In an era of widespread influence of Natural Language Processing (NLP), there have been multiple research efforts to supplant traditional manual coding techniques with automated systems capable of generating solutions autonomously. With rapid research for code generation and a sole focus on large language models, there emerges a need to compare and evaluate the performance of transformer architectures based on several complexities of the model. This paper introduces the concept of a "A Comparative Study on Code Generation with Transformers," a model based on Transformer architecture, and NLP methodologies to automatically generate C++ source code for different varieties of problems. Here, a comparative study is performed to evaluate the robustness of transformer-based models on the basis of their architecture complexities and their capability to handle diverse problem sets, from basic arithmetic to complex computations.
As the scale and complexity of spatiotemporal data continue to grow rapidly, the use of geospatial modeling on the Google Earth Engine (GEE) platform presents dual challenges: improving the coding efficiency of domain experts and enhancing the coding capabilities of interdisciplinary users. To address these challenges and improve the performance of large language models (LLMs) in geospatial code generation tasks, we propose a framework for building a geospatial operator knowledge base tailored to the GEE JavaScript API. This framework consists of an operator syntax knowledge table, an operator relationship frequency table, an operator frequent pattern knowledge table, and an operator relationship chain knowledge table. By leveraging Abstract Syntax Tree (AST) techniques and frequent itemset mining, we systematically extract operator knowledge from 185,236 real GEE scripts and syntax documentation, forming a structured knowledge base. Experimental results demonstrate that the framework achieves over 90% accuracy, recall, and F1 score in operator knowledge extraction. When integrated with the Retrieval-Augmented Generation (RAG) strategy for LLM-based geospatial code generation tasks, the knowledge base improves performance by 20-30%. Ablation studies further quantify the necessity of each knowledge table in the knowledge base construction. This work provides robust support for the advancement and application of geospatial code modeling techniques, offering an innovative approach to constructing domain-specific knowledge bases that enhance the code generation capabilities of LLMs, and fostering the deeper integration of generative AI technologies within the field of geoinformatics.
Automated code generation with large language models has gained significant traction, but there remains no guarantee on the correctness of generated code. We aim to use formal verification to provide mathematical guarantees that the generated code is correct. However, generating formally verified code with LLMs is hindered by the scarcity of training data and the complexity of formal proofs. To tackle this challenge, we introduce AlphaVerus, a self-improving framework that bootstraps formally verified code generation by iteratively translating programs from a higher-resource language and leveraging feedback from a verifier. AlphaVerus operates in three phases: exploration of candidate translations, Treefinement -- a novel tree search algorithm for program refinement using verifier feedback, and filtering misaligned specifications and programs to prevent reward hacking. Through this iterative process, AlphaVerus enables a LLaMA-3.1-70B model to generate verified code without human intervention or model finetuning. AlphaVerus shows an ability to generate formally verified solutions for HumanEval and MBPP, laying the groundwork for truly trustworthy code-generation agents.
Code large language models (codeLLMs) have made significant strides in code generation. Most previous code-related benchmarks, which consist of various programming exercises along with the corresponding test cases, are used as a common measure to evaluate the performance and capabilities of code LLMs. However, the current code LLMs focus on synthesizing the correct code snippet, ignoring the alignment with human preferences, where the query should be sampled from the practical application scenarios and the model-generated responses should satisfy the human preference. To bridge the gap between the model-generated response and human preference, we present a rigorous human-curated benchmark CodeArena to emulate the complexity and diversity of real-world coding tasks, where 397 high-quality samples spanning 40 categories and 44 programming languages, carefully curated from user queries. Further, we propose a diverse synthetic instruction corpus SynCode-Instruct (nearly 20B tokens) by scaling instructions from the website to verify the effectiveness of the large-scale synthetic instruction fine-tuning, where Qwen2.5-SynCoder totally trained on synthetic instruction data can achieve top-tier performance of open-source code LLMs. The results find performance differences between execution-based benchmarks and CodeArena. Our systematic experiments of CodeArena on 40+ LLMs reveal a notable performance gap between open SOTA code LLMs (e.g. Qwen2.5-Coder) and proprietary LLMs (e.g., OpenAI o1), underscoring the importance of the human preference alignment.\footnote{\url{https://codearenaeval.github.io/ }}
