chief-engineer-field-log
收藏资源简介:
Microfactory Node: 3D打印机现场日志数据集记录了在Microfactory Node实时平台上发生的所有交互事件,包括构建、第二意见、模拟打印、迭代运行和记录结果。该数据集作为节点的现场笔记本,旨在积累和共享3D打印工艺知识,超越单个创客的局限。数据仅包含工作配置和结果,不涉及个人数据或上传的网格文件(仅存储工程师推断的几何类别)。每一行代表一个交互事件,是候选条目,需经人工审核后才能提升为精选课程。数据集采用扁平表结构,每行包含时间戳、事件类型、工作参数(如材料、几何形状、环境温度、湿度、床位置、打印机)、建议设置(如喷嘴温度、床温、风扇百分比、回抽距离)、模型路径(后端和回退使用情况)、风险信息(OBrien标记的风险区域和数量)、检查结果(La Forge的判决和预测一致性)以及结果质量(模拟或实际结果)。不相关字段为空。数据集适用于3D打印过程分析、机器学习模型训练(如LLM交互日志)、增材制造优化和工艺监控等任务。
The Microfactory Node 3D Printer On-Site Log Dataset documents all interactive events occurring on the Microfactory Node real-time platform, including build jobs, second opinion consultations, simulated printing runs, iterative runs, and recorded results. This dataset serves as the on-site notebook for the node, aiming to accumulate and share 3D printing process knowledge and transcend the limitations of individual makers. The dataset only contains job configurations and results, excluding any personal data or uploaded mesh files—only geometric categories inferred by engineers are stored. Each row represents an interactive event as a candidate entry, which requires manual review before being promoted to a curated course. The dataset adopts a flat table structure, with each row containing timestamp, event type, job parameters (e.g., materials, geometry, ambient temperature, humidity, bed position, printer), suggested settings (e.g., nozzle temperature, bed temperature, fan speed percentage, retraction distance), model paths (backend and fallback usage details), risk information (risk areas and their counts marked by O'Brien), inspection results (La Forge's verdict and prediction consistency), and result quality (simulated or actual results). Irrelevant fields are left blank. This dataset is applicable to tasks such as 3D printing process analysis, machine learning model training (e.g., LLM interaction logs), additive manufacturing optimization, and process monitoring.
数据集概览:Microfactory Node: Field Log
数据集名称:Microfactory Node: Field Log(微工厂节点:现场日志)
许可证:MIT
语言:英语
类别:3D打印、增材制造、LLM、交互日志
数据规模:少于10,000行
数据集描述
该数据集记录了与 Microfactory Node 实时交互的每一次操作,包括构建、二次意见、模拟打印、迭代运行和结果记录。它相当于一个现场笔记本,随着使用不断积累工艺经验。日志写入需授权令牌,否则静默跳过。
数据集内容说明
- 包含内容:仅记录作业配置和结果。
- 不包含内容:无个人数据,无上传的网格文件(仅包含工程师推断的几何类别)。
- 数据状态:所有行均为“候选”行,不会自动进入精选数据集 lesson ledger,需人工审核后方可成为“课程”状态。
数据模式(单行扁平结构)
每一行代表一个交互,按 kind 区分,表格结构如下:
| 列名 | 含义 |
|---|---|
ts, kind |
时间戳和交互类型(build, second_opinion, simulate, print_run, record) |
material, geometry, env_temp, env_humidity, bed_position, printer |
作业参数 |
nozzle_temp, bed_temp, fan_pct, retraction_mm, first_layer_fan_pct |
建议的设置参数 |
backend, used_fallback |
用于构建的模型路径(实时Gemma或确定性模型) |
risks, risk_count |
OBrien标记的故障区域 |
inspector_stance, inspector_headline, agreement |
La Forge的评判结果(以及分级打印中预测是否成立) |
outcome, quality |
模拟或真实结果 |
iterations, q_start, q_end, first_clean |
打印循环运行总结 |
与某行 kind 无关的字段为 null。判断由两个命名智能体产出:首席工程师 OBrien 提出建议,La Forge 进行检验。




