Generate Synthetic Data

ai-evals-course/evals-skills/skills/generate-synthetic-data

by ai-evals-course80d5f7b0127c7572ed9e9339937adbfd7240ffebNo license1.4K starsListed Oct 9, 2026Updated Oct 9, 2026Repository updated 2 weeks ago

Create diverse synthetic test inputs for LLM pipeline evaluation using dimension-based tuple generation. Use when bootstrapping an eval dataset, when real user data is sparse, or when stress-testing specific failure hypotheses. Do NOT use when you already have 100+ representative real traces (use stratified sampling instead), or when the task is collecting production logs.

Only the file list is public. File contents are available once the skill is installed in a workspace.

PathSizeType
agents/openai.yaml258 Bapplication/yaml
SKILL.md5.3 KBtext/markdown

Source and attribution

Source:ai-evals-course/evals-skillsinskills/generate-synthetic-dataat commit80d5f7b

License: No license

Content belongs to its original authors. SourceWeft indexes it from a public repository.

Report or request removal