experiment-log-summarizer
Reads machine learning experiment notes and results and writes a clear summary of what worked best. Helps researchers track progress and share findings.
Installation
Paste this into Claude Code, Cursor, or any agent that can run commands.
What this skill does
What it does: - Organizes messy machine learning experiment notes into a clear summary - Sorts through training logs, test results, and parameter changes - Separates what actually happened from what you think might have happened - Identifies which settings worked best - Explains why experiments failed or succeeded - Creates a short summary ready for a weekly report. When to use it: - You ran multiple machine learning experiments and need to understand what worked - You want to compare different settings and their results - You need to figure out why an experiment did not work as expected - You are writing a weekly update and need a summary of your experiments - You have scattered notes and logs that need to be organized into one clear record.
SKILL.mdShow the author's original SKILL.md (not in English)
--- name: experiment-log-summarizer description: summarize machine learning experiment logs in chinese when the input includes training logs, eval results, hyperparameter changes, user notes, or multiple runs and the user needs a grounded experiment summary, error analysis, best configuration recap, or a weekly update ready abstract. --- # Experiment Log Summarizer 用这个 skill 整理实验日志、参数改动、训练结果和失败记录，输出中文实验总结。重点是区分证据与猜测，并把分散实验整理成可复盘的研究记录。 ## 工作流 1. 先把输入按实验轮次、配置、结果和备注拆开。 2. 参考 `references/experiment_template.md` 汇总主要结论。 3. 在需要失败归因或误差分析时，参考 `references/error_analysis_template.md`。 4. 输出完整实验总结，并附周报版摘要。 ## 输入处理规则 - 接收训练日志、eval 结果、参数表、用户备注和多轮实验对比。 - 如果日志不完整，优先整理可确认事实，再列缺口。 - 如果同一实验有多次重复运行，优先总结稳定趋势，不要被单次波动误导。 ## 输出规则 - 默认输出： - 本次实验目标 - 做了哪些改动 - 结果变化 - 可能原因 - 当前最佳配置 - 失败实验总结 - 下一步建议 - 周报版摘要 - “结果变化”只写有数字、日志或明确记录支撑的内容。 - “可能原因”必须明确标为推测，不要伪装成已验证结论。 ## 证据与表述约束 - 明确区分： - `证据`：日志、指标、配置表、用户明确说明 - `推测`：对涨跌原因的解释、潜在 bug 假设、过拟合猜测 - 不要把失败实验简化成“无效”，要指出失败是因为假设错误、实现问题、数据问题还是评测问题。 ## 何时读引用文件 - 始终读取 `references/experiment_template.md`。 - 在需要拆失败原因、错误模式或后续验证动作时读取 `references/error_analysis_template.md`。
Ships with 3 supporting files:
- agents/openai.yaml
- references/error_analysis_template.md
- references/experiment_template.md
Mirrored from the author's public source. Install counts from the open skills registry.