tegwick
a78187c33e
The weekday brief only counts when the file is on origin/main. Hub events without git objects are failures. The open-weight SoT moves from the VAULT 850 GiB quota to a dedicated Scaleway bucket (One Zone IA for R, Glacier for S) so V4-Flash and K3 are economically in-scope. Catalogs the real DeepSeek-V4-Flash-0731 (MIT ~167 GiB MoE, not 12B) and nominates Qwen3.8-27B. Bucket create remains FI-WP-0004-T08. Assistant: grok Assistant-Session: 01a09c6a-1cfc-75b1-a78d-c13eaf22241d
92 lines
3 KiB
Markdown
92 lines
3 KiB
Markdown
# Hardware envelope — run vs reserve
|
||
|
||
**Status:** updated 2026-09-13 (Scaleway object reserve; run envelope unchanged)
|
||
**Used by:** inventory `hardware_class`, axis B/C, collection tiers R/S
|
||
|
||
---
|
||
|
||
## Purpose
|
||
|
||
Separate two questions that used to be conflated:
|
||
|
||
1. **What can we run or fine-tune now?** → *run envelope*
|
||
2. **What should we store just in case?** → *reserve envelope* (Scaleway Glacier / One Zone IA)
|
||
|
||
Collecting a model **does not** require it to fit the run envelope.
|
||
See `docs/decisions/2026-09-13-scaleway-object-reserve.md`.
|
||
|
||
---
|
||
|
||
## Run envelope (inference / training *today*)
|
||
|
||
| Tier | Assumed capacity | Typical runnable class |
|
||
| ---- | ---------------- | ---------------------- |
|
||
| **T0 — CPU / edge** | 16–64 GB system RAM, no GPU | ≤3B Q4; embeddings ≤0.5B |
|
||
| **T1 — consumer GPU** | 8–12 GB VRAM | 7–8B Q4/Q5; 3B fp16 |
|
||
| **T2 — enthusiast GPU** | 16–24 GB VRAM | 14B Q4; 8B fp16; light 32B Q4 |
|
||
| **T3 — lab multi-GPU** | 2×24 GB+ or 48 GB+ | 32B fp16 / 70B Q4; small MoE |
|
||
| **T4 — beyond current lab** | multi-node / datacenter class | Full large MoE / 70B+ fp16 dense |
|
||
|
||
**Runnable spine (tier R)** targets **T0–T2** (stretch T3 when hardware exists).
|
||
|
||
**Strategic reserve (tier S)** may be **T3–T4 or above** for *execution* — still
|
||
valid for NAS collection.
|
||
|
||
---
|
||
|
||
## Reserve envelope (object store)
|
||
|
||
| Parameter | Value |
|
||
| --------- | ----- |
|
||
| Facility | Scaleway Object Storage `nl-ams` (`railiance-fi-open-weight-reserve`) |
|
||
| Soft budget | **€15 / month** |
|
||
| R class | One Zone IA |
|
||
| S class | Glacier |
|
||
| Staging | VAULT HD `D:\vault\coulomb\freedom-intelligence\` |
|
||
| Policy | `docs/backup-storage-policy.md` |
|
||
|
||
---
|
||
|
||
## Lab hosts (fill in)
|
||
|
||
| Host | Role | GPU | VRAM | RAM | Notes |
|
||
| ---- | ---- | --- | ---- | --- | ----- |
|
||
| TBD | primary local inference | TBD | TBD | TBD | Run envelope |
|
||
| TBD | training experiments | TBD | TBD | TBD | Run envelope |
|
||
| **VAULT HD (D:)** | staging / R cache | n/a | n/a | n/a | **Not SoT** — downloads and optional local R |
|
||
| **Scaleway nl-ams** | weight SoT | n/a | n/a | n/a | **Reserve envelope** — €15/mo soft |
|
||
| railiance01 | cluster | n/a | n/a | n/a | Do not store weight trees on hot disks |
|
||
|
||
---
|
||
|
||
## Inference runtime defaults (when runnable)
|
||
|
||
| Runtime | When |
|
||
| ------- | ---- |
|
||
| llama.cpp / GGUF | Desktop & edge |
|
||
| vLLM | Server multi-request |
|
||
| MLX | Apple silicon if present |
|
||
| Ollama | Quick operator UX only |
|
||
|
||
Strategic giants may remain **cold storage only** until hardware upgrades.
|
||
|
||
---
|
||
|
||
## Training defaults (when runnable)
|
||
|
||
| Method | Envelope |
|
||
| ------ | -------- |
|
||
| QLoRA 7–8B | T1–T2 |
|
||
| QLoRA 14B | T2 |
|
||
| Full FT 7B | T2–T3 |
|
||
| 70B+ / large MoE FT | Future facility — weights may already sit on Glacier |
|
||
|
||
---
|
||
|
||
## Pin log
|
||
|
||
| Date | Change |
|
||
| ---- | ------ |
|
||
| 2026-07-24 | Draft tiers |
|
||
| 2026-07-24 | Split run vs reserve; NAS 1 TB; S tier may exceed run envelope |
|
||
| 2026-09-13 | Reserve envelope moved to Scaleway; VAULT is staging; euro gate |
|