Laguna-S-2.1-Spark-Agentic / results /freeze_notes_2026-07-29.md
hizrianraz's picture
docs(public): professional cards, PII-clean measured surface, exclusive research/prompts off HF
1993625 verified
|
Raw
History Blame
2.3 kB

Freeze notes — 2026-07-29

Gate

  • Freeze gate: model_card + lock files
  • Status: FILLED at 2026-07-29T14:43:26+07:00
  • Receipt: results/freeze_gate_model_card_lock_set_2026-07-29.json
  • Freeze clock: still 2026-08-02 18:00 local (do not freeze early)
  • HF go-live: 2026-08-03 12:00 local

Live tip bound into card + locks

Field Value
Pack measure tip bf82eab
Q4 sha256 a8b55c75714ea73fd90ec85de5defdc0b8d88ca0ad2108343cdd8fc22f7583e4
agent_smoke 40/40 · 84.86s
hermes v2 27/27 · 100.1s
gen128 ~21.47 t/s
Engine poolsideai/llama.cpp 04b2b72

Locks held

  • diy_gguf: false
  • weight_host: Spark-only
  • Mac ≤32 GB: client-only (no local S)
  • public_promo_before_launch: false
  • XS: parallel, not in S freeze
  • branding: personal-only
  • DFlash: measured, DO_NOT_PROMOTE

Still open before freeze day

  1. Optional DFlash status row finaldone 2026-07-29 · DO_NOT_PROMOTE (results/dflash_2026-07-29/)
  2. Stranger-path dry re-runDRY PASS (results/stranger_path_dry_2026-07-29.md) · serve flag aligned to -ngl -1
  3. Baseline serve re-health after DFlash teardownHEALTH_OK + chat OK 2026-07-29 15:50 local (results/baseline_rehealth_2026-07-29.md)
  4. HF tip push of DFlash/last-green/stranger/jury receipts (docs-only)
  5. Clean pushed HEAD on freeze day (2026-08-02 18:00 local)
  6. Public go-live 2026-08-03 12:00 local (no early promo)

Non-actions

  • No accidental freeze today
  • No public promo before Aug 3 12:00 local
  • No XS inside S freeze artifact
  • No re-bench required unless weights/runners change
  • No DFlash speedup claim

2026-07-29 jury-fix rev (post dual seat)

  • S freeze lock_set is S-only. Dropped results/xs_mac_track_lock.json from S lock_set.
  • XS remains parallel prepared_research under its own lock; not an S freeze hash input.
  • Rehash at freeze: on clean pushed HEAD (not dirty WT), recompute sha256 for every S lock_set member and stamp freeze gate. Docs tip may move; measure tip stays bf82eab until smoke/weights change.
  • Gen headline sole: ~21.47 tok/s from multi suite in results/measured.json. No 22.50 on launch surfaces.
  • SHA256SUMS is GNU 2-column for sha256sum -c.