As deep learning models scale, their training cost has surged significantly. Due to both hardware advancements and limitations in current software stacks, the need for data efficiency has risen. Data efficiency refers to the effective hiding of data access latency and the avoidance of unnecessary data movements. Major challenges arise from the growing disparity between GPU memory bandwidth and computational throughput, imminent GPU memory capacity limitations, and inefficiencies in the PyTorch software stack, including a lack of device-specific PCIe transfer optimizations and high-level domain-specific abstractions. To effectively mitigate these data inefficiencies for deep learning training, this dissertation analyzes data inefficiency in representative deep training tasks, specifically in graph neural networks (GNNs) and large language models (LLMs). It then proposes novel runtime and code generation techniques to mitigate these challenges and implements these optimizations seamlessly within the PyTorch stack while maintaining strong programmability and interoperability. First, PyTorch-Direct is devised to incorporate the GPU-centric PCIe data transfer paradigm in PyTorch for GNN training. Next, Hector intermediate representation (IR) and its code generator are proposed to introduce domain-specific high-level abstraction and systematically address memory-intensive performance challenges for relational GNNs. Finally, in LLM training, the throughput has been increasingly constrained by GPU memory capacity. To mitigate this, the SSDTrain offloading framework is designed and implemented. Together, these contributions show that code generation and runtime techniques can systematically mitigate the data management bottlenecks in deep learning training, which stem from the data-intensive nature of workloads and the oversimplification inherent in the deep learning training software stack.
Large Language Models (LLMs) have become increasingly popular for generating RTL code. However, producing error-free RTL code in a zero-shot setting remains highly challenging for even state-of-the-art LLMs, often leading to issues that require manual, iterative refinement. This additional debugging process can dramatically increase the verification workload, underscoring the need for robust, automated correction mechanisms to ensure code correctness from the start. In this work, we introduce AIvril2, a self-verifying, LLM-agnostic agentic framework aimed at enhancing RTL code generation through iterative corrections of both syntax and functional errors. Our approach leverages a collaborative multi-agent system that incorporates feedback from error logs generated by EDA tools to automatically identify and resolve design flaws. Experimental results, conducted on the VerilogEval-Human benchmark suite, demonstrate that our framework significantly improves code quality, achieving nearly a 3.4$\times$ enhancement over prior methods. In the best-case scenario, functional pass rates of 77% for Verilog and 66% for VHDL were obtained, thus substantially improving the reliability of LLM-driven RTL code generation.
Recent advancements in code completion models have primarily focused on local file contexts. However, these studies do not fully capture the complexity of real-world software development, which often requires the use of rapidly-evolving public libraries. To fill the gap, we introduce LibEvolutionEval, a detailed study requiring an understanding of library evolution to perform in-line code completion accurately. LibEvolutionEval provides a version-specific code-completion task comprised of eight libraries (torch, torchvision, scipy, pil, tqdm, pyyaml, matplotlib, and pandas) as they evolve over the year along with a detailed analysis of the evolution of two popular and well-maintained public libraries: PyTorch and Matplotlib. We evaluate popular public models and find that public library evolution significantly influences model performance. We explored mitigation methods by studying how retrieved version-specific library documentation and prompting can improve the model's capability in handling these fast-evolving packages, paving a promising future path in better handling fast-evolving libraries.
Multimodal AI has the potential to significantly enhance document-understanding tasks, such as processing receipts, understanding workflows, extracting data from documents, and summarizing reports. Code generation tasks that require long-structured outputs can also be enhanced by multimodality. Despite this, their use in commercial applications is often limited due to limited access to training data and restrictive licensing, which hinders open access. To address these limitations, we introduce BigDocs-7.5M, a high-quality, open-access dataset comprising 7.5 million multimodal documents across 30 tasks. We use an efficient data curation process to ensure our data is high-quality and license-permissive. Our process emphasizes accountability, responsibility, and transparency through filtering rules, traceable metadata, and careful content analysis. Additionally, we introduce BigDocs-Bench, a benchmark suite with 10 novel tasks where we create datasets that reflect real-world use cases involving reasoning over Graphical User Interfaces (GUI) and code generation from images. Our experiments show that training with BigDocs-Bench improves average performance up to 25.8% over closed-source GPT-4o in document reasoning and structured output tasks such as Screenshot2HTML or Image2Latex generation. Finally, human evaluations showed a preference for outputs from models trained on BigDocs over GPT-4o. This suggests that BigDocs can help both academics and the open-source community utilize and improve AI tools to enhance multimodal capabilities and document reasoning. The project is hosted at https://bigdocs.github.io .
