Home / Skills / bad-case-to-dpo
bad-case-to-dpo
把生产 bad case / 失败轨迹转成训练数据时使用——从失败归因、轨迹前缀截取、DPO 偏好对构造、LoRA 训练到边界集与保留集双集验证的完整链路；也用于判断某类问题该走 SFT、DPO、RL 还是先修 Harness/工具/tokenizer，以及如何从运行轨迹中提取带证据的结构化学习信号。触发词：bad case、失败归因、DPO、偏好对、轨迹前缀、拒绝采样、过度矫正、边界集、保留集、过早结束、reward hacking、LoRA 微调、训练评估隔离。
Why this score
This repo hosts 10 skills, so its 54,962 stars are split between them (about 5,496 each) for scoring.
Popularity score: 88/100 from public GitHub signals only. It shows how widely used and maintained this is, not how well it works. Not included yet: the fake-star check. The security scan is shown separately and doesn't change the score. How we rank
Security
Scanned Oct 20, 2026 at commit dbc046eb896a, with Cisco's open-source skill-scanner.
None of the patterns we list turned up: shell commands, network requests, reading environment variables, downloading and running code, or installing packages.
A scan lists patterns in the code. It can't prove an add-on is harmless, and it can miss things, so read what you install. How we scan
Alternatives
Other makers' add-ons that do the same job, next to this one.
| Name | Score | Stars | Context | Updated |
|---|---|---|---|---|
| This onebad-case-to-dpo bojieli/ai-agent-book · Skill | 88 | ★ 55k | ≈63 always | 27d ago |
| lgpd-dpo-encarregado goul4rt/lgpd-skills · Skill Both about dpo | 47 | ★ 74 | ≈77 always | 5mo ago |
| use-case-triage anthropics/claude-for-legal · Skill Both about case | 81 | ★ 9.7k | ≈24 always | 21d ago |
| case-management DataDog/pup · Subagent Both about case | 77 | ★ 1.0k | ≈28 in chat | 16d ago |
| mcp-case-overrides alibaba/skill-up · Skill Both about case | 69 | ★ 1.2k | ≈22 always | 16d ago |
| clinical-case-report nexu-io/open-design · Skill Both about case | 87 | ★ 103k | ≈23 always | 16d ago |