mirror of
https://github.com/ruvnet/RuView
synced 2026-06-23 12:33:18 +00:00
ruv
8ff7c2c35a
Add `plugins/ruview` — an end-to-end toolkit for working with RuView (WiFi-DensePose) from Claude Code, mirrored as Codex prompts. Marketplace: `plugins/.claude-plugin/marketplace.json` (one plugin, `ruview`). Skills (9): ruview-quickstart, ruview-hardware-setup, ruview-configure, ruview-applications, ruview-model-training, ruview-advanced-sensing, ruview-cli-api, ruview-mmwave, ruview-verify — shell-first (cargo / python / idf.py / docker / node), no claude-flow MCP dependency. Commands (7): /ruview-start, /ruview-flash, /ruview-provision, /ruview-app, /ruview-train, /ruview-advanced, /ruview-verify. Agents (3): ruview-onboarding-guide, ruview-config-engineer, ruview-training-engineer. Codex mirror: codex/AGENTS.md + codex/README.md + codex/prompts/*.md (full command parity, enforced by scripts/smoke.sh). Docs: docs/adrs/0001-ruview-plugin-contract.md (Proposed). Verification: scripts/smoke.sh (13 structural checks). Provisioning docs reflect the full `provision.py` flag set (TDM mesh, edge tiers, vitals, hop channels, Cognitum Seed, swarm intervals) and the issue #391 NVS-namespace-replace gotcha. Verified: `claude plugin validate` (plugin + marketplace), loads via `claude --plugin-dir`, smoke 13/13, and confirmed against an attached ESP32-S3 on COM8 running the RuView CSI firmware (live adaptive_ctrl + csi_collector serial output). Co-Authored-By: claude-flow <ruv@ruv.net>
15 lines
2.4 KiB
Markdown
15 lines
2.4 KiB
Markdown
# /ruview-train — train a RuView model
|
|
|
|
Train / evaluate / publish a RuView model. Track: `$ARGUMENTS` (one of `camera-free`, `camera-supervised`, `embeddings`, `domain-gen`, `snn`, `gpu`; if empty, ask).
|
|
|
|
- **camera-free** (WiFlow pose, no labels): `cd v2 && cargo run -p wifi-densepose-sensing-server -- --pretrain --dataset data/csi/ --pretrain-epochs 50`, then `-- --train --dataset data/mmfi/ --epochs 100 --save-rvf model.rvf`. ~84 s on M4 Pro, modest accuracy. Bench `node scripts/benchmark-wiflow.js`, eval `node scripts/eval-wiflow.js`.
|
|
- **camera-supervised** (ADR-079, 92.9% PCK@20, ~19 min): `python scripts/collect-ground-truth.py` (MediaPipe landmarks; needs `data/pose_landmarker_lite.task`), `python scripts/collect-training-data.py` (CSI capture), `node scripts/align-ground-truth.js` (timestamp align), then `cd v2 && cargo run -p wifi-densepose-sensing-server -- --train --dataset data/paired/ --epochs <N> --save-rvf model.rvf`, eval `node scripts/eval-wiflow.js` (reports PCK@20).
|
|
- **embeddings** (AETHER ADR-024 / spectrogram ADR-076): `wifi-densepose-train` + `wifi-densepose-ruvector`; `-- --model model.rvf --embed`, `-- --model model.rvf --build-index env`. 171K emb/s on M4 Pro.
|
|
- **domain-gen** (MERIDIAN ADR-027): domain-generalization options in the training pipeline + `ruview_metrics`.
|
|
- **snn** (local env adaptation, <30 s): `node scripts/snn-csi-processor.js --port 5006`; `docs/tutorials/cognitum-seed-pretraining.md`; ADR-084/085 (RaBitQ), ADR-086 (novelty gate).
|
|
- **gpu**: `gcloud auth login && gcloud config set project cognitum-20260110`, then `bash scripts/gcloud-train.sh --dry-run` (smoke), `bash scripts/gcloud-train.sh --gpu l4 --hours 2` (proto, ~$0.80/hr), `bash scripts/gcloud-train.sh --gpu a100 --config scripts/training-config-sweep.json` (~$3.60/hr), `bash scripts/gcloud-train.sh --sweep` (full sweep). VM auto-deletes unless `--keep-vm`. Local Mac: `bash scripts/mac-mini-train.sh`. Bench: `python scripts/benchmark-model.py`.
|
|
|
|
Data: `data/recordings/` raw CSI · `data/csi/` pretrain · `data/mmfi/` MM-Fi · `data/paired/` camera↔CSI · `data/ground-truth/` MediaPipe · `models/` artifacts. Record more: `python scripts/record-csi-udp.py`.
|
|
|
|
After training: `cd v2 && cargo test --workspace --no-default-features`, `cd .. && python archive/v1/data/proof/verify.py` (VERDICT: PASS). Publish: `python scripts/publish-huggingface.py` (or `.sh`; `docs/huggingface/`). Then run `/ruview-verify`.
|