This paper presents HyperGraphOS, an innovative Operating System designed for the scientific and engineering domains. It combines model based engineering, graph modeling, data containers, and computational tools, offering users a dynamic workspace for creating and managing complex models represented as customizable graphs. Using a web based architecture, HyperGraphOS requires only a modern browser to organize knowledge, documents, and content into interconnected models. Domain Specific Languages drive workspace navigation, code generation, AI integration, and process organization.The platform models function as both visual drawings and data structures, enabling dynamic modifications and inspection, both interactively and programmatically. HyperGraphOS was evaluated across various domains, including virtual avatars, robotic task planning using Large Language Models, and meta modeling for feature based code development. Results show significant improvements in flexibility, data management, computation, and document handling.
We introduce a novel method to enhance cross-language code translation from Fortran to C++ by integrating task-specific embedding alignment into a Retrieval-Augmented Generation (RAG) framework. Unlike conventional retrieval approaches that utilize generic embeddings agnostic to the downstream task, our strategy aligns the retrieval model directly with the objective of maximizing translation quality, as quantified by the CodeBLEU metric. This alignment ensures that the embeddings are semantically and syntactically meaningful for the specific code translation task. Our methodology involves constructing a dataset of 25,000 Fortran code snippets sourced from Stack-V2 dataset and generating their corresponding C++ translations using the LLaMA 3.1-8B language model. We compute pairwise CodeBLEU scores between the generated translations and ground truth examples to capture fine-grained similarities. These scores serve as supervision signals in a contrastive learning framework, where we optimize the embedding model to retrieve Fortran-C++ pairs that are most beneficial for improving the language model's translation performance. By integrating these CodeBLEU-optimized embeddings into the RAG framework, our approach significantly enhances both retrieval accuracy and code generation quality over methods employing generic embeddings. On the HPC Fortran2C++ dataset, our method elevates the average CodeBLEU score from 0.64 to 0.73, achieving a 14% relative improvement. On the Numerical Recipes dataset, we observe an increase from 0.52 to 0.60, marking a 15% relative improvement. Importantly, these gains are realized without any fine-tuning of the language model, underscoring the efficiency and practicality of our approach.
In recent years, table reasoning has garnered substantial research interest, particularly its integration with Large Language Models (LLMs) which revolutionize natural language applications. Existing typical LLM-based studies realize step-by-step reasoning, promoting the capabilities in table understanding and analyzing. While these approaches emphasize autonomous exploration to accomplish the task objective, they overlook systematic thinking in the reasoning process, leading to potential risks of omitted steps, disorganized logic and misleading results. In this paper, we propose PoTable, a novel stage-oriented plan-then-execute reasoning approach that achieves systematic thinking on tables. Specifically, PoTable deploys several distinct tabular analytical stages with clear objectives and achieves stage-by-stage reasoning. To accomplish the stage-specific goal, PoTable conducts plan-then-execute reasoning, which first plans the operation chain under the stage objective, and then executes each operation sequentially through code generation, real-time running and feedback processing. As a result, PoTable can produce reliable table reasoning results with highly accurate, steply commented and completely executable programs. It possesses a high degree of alignment with a distinguished tabular data analyst, offering advantages of high accuracy and explainability. Finally, we conduct extensive experiments over four evaluation datasets from WikiTQ and TabFact benchmarks, where the results demonstrate the effectiveness of PoTable, as well as the efficiency and explainability.
Model merging has shown great promise at combining expert models, but the benefit of merging is unclear when merging ``generalist'' models trained on many tasks. We explore merging in the context of large (~100B) models, by recycling checkpoints that exhibit tradeoffs among different tasks. Such checkpoints are often created in the process of developing a frontier model, and many suboptimal ones are usually discarded. Given a pool of model checkpoints obtained from different training runs (e.g., different stages, objectives, hyperparameters, and data mixtures), which naturally show tradeoffs across different language capabilities (e.g., instruction following vs. code generation), we investigate whether merging can recycle such suboptimal models into a Pareto-optimal one. Our optimization algorithm tunes the weight of each checkpoint in a linear combination, resulting in a Pareto-optimal models that outperforms both individual models and merge-based baselines. Further analysis shows that good merges tend to include almost all checkpoints with non-zero weights, indicating that even seemingly bad initial checkpoints can contribute to good final merges.
