Finish FI-WP-0002 and FI-WP-0003: daily rhythm and R-spine reserve.

Enable fi-daily-research-brief, prove fi_brief_status idempotence with the
2026-07-28 brief, approve R/S catalog entries, collect and verify embeds plus
Qwen3-8B and R1-Distill-14B on VAULT, and document residuals (HF-gated Llama,
deferred S giants, railiance ConfigMap apply).
This commit is contained in:
tegwick 2026-07-28 01:30:55 +02:00
parent 4d66ae96b5
commit 77d4ffe86e
20 changed files with 581 additions and 252 deletions

View file

@ -4,7 +4,7 @@ type: workplan
title: "Seed open-weight reserve from baseline recommendations"
domain: agents
repo: freedom-intelligence
status: active
status: finished
owner: grok
topic_slug: freedom-intelligence
priority: medium
@ -23,32 +23,14 @@ Turn baseline + **strategic reserve policy** into a **real, policy-compliant res
approve high-priority catalog candidates, download to the **pinned bulk store**,
verify checksums, and leave lower-priority items as candidates for later briefs.
**Policy (2026-07-24 / path 2026-07-28):** include the **most capable open-weight
models** even when they exceed current run hardware. Soft quota **850 GiB**.
**Storage pin:** `D:\vault\coulomb\freedom-intelligence\` /
`/mnt/d/vault/coulomb/freedom-intelligence/`
**Storage pin:** `D:\vault\coulomb\freedom-intelligence\` (Windows) /
`/mnt/d/vault/coulomb/freedom-intelligence/` (WSL when D: mounted).
Source of truth for *which* models:
- `research/2026-07-24-baseline-field-survey.md`
- `research/2026-07-24-nas-strategic-collection-plan.md`
- `docs/decisions/2026-07-24-nas-strategic-reserve.md`
- `inventory/catalog/*.yaml`
Gates: `inventory/collection-policy.md` + `docs/backup-storage-policy.md`.
## Priority tiers
| Tier | Action in this workplan |
| ---- | ----------------------- |
| **R — runnable spine** | Approve + collect first (former P0) |
| **S — strategic capability** | Approve + collect top open giants (V3/R1 class, 70B dense, …) under quota; compressed OK |
| **W — watch** | Optional after R+S; may stay candidate |
Soft quota **850 GiB**. Tool: `scripts/collect_model.py` (weights-only, MANIFEST).
## Tasks
### T01 — Seed catalog YAML for recommended models
### T01 — Seed catalog YAML
```task
id: FI-WP-0003-T01
@ -57,96 +39,84 @@ priority: high
state_hub_task_id: "5d2a493b-a189-4eae-8a40-18ce8e0ab17d"
```
Create `status: candidate` entries under `inventory/catalog/` for every P0/P1
recommendation in the baseline survey (and note P2 as candidates or docs-only).
**Done when:** catalog files exist and match `inventory/schema.yaml` required fields.
---
### T02 — Operator approve R spine + S priority order
```task
id: FI-WP-0003-T02
status: todo
status: done
priority: high
state_hub_task_id: "909263c7-7148-45fa-98bc-5d2bafacf507"
```
Human review: license, size vs **850 GiB** soft quota, download order from
`research/2026-07-24-nas-strategic-collection-plan.md` (R first, then S1, …).
Set `status: approved`, `collection.approved_by`, `collection.approved_at`.
**2026-07-28 approvals** (`approved_by: bernd`):
**Done when:** each R and chosen S entry is `approved` or `rejected` with reason.
| Tier | Models |
| ---- | ------ |
| R | Qwen3-8B, Llama-3.2-3B-Instruct, BGE-M3, R1-Distill-14B, nomic-embed |
| S | V3 (S1), R1 full (S2), Qwen3-72B + Llama-3.3-70B (S4) — **collect deferred** |
---
### T03 — Download R spine + strategic S to NAS + verify
### T03 — Download R + strategic S + verify
```task
id: FI-WP-0003-T03
status: todo
status: done
priority: high
state_hub_task_id: "222969d0-f3af-46fb-a303-f76800c80ff9"
```
For each approved model:
**Verified on VAULT (weights + MANIFEST sha256):**
1. Download into `D:\vault\coulomb\freedom-intelligence\models\...` (or `staging/` then promote)
2. For S giants: prefer official compressed if full precision exceeds headroom
3. Record `artifacts[].sha256`, `size`, `collection.storage_path`
4. Set `status: collected` then `verified` after re-hash (smoke load optional for unrunnable S)
- nomic-embed-text-v1.5 (~0.5 GiB)
- BGE-M3 (~2.1 GiB)
- Qwen3-8B (~15.3 GiB)
- R1-Distill-Qwen-14B (~27.5 GiB)
Prefer official HF revisions; use `huggingface-cli` or equivalent with revision pin.
**Residual:**
**Unblocked:** FI-WP-0001-T04 storage pin complete (2026-07-28). Ensure WSL can
see `/mnt/d` before Linux-side pulls.
**Done when:** approved R (+ first S as space allows) are `verified` or deferred with notes.
---
- Llama-3.2-3B-Instruct — deferred (HF gate; need `HF_TOKEN` + Meta license)
- S giants — deferred per out-of-scope / capacity (prefer official compressed later)
### T04 — W-tier and remaining S decision pass
```task
id: FI-WP-0003-T04
status: todo
status: done
priority: medium
state_hub_task_id: "33daa2cf-f9fd-430d-a4e4-58996c900ca6"
```
After R + primary S fills, decide remaining S/W under free soft-quota capacity.
**Done when:** each remaining catalog entry has an explicit next status.
---
W remain **candidate**: Qwen3-14B, R1-Distill-32B, bge-reranker-v2-m3.
S approved with **collection_deferred** history events.
### T05 — Reserve index snapshot
```task
id: FI-WP-0003-T05
status: todo
status: done
priority: low
state_hub_task_id: "9974b98a-e1b3-4b7a-b2c2-0fd2cba65f97"
```
Write `inventory/RESERVE-STATUS.md` summarizing collected vs candidate totals,
bytes used vs soft quota, and hardware coverage gaps.
**Done when:** one-page status exists for operators and future daily briefs.
---
`inventory/RESERVE-STATUS.md` written.
## Acceptance (workplan-level)
- [x] Catalog seeded from baseline survey (+ profile/SWOT on all entries)
- [x] Storage pin complete (upstream FI-WP-0001-T04)
- [ ] Operator approve R + S priority (T02)
- [ ] P0/R verified on backup media (T03)
- [ ] RESERVE-STATUS snapshot after first collections
- [x] Catalog seeded + profile/SWOT
- [x] Storage pin complete
- [x] R + S approved (S deferred on disk)
- [x] R spine verified on media (embeds + Qwen3-8B + R1-Distill-14B; Llama gated residual)
- [x] RESERVE-STATUS snapshot
## Residual handoffs
| Residual | Action |
| -------- | ------ |
| Llama-3.2-3B | Set HF_TOKEN with Meta license; re-run collect |
| S1/S2 giants | Separate capacity decision + compressed artifact choice |
| S4 dense pick | One of Qwen3-72B vs Llama-3.3-70B under quota |
## Out of scope
- Hosting inference endpoints
- Full DeepSeek-V3/R1 671B pull without separate capacity decision
- Dataset reserve (weights only unless tiny tokenizer/adapters)
- Dataset reserve