README: reflect deployed state #12

2026-05-25T17:51:17-04:00

justin commented

2026-05-25 17:51:17 -04:00

Rewrites the scaffold-era README to match the deployed state.

Corpus inventory: 760 variety records + 4,313 trial documents = 5,073 total chunks across 6 sources
All 6 MCP tools documented (was missing lookup_variety and search_trials; had a nonexistent corpus_status)
Eval baseline table — hybrid+rerank wins 100% pass / 90% P@1 / MRR 0.905
Surprising findings: dense alone is noise; hybrid w/o rerank is WORSE than BM25 alone
Deploy mechanics (Watchtower chain, 4-GPU pool, shared llama-rerank with network-attach gotcha)
Status table updated — Phases 1-8 + 11 all ✅

Rewrites the scaffold-era README to match the deployed state. - Corpus inventory: 760 variety records + 4,313 trial documents = **5,073 total chunks** across 6 sources - All 6 MCP tools documented (was missing `lookup_variety` and `search_trials`; had a nonexistent `corpus_status`) - Eval baseline table — hybrid+rerank wins 100% pass / 90% P@1 / MRR 0.905 - Surprising findings: dense alone is noise; hybrid w/o rerank is WORSE than BM25 alone - Deploy mechanics (Watchtower chain, 4-GPU pool, shared llama-rerank with network-attach gotcha) - Status table updated — Phases 1-8 + 11 all ✅

justin added 1 commit 2026-05-25 17:51:17 -04:00

README: reflect deployed state — 5,073 chunks, eval numbers, 6 tools d65c7d0d67

The scaffold-era README was out of sync with the shipped product:
- Vendor counts stale (recon estimates, not actual deployed counts)
- Trial data sources (gh_plot_reports + agripro_trials) entirely
  unmentioned
- Tool list listed `corpus_status` (doesn't exist) and missed both
  `lookup_variety` and `search_trials`
- Build-phase table showed everything as "pending" / "next" but
  Phases 1-8 + 11 all shipped

Rewrite to reflect the deployed state:
- Corpus inventory: 760 variety records + 4,313 trial documents =
  5,073 chunks across 6 sources
- All 6 MCP tools documented with their purpose
- Eval baseline table (hybrid+rerank wins 100%, P@1 90%, MRR 0.905)
  with the surprising findings (dense alone is noise; hybrid w/o
  rerank is WORSE than BM25 alone)
- Deploy mechanics: Watchtower chain, 4-GPU embedder pool, shared
  llama-rerank sidecar with the network-attach gotcha
- Status table: ✅ on the phases that shipped, deferred work list
  (becks_pfr, 2023 plot backfill, NK trials, Channel Seed brand)

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

justin merged commit c76df4c44a into main

2026-05-25 17:51:19 -04:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: justin/seed-mcp#12