Transforming mathematical expressions into LaTeX poses a significant challenge. In this paper, we examine the application of advanced transformer-based architectures to address the task of converting handwritten or digital mathematical expression images into corresponding LaTeX code. As a baseline, we utilize the current state-of-the-art CNN encoder and LSTM decoder. Additionally, we explore enhancements to the CNN-RNN architecture by replacing the CNN encoder with the pretrained ResNet50 model with modification to suite the grey scale input. Further, we experiment with vision transformer model and compare with Baseline and CNN-LSTM model. Our findings reveal that the vision transformer architectures outperform the baseline CNN-RNN framework, delivering higher overall accuracy and BLEU scores while achieving lower Levenshtein distances. Moreover, these results highlight the potential for further improvement through fine-tuning of model parameters. To encourage open research, we also provide the model implementation, enabling reproduction of our results and facilitating further research in this domain.
Large Language Models (LLMs) represent a class of deep learning models adept at understanding natural language and generating coherent responses to various prompts or queries. These models far exceed the complexity of conventional neural networks, often encompassing dozens of neural network layers and containing billions to trillions of parameters. They are typically trained on vast datasets, utilizing architectures based on transformer blocks. Present-day LLMs are multi-functional, capable of performing a range of tasks from text generation and language translation to question answering, as well as code generation and analysis. An advanced subset of these models, known as Multimodal Large Language Models (MLLMs), extends LLM capabilities to process and interpret multiple data modalities, including images, audio, and video. This enhancement empowers MLLMs with capabilities like video editing, image comprehension, and captioning for visual content. This survey provides a comprehensive overview of the recent advancements in LLMs. We begin by tracing the evolution of LLMs and subsequently delve into the advent and nuances of MLLMs. We analyze emerging state-of-the-art MLLMs, exploring their technical features, strengths, and limitations. Additionally, we present a comparative analysis of these models and discuss their challenges, potential limitations, and prospects for future development.
Large language models (LLMs) have made significant strides at code generation through improved model design, training, and chain-of-thought. However, prompt-level optimizations remain an important yet under-explored aspect of LLMs for coding. This work focuses on the few-shot examples present in most code generation prompts, offering a systematic study on whether few-shot examples improve LLM's coding capabilities, which few-shot examples have the largest impact, and how to select impactful examples. Our work offers 2 approaches for selecting few-shot examples, a model-free method, CODEEXEMPLAR-FREE, and a model-based method, CODEEXEMPLAR-BASED. The 2 methods offer a trade-off between improved performance and reliance on training data and interpretability. Both methods significantly improve CodeLlama's coding ability across the popular HumanEval+ coding benchmark. In summary, our work provides valuable insights into how to pick few-shot examples in code generation prompts to improve LLM code generation capabilities.
Effective code retrieval plays a crucial role in advancing code generation, bug fixing, and software maintenance, particularly as software systems increase in complexity. While current code embedding models have demonstrated promise in retrieving code snippets for small-scale, well-defined tasks, they often underperform in more demanding real-world applications such as bug localization within GitHub repositories. We hypothesize that a key issue is their reliance on noisy and inconsistent datasets for training, which impedes their ability to generalize to more complex retrieval scenarios. To address these limitations, we introduce CoRNStack, a large-scale, high-quality contrastive training dataset for code that spans multiple programming languages. This dataset is curated using consistency filtering to eliminate noisy positives and is further enriched with mined hard negatives, thereby facilitating more effective learning. We demonstrate that contrastive training of embedding models using CoRNStack leads to state-of-the-art performance across a variety of code retrieval tasks. Furthermore, the dataset can be leveraged for training code reranking models, a largely underexplored area compared to text reranking. Our finetuned code reranking model significantly improves the ranking quality over the retrieved results. Finally, by employing our code retriever and reranker together, we demonstrate significant improvements in function localization for GitHub issues, an important component of real-world software development.
