experiments-analysis
Reuse OASIS legacy experiment patterns, user-generation notes, and post-simulation analysis workflows without assuming unsafe execution.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- experiments-analysis
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Reuse OASIS legacy experiment patterns, user-generation notes, and post-simulation analysis workflows without assuming unsafe execution.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/legacy-experiments.md6.3 KB
- 参考文档references/profile-generation-and-analysis.md4.7 KB
- 参考文档references/troubleshooting.md4.7 KB
03 SKILL.md 正文
Experiments analysis
Use this sub-skill when a request is about legacy OASIS experiment families, generated-user pipelines, or post-simulation visualization and triage.
Start here
- Identify the experiment family and artifact type.
- Read legacy experiments.
- Read profile generation and analysis if the request touches generated profiles, score plots, counterfactual scoring, or Neo4j follow graphs.
- If the task starts from a SQLite database and needs schema or table inspection, hand off first to
platform-actionsDB summary before custom analysis. - If the request is really about profile validation, env lifecycle, or action arguments, route to
agent-profiles,simulation-workflows, orplatform-actions.
What this sub-skill owns
- Mapping legacy experiment folders to their config knobs and result artifacts.
- Explaining how to downscale or adapt a legacy config without changing the scenario unnecessarily.
- Reading generated DB, JSON, CSV, or graph outputs and deciding whether a plot, table, or graph export is the right next step.
- Flagging blocked runs caused by missing credentials, GPU or server pieces, or large-scale cost.
What it does not own
- Root install, import, or API smoke.
- Profile schema validation.
- Environment lifecycle or step-by-step execution.
- Detailed action argument matrices or recsys internals.
Safe defaults
- Treat OpenAI, Hugging Face, VLLM, Slurm, and Neo4j dependencies as conditional.
- Prefer analysis of existing outputs over rerunning simulations.
- For tiny budgets, lower
num_timesteps,round_post_num, andactivate_probbefore changing the scenario.
Deliverables
- Adaptation notes for legacy YAMLs.
- Analysis plans for existing outputs.
- Blocker lists when a requested run is not safe to attempt.
References
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.