🎁 Get the FREE AI Skills Starter Guide β€” Subscribe β†’
BytesAgainBytesAgain
BytesAgain β€Ί Skills β€Ί Lora Pipeline
πŸ¦€ ClawHub

Lora Pipeline

by @iskwang

Manages end-to-end LoRA training: collects and verifies photos, scrapes datasets, applies quality checks, captions, and trains the LoRA model locally.

Versionv1.0.0
Downloads930
Stars⭐ 1
#devops#data#image-gen#legal#gaming
TERMINAL
clawhub install lora-pipeline

πŸ“– About This Skill

name: lora-pipeline description: End-to-end LoRA training pipeline: reference photo collection β†’ face verification β†’ dataset scraping β†’ quality check β†’ WD14 captioning β†’ RunPod training. Use this skill whenever a user asks to build a training dataset, collect photos for a LoRA, or train a LoRA model.

LoRA Pipeline

Orchestrates the full LoRA dataset-to-model pipeline. Each phase is self-contained and can be delegated to a sub-agent independently.

Pipeline Overview

Phase 1: 蒐集範例照片   β†’ collect 3–6 reference face photos
Phase 2: η’ΊθͺδΊΊθ‡‰ζ­£η’Ί   β†’ user confirms refs; deepface cross-check
Phase 3: 蒐集 datasets  β†’ scrape web sources guided by face features
Phase 4: η’Ίθͺη…§η‰‡ζ­£η’Ί   β†’ face verify + dedup + quality filter + crop
Phase 5: 開始 caption   β†’ WD14 local tagging + trigger word
Phase 6: LoRA training  β†’ RunPod Kohya training β†’ retrieve outputs

Phase Index

| Phase | File | Can Sub-Agent | Model | Est. Time | |-------|------|:---:|---|---| | 01 β€” Reference Collection | phases/01-reference.md | βœ… | Haiku (Worker) | 5–10 min | | 02 β€” Scraping | phases/02-scraping.md | βœ… | Haiku (Worker) | 10–30 min | | 03 β€” Verify & Clean | phases/03-verify.md | βœ… | Haiku (Worker) | 2–5 min | | 04 β€” Caption | phases/04-caption.md | βœ… | Haiku (Worker) | 1–3 min | | 05 β€” Training | phases/05-training.md | βœ… | Haiku (Worker) + Sentry | 15–30 min |

To load a specific phase: read skills/lora-pipeline/phases/ β€” each file is independently readable.

Directory Structure

~/.openclaw/workspace/
└── datasets/
    β”œβ”€β”€ face_references/
    β”‚   └── /          # Phase 1–2: Gold standard refs (3–6 photos)
    β”‚       β”œβ”€β”€ ref_01.jpg
    β”‚       └── ...
    β”œβ”€β”€ _raw/          # Phase 3: Raw scraped images (pre-verification)
    β”‚   └── ...
    └── /              # Phase 4–5: Verified + captioned training set
        β”œβ”€β”€ image001.png
        β”œβ”€β”€ image001.txt
        └── ...

Privacy Rules (CRITICAL β€” All Phases)

  • NO DATA INSPECTION: Do NOT cat, read, or analyze image file contents or .txt caption files.
  • NO CLOUD UPLOAD: All face verification (DeepFace) must run locally. Never send images to cloud APIs.
  • NO DATA LEAKAGE: Do not describe dataset details (person names, attributes) to the LLM unnecessarily.
  • Treat datasets as opaque binary blobs except when running local scripts.

    • Quality Standards (SDXL)

  • Resolution: 1024Γ—1024 minimum after crop
  • Format: Convert all to PNG before training
  • No black borders: Run autocrop before final save
  • Dataset diversity: β‰₯30% clothed/natural skin shots

    • Scripts

    | Script | Location | Purpose | |--------|----------|---------| | tag_batch.py | skills/lora-pipeline/scripts/tag_batch.py | Local WD14 ONNX tagger for a directory | | smart_crop.py | skills/lora-pipeline/scripts/smart_crop.py | Interactive or automated single-subject cropping | | batch_lora_train.py | skills/lora-pipeline/scripts/batch_lora_train.py | Kohya batch training runner for RunPod |

    Sub-Agent Protocol

    Each phase file contains: 1. Input Contract β€” what must already exist before this phase starts 2. Output Contract β€” what this phase produces 3. Completion Signal β€” how to report back (sessions_send + status file fallback) 4. Error Escalation β€” sub-agent reports to parent, never self-escalates model tier