dotfiles/claude/memory/training-pipeline.md at e9e772d69c59d65cfcaba596a1b4bcd64f2b6801

Files

T

David F GliddenandClaude Opus 4.7 119285cf43 claude: back up custom skills, memory, and settings with symlink pattern

The ~/.claude/ directory was previously local-only — a machine wipe
would have lost the accumulated memory, custom skills, and settings.
This commit moves the durable parts into dotfiles with the same
symlink-to-home pattern used for CLAUDE.md, PENDING.md, REVIEWED.md,
and L2-BOOTSTRAP.md.

Preserved (symlinked from ~/.claude/* into here):
  skills/audit/              — thinking-folder drift scanner
  skills/symmetria/          — practice-of-return discipline
  skills/vault-update-people/ — Obsidian People-file maintainer
  skills/wake-up/            — session restoration
  skills/wrap-up/            — session state capture
  memory/                    — 55+ memory files (MEMORY.md, sessions,
                               ledgers, project state, feedback, etc.)
  settings/settings.json     — user preferences (hooks, flags, no secrets)

Deliberately NOT backed up:
  settings.local.json   — contains operational secrets (HF_TOKEN,
                           SSH password in expect scripts); by naming
                           convention, *.local.* is not synced.
                           Needs separate review and probable rotation.
  sessions/, history.jsonl, caches, telemetry — ephemeral
  plugins/, marketplace skills and agents — reinstallable

The working copies at ~/.claude/skills/* and
~/.claude/projects/-Users-davidglidden/memory are symlinks into this
directory, so every write flows here automatically. install.sh
recreates the symlinks on a fresh machine.

FOLLOW-ON (flagged, not in this commit):
  settings.local.json contains a HuggingFace token and an SSH password
  as plaintext strings inside allowed Bash command patterns. These
  should be rotated and moved to secure storage (keychain / pass /
  env file outside the settings file).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-04-17 13:51:04 +02:00

3.3 KiB

Raw Blame History

name, description, type

name	description	type
training-pipeline-state	LoRA training pipeline — model trained, GGUF transfer failed over VPN. Must download on home LAN. Then rebuild BMF + reconnect MCP.	project

Training Pipeline — State as of 2026-03-16 (evening)

MODEL TRANSFER — BLOCKED ON LAN ACCESS

6.2 GB GGUF on clasp at /tmp/lora-david/model-f16.gguf
~30 rsync attempts over VPN from Berlin failed: hash mismatches from partial corruption + 120s timeout too short for 1.4 MB/s VPN throughput
Partial file at /tmp/cm-david-model.gguf is corrupt — delete before next attempt
Script ready: /tmp/pull-model-vpn.sh (600s timeout, SSH keepalive, auto-registers in Ollama)
Action when home (2026-03-17): rm /tmp/cm-david-model.gguf && nohup /tmp/pull-model-vpn.sh > /dev/null 2>&1 &
On LAN (~200 MB/s) a clean transfer takes ~30 seconds
Log: tail -f /tmp/pull-model.log

WHEN MODEL LANDS

Verify: ollama list | grep cm-david
Rebuild BMF: cd ~/_ Dev/BetterMemories.io && npm run build
Reconnect MCP in Claude Code: /mcp
This activates Seb's fixes: teacher during ingestion (#49), chat export (#47), pairing fix (#43), graduation fallback (#45)

Seb's Fixes (pulled to main, 2026-03-16)

8dd17a0 #39 — 'auto' backend default
eca8da7 #40 — llama.cpp GGUF fallback (auto-clones)
7fec033 #43 — Pairing ID mismatch detection
422212f #45 — Graduation fallback when teacher unavailable
6c332b7 #47 — Chat format in export_training_pairs
ce03770 #49 — Teacher enrichment during connector ingestion (rate-limited)
39915cc — Flush timer fix for circle pairings

Second Training Run (COMPLETE)

14,199 deduplicated pairs from David's instance (vault + git + BMF sessions)
Qwen2.5-3B-Instruct, rank-16, 2,000 iters, 21 min on M4 Pro
Output: /tmp/lora-david/ on clasp, registered as cm-david-all-slots:lora-v1

Clasp State (cleaned up this session)

122 stale Modelfile models removed
PATH fixed (#42): node/npm available in SSH sessions via ~/.zshenv
TypeScript rebuilt with #39 fix
mlx-fine-tune.py updated with llama.cpp fallback
BMF v0.48.0 running, healthy

Ingestion

Claude transcripts: idle / needs restart — was at 7,965/72,938 (11%) but BMF process restart lost the job
Restart after model lands + BMF rebuild: connect_service with claude_transcripts
With Seb's #49 fix + MCP reconnect, remaining transcripts WILL generate training pairs via teacher

Remaining Issues (not yet resolved by Seb)

#41 — Qwen3.5 GGUF broken (upstream, informational)
#42 — Clasp SSH PATH (we fixed manually via .zshenv)
#44 — Clasp teacher not configured (decided to skip — unnecessary token burn)
#46 — Stale models (we cleaned manually)
#48 — GGUF distribution pipeline (future feature)

Architecture Finding

Training pairs only generated via sampling/teacher provider in slot chains
All slots start SIGNALING with rules-only → no sampling → no pairs from ingestion
Seb's #49 fix adds optional teacher enrichment during ingestion (rate-limited)
#45 fix helps graduation progress when teacher unavailable

Why: Training flywheel = sovereignty. Local models that learn from your data. How to apply: Model landing → reconnect MCP → Chamber testable. Next training round after ingestion completes (more data = better model).

3.3 KiB Raw Blame History