thwrhrt/ASM

--- license: bsd-2-clause --- # Low-Level Assembly & Systems Dataset A high-quality instruction-based dataset focused on **assembly programming, system calls, and low-level OS concepts** across multiple CPU architectures. --- ## 📦 Overview This dataset is designed for training and evaluating AI models on **low-level programming tasks**, including: - 🧾 Assembly code generation - ⚙️ System call wrappers - 🧠 Kernel & interrupt concepts - 🧩 Cross-architecture reasoning It follows an **instruction → response** format, making it ideal for instruction-tuned models. --- ## Supported Architectures The dataset includes multiple architectures: - 🟢 AVR (Embedded systems) - 🔵 ARMv8 (Modern mobile & servers) - 🟡 MIPS (Classic Unix systems) - 🟠 RISC-V (Modern open ISA) - 🔴 x86 / x86-64 (Desktop & servers) --- ## Data Format Each entry follows: ```json { "instruction": "Task description", "input": "Optional input context", "output": "Assembly code or explanation" }license: BSD-2-Clause # 底层汇编与系统数据集 本数据集为高质量基于指令的数据集，聚焦多CPU架构下的**汇编编程、系统调用与底层操作系统概念**。 --- ## 📦 概览 本数据集旨在针对**底层编程任务**训练与评估人工智能模型，涵盖以下场景： - 🧾 汇编代码生成 - ⚙️ 系统调用包装器 - 🧠 内核与中断概念 - 🧩 跨架构推理 本数据集采用**指令→响应**的格式，非常适合用于指令微调模型的训练。 --- ## 🛠️ 支持的架构 本数据集涵盖以下多种CPU架构： - 🟢 AVR（嵌入式系统） - 🔵 ARMv8（现代移动与服务器设备） - 🟡 MIPS（经典Unix系统） - 🟠 RISC-V（现代开源指令集架构） - 🔴 x86 / x86-64（桌面与服务器场景） --- ## 📄 数据格式 每条数据条目格式如下： json { "instruction": "任务描述", "input": "可选输入上下文", "output": "汇编代码或相关解释" }