github ConardLi/easy-dataset 1.7.3
[1.7.3] 2026-04-09

17 hours ago

🔧 修复

  1. 修复大量蒸馏数据场景下,数据蒸馏页面渲染卡顿的问题
  2. 修复高并发场景下,全自动蒸馏进度弹框卡顿的问题 #694
  3. 修复使用 Ollama 进行模型评估时,思维链处理异常的问题 #714
  4. 修复自定义模型评估提示词不生效的问题 #687

⚡ 优化

  1. 优化 OpenAI-Compatible API 兼容性问题
  2. 优化模型设置中 Temperature 和 TOP-P 参数的配置问题 #717

✨ 新功能

  1. 新增 MiniMax 作为内置模型提供商支持
  2. 新增主动 COT 合成功能(未提取到原始模型 COT 时自动启用),确保所有数据集均携带思维链 #722

🌍 国际化支持

  1. 新增土耳其语支持,完善大量国际化文案兼容 #702 #706 #708
  2. 新增意大利语支持

🙏 特别感谢


🔧 Fixes

  1. Fixed the rendering lag of the data distillation page when processing a large amount of distillation data
  2. Fixed the lag of the fully automatic distillation progress popup in high-concurrency scenarios #694
  3. Fixed the thought chain processing issue during model evaluation with Ollama #714
  4. Fixed the issue that custom model evaluation prompts did not take effect #687

⚡ Optimizations

  1. Optimized the compatibility of OpenAI-Compatible API
  2. Optimized the configuration of Temperature and TOP-P parameters in model settings #717

✨ New Features

  1. Added MiniMax as a built-in model provider
  2. Added active COT synthesis function (automatically enabled when the original model COT is not extracted), ensuring all datasets are equipped with thought chains #722

🌍 Internationalization

  1. Added Turkish support and improved compatibility with a large number of international copy #702 #706 #708
  2. Added Italian support

🙏 Special Thanks

Don't miss a new easy-dataset release

NewReleases is sending notifications on new releases.