vinn-offline
Routes VINN feature caching and non-interactive k-selection workflows for ACT++ BYOL/ResNet episode features.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- vinn-offline
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Routes VINN feature caching and non-interactive k-selection workflows for ACT++ BYOL/ResNet episode features.
- disable-model-invocation
- true
02 资源文件 2
- 参考文档references/troubleshooting.md1.6 KB
- 参考文档references/workflows.md2.5 KB
03 SKILL.md 正文
vinn-offline
Use this sub-skill when the task is about caching image features for VINN or choosing a k value from cached episode features.
Typical triggers
- "Cache BYOL features for the simulated cube dataset"
- "Choose k for VINN"
- "How are feature files named?"
- "Why does the raw k-selection script stop in IPython?"
What this sub-skill covers
- Feature caching from per-camera ResNet18 checkpoints.
- Feature-file naming and layout for simulated and cotrain variants.
- Offline nearest-neighbor k selection over cached features.
- CUDA requirements and dataset-index assumptions for VINN preprocessing.
What it excludes
- ACT/CNNMLP/Diffusion training and eval -> policy-training.
- Simulation episode generation / replay / visualization -> simulation-data.
- Real-robot VINN deployment -> root troubleshooting only.
Read these first
Run this helper first
Before a long cache or k-selection job, use check_vinn_stack.py to confirm the repo checkout imports and the CUDA backend is visible.
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.