). In addition, the active beamforming at the legitimate access point (AP) is designed to maximize the signal-to-jamming-plus-noise ratios (SJNRs). Numerical results are presented to evaluate the effectiveness of the proposed IRS-enhanced anti-jamming precoder against DIRS-based FPJs.
The emergence of large language models (LLMs) has substantially influenced natural language processing, demonstrating exceptional results across various tasks. In this study, we employ ``Introspective Tips" to facilitate LLMs in self-optimizing their decision-making. By introspectively examining trajectories, LLM refines its policy by generating succinct and valuable tips. Our method enhances the agent's performance in both few-shot and zero-shot learning situations by considering three essential scenarios: learning from the agent's past experiences, integrating expert demonstrations, and generalizing across diverse games. Importantly, we accomplish these improvements without fine-tuning the LLM parameters; rather, we adjust the prompt to generalize insights from the three aforementioned situations. Our framework not only supports but also emphasizes the advantage of employing LLM in in-contxt decision-making. Experiments involving over 100 games in TextWorld illustrate the superior performance of our approach.
The development of autonomous agents which can interact with other agents to accomplish a given task is a core area of research in artificial intelligence and machine learning. Towards this goal, the Autonomous Agents Research Group develops novel machine learning algorithms for autonomous systems control, with a specific focus on deep reinforcement learning and multi-agent reinforcement learning. Research problems include scalable learning of coordinated agent policies and inter-agent communication; reasoning about the behaviours, goals, and composition of other agents from limited observations; and sample-efficient learning based on intrinsic motivation, curriculum learning, causal inference, and representation learning. This article provides a broad overview of the ongoing research portfolio of the group and discusses open problems for future directions.
Learning latent representations of nodes in graphs is an important and ubiquitous task with widespread applications such as link prediction, node classification, and graph visualization. Previous methods on graph representation learning mainly focus on static graphs, however, many real-world graphs are dynamic and evolve over time. In this paper, we present Dynamic Self-Attention Network (DySAT), a novel neural architecture that operates on dynamic graphs and learns node representations that capture both structural properties and temporal evolutionary patterns. Specifically, DySAT computes node representations by jointly employing self-attention layers along two dimensions: structural neighborhood and temporal dynamics. We conduct link prediction experiments on two classes of graphs: communication networks and bipartite rating networks. Our experimental results show that DySAT has a significant performance gain over several different state-of-the-art graph embedding baselines.
Graph neural networks (GNNs) are a popular class of machine learning models whose major advantage is their ability to incorporate a sparse and discrete dependency structure between data points. Unfortunately, GNNs can only be used when such a graph-structure is available. In practice, however, real-world graphs are often noisy and incomplete or might not be available at all. With this work, we propose to jointly learn the graph structure and the parameters of graph convolutional networks (GCNs) by approximately solving a bilevel program that learns a discrete probability distribution on the edges of the graph. This allows one to apply GCNs not only in scenarios where the given graph is incomplete or corrupted but also in those where a graph is not available. We conduct a series of experiments that analyze the behavior of the proposed method and demonstrate that it outperforms related methods by a significant margin.
Multi-relation Question Answering is a challenging task, due to the requirement of elaborated analysis on questions and reasoning over multiple fact triples in knowledge base. In this paper, we present a novel model called Interpretable Reasoning Network that employs an interpretable, hop-by-hop reasoning process for question answering. The model dynamically decides which part of an input question should be analyzed at each hop; predicts a relation that corresponds to the current parsed results; utilizes the predicted relation to update the question representation and the state of the reasoning process; and then drives the next-hop reasoning. Experiments show that our model yields state-of-the-art results on two datasets. More interestingly, the model can offer traceable and observable intermediate predictions for reasoning analysis and failure diagnosis.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191