With the goal of benchmarking generative systems beyond expert software development ability, we introduce Commit0, a benchmark that challenges AI agents to write libraries from scratch. Agents are provided with a specification document outlining the library's API as well as a suite of interactive unit tests, with the goal of producing an implementation of this API accordingly. The implementation is validated through running these unit tests. As a benchmark, Commit0 is designed to move beyond static one-shot code generation towards agents that must process long-form natural language specifications, adapt to multi-stage feedback, and generate code with complex dependencies. Commit0 also offers an interactive environment where models receive static analysis and execution feedback on the code they generate. Our experiments demonstrate that while current agents can pass some unit tests, none can yet fully reproduce full libraries. Results also show that interactive feedback is quite useful for models to generate code that passes more unit tests, validating the benchmarks that facilitate its use.
Producing emotionally dynamic 3D facial avatars with text derived from spoken words (Emo3D) has been a pivotal research topic in 3D avatar generation. While progress has been made in general-purpose 3D avatar generation, the exploration of generating emotional 3D avatars remains scarce, primarily due to the complexities of identifying and rendering rich emotions from spoken words. This paper reexamines Emo3D generation and draws inspiration from human processes, breaking down Emo3D into two cascading steps: Text-to-3D Expression Mapping (T3DEM) and 3D Avatar Rendering (3DAR). T3DEM is the most crucial step in determining the quality of Emo3D generation and encompasses three key challenges: Expression Diversity, Emotion-Content Consistency, and Expression Fluidity. To address these challenges, we introduce a novel benchmark to advance research in Emo3D generation. First, we present EmoAva, a large-scale, high-quality dataset for T3DEM, comprising 15,000 text-to-3D expression mappings that characterize the aforementioned three challenges in Emo3D generation. Furthermore, we develop various metrics to effectively evaluate models against these identified challenges. Next, to effectively model the consistency, diversity, and fluidity of human expressions in the T3DEM step, we propose the Continuous Text-to-Expression Generator, which employs an autoregressive Conditional Variational Autoencoder for expression code generation, enhanced with Latent Temporal Attention and Expression-wise Attention mechanisms. Finally, to further enhance the 3DAR step on rendering higher-quality subtle expressions, we present the Globally-informed Gaussian Avatar (GiGA) model. GiGA incorporates a global information mechanism into 3D Gaussian representations, enabling the capture of subtle micro-expressions and seamless transitions between emotional states.
Programmable Logic Controllers (PLCs) are microcomputers essential for automating factory operations. Structured Text (ST), a high-level language adhering to the IEC 61131-3 standard, is pivotal for PLCs due to its ability to express logic succinctly and to seamlessly integrate with other languages within the same standard. However, vendors develop their own customized versions of ST, and the lack of comprehensive and standardized documentation for the full semantics of ST has contributed to inconsistencies in how the language is implemented. Consequently, the steep learning curve associated with ST, combined with ever-evolving industrial requirements, presents significant challenges for developers. In response to these issues, we present AutoPLC, an LLM-based approach designed to automate the generation of vendor-specific ST code. To facilitate effective code generation, we first built a comprehensive knowledge base, including Rq2ST Case Library (requirements and corresponding implementations) and Instruction libraries. Then we developed a retrieval module to incorporate the domain-specific knowledge by identifying pertinent cases and instructions, guiding the LLM to generate code that meets the requirements. In order to verify and improve the quality of the generated code, we designed an adaptable code checker. If errors are detected, we initiate an iterative self-improvement process to instruct the LLM to revise the generated code. We evaluate AutoPLC's performance against seven state-of-the-art baselines using three benchmarks, one for open-source basic ST and two for commercial Structured Control Language (SCL) from Siemens. The results show that our approach consistently achieves superior performance across all benchmarks. Ablation study emphasizes the significance of our modules. Further manual analysis confirm the practical utility of the ST code generated by AutoPLC.
Medical vision-language model models often struggle with generating accurate quantitative measurements in radiology reports, leading to hallucinations that undermine clinical reliability. We introduce FactCheXcker, a modular framework that de-hallucinates radiology report measurements by leveraging an improved query-code-update paradigm. Specifically, FactCheXcker employs specialized modules and the code generation capabilities of large language models to solve measurement queries generated based on the original report. After extracting measurable findings, the results are incorporated into an updated report. We evaluate FactCheXcker on endotracheal tube placement, which accounts for an average of 78% of report measurements, using the MIMIC-CXR dataset and 11 medical report-generation models. Our results show that FactCheXcker significantly reduces hallucinations, improves measurement precision, and maintains the quality of the original reports. Specifically, FactCheXcker improves the performance of all 11 models and achieves an average improvement of 94.0% in reducing measurement hallucinations measured by mean absolute error.