Automate end-to-end ML experiment workflows from setup to reporting.
npx skills add https://github.com/Gonglitian/agent-skills --skill auto-experiment
Please help me install this Agent Skill. Skill: auto_experiment Source: https://github.com/Gonglitian/agent-skills/tree/main/skills/auto_experiment Command: npx skills add https://github.com/Gonglitian/agent-skills --skill auto-experiment
Automated end-to-end ML experiment workflows by coordinating setup, planning, execution, evaluation, and reporting.
For example, researchers can run multiple rounds with different hyperparameters, automatically generate baselines, and produce a comprehensive final report.
Provide code path, data path, and workspace, plus an instruction, and the skill will run the full experiment cycle.
High-intent search queries and answers about installing and using this skill.
You can automate end-to-end ML experiment workflows by providing a code path, data path, workspace, and instruction to orchestrate setup, planning, execution, evaluation, and reporting automatically.
Reproducibility for iterative ML experiments is maintained through strict data-symlink handling, branch-based rounds, per-round logging, and automatic sketch updates to ensure full context documentation.
You need a code repository, data path, and workspace to run automated experiments. The workflow uses branch-based rounds and symlinks to keep outputs separate from source data.
The best way to compare baselines across research cycles is using automated per-round logging and evaluation to track different hyperparameters and generate comprehensive final reports.
No, automating your ML workflow will not modify source data because the system uses strict data-symlink handling and keeps all outputs separate from source data to prevent accidental modification.
alpoxdevCreate and manage isolated Git worktree workspaces for parallel development.
tbroadleyExtract samples, messages, events, and metrics from Inspect AI .eval logs.
jwangkunTrack investment theses for A-share positions with structured components and data monitoring.
cookeyholderRun three review-pr workflow rounds on a single pull request with 10-minute pauses.
