EpiCoder-meta-features

EpiCoder Meta Features This dataset contains the hierarchical meta-feature taxonomy and corresponding frequency statistics used in EpiCoder. These meta-features capture fine-grained code characteristics extracted from real-world repositories and serve as the foundation for controlled, feature-conditioned code generation. Dataset Description The dataset consists of two files: 1. `epicoder_features.json` A hierarchical taxonomy of **17 top-level code feature categories** (stored as a `dict`), each organized into nested subcategories (values are `dict` or `list`). This taxonomy defines the feature space used by EpiCoder to characterize and generate code. Top-level categories:** | Category | Description | |---|---| | `programming language` | Supported programming languages (Python, JavaScript, Java, C++, etc.) | | `implementation style` | Coding paradigms (procedural, object-oriented, functional, modular, etc.) | | `workflow` | Program workflow patterns (argument parsing, initialization, data loading, etc.) | | `functionality` | Functional capabilities (file handling, network communication, UI, etc.) | | `resource usage` | System resource utilization (file system, memory, CPU, network, etc.) | | `computation operation` | Computational operations (mathematical, algorithmic, statistical, etc.) | | `security` | Security-related features (authentication, cryptography, access control, etc.) | | `error handling` | Error handling patterns (exceptions, assertions, validation, etc.) | | `logging` | Logging approaches (debug, error, info logging, etc.) | | `data structures` | Common data structures (list, dict, tuple, class, etc.) | | `implementation logic` | Logic patterns (iterative, conditional, recursive, etc.) | | `user interaction` | User I/O features (input handling, UI components, display, etc.) | | `data processing` | Data operations (preparation, retrieval, transformation, etc.) | | `file operation` | File I/O operations (reading, writing, deletion, etc.) | | `dependenc