ConardLi/easy-dataset 1.3.5

[1.3.5] 2025-05-21

on GitHub

如果遇到 Github 下载慢的问题可以使用网盘下载：https://pan.quark.cn/s/194b7eedf16e

🔧 修复

1. 数据集确认/保存失败
   → 修复因权限校验异常或网络波动导致的数据集保存失败问题，提升操作稳定性。
2. 修改文本块后筛选条件失效
   → 解决文本块内容更新后，筛选条件（如标签、状态）未同步刷新的问题。
3. 硅基流动默认 API 错误
   → 修正默认配置中硅基流动 API 地址及认证参数，确保模型调用正常。
4. 导出自定义格式数据集丢失标签
   → 恢复自定义格式导出时标签字段的正常提取，支持保留完整元数据。

⚡ 优化

1. Windows 安装路径自定义
   → 安装程序新增路径选择功能，默认不再强制安装至 C 盘，支持用户指定安装目录。
2. Alpaca 数据集导出配置优化
   • 字段选择：支持切换问题使用 instruction 或 input 字段，适配不同模型训练需求。
   • 自定义指令：允许手动输入或修改 instruction 内容，提升数据生成灵活性。

🔧 Fixes

1. Dataset confirmation/saving failures
   → Fixed issues with dataset saving due to permission errors or network fluctuations, improving operational stability.
2. Filter criteria失效 after text block modification
   → Resolved synchronization issues where filter conditions (e.g., labels, status) failed to update after text block edits.
3. Default API error for SiliconFlow
   → Corrected the default API endpoint and authentication parameters for SiliconFlow to ensure proper model invocation.
4. Missing labels in custom-format dataset exports
   → Restored label fields in custom exports to preserve complete metadata during data export.

⚡ Optimizations

1. Windows installation path customization
   → Added a path selection feature during installation, allowing users to specify a directory instead of forcing C:\ by default.
2. Alpaca dataset export configuration
   • Field selection: Supported switching between instruction and input fields for questions, adapting to different model training needs.
   • Custom instruction: Allowed manual input or modification of instruction content for more flexible data generation.