Files
whetstone_DSL/specialists/data/combined/prereq_op_eval.tsv
Bill 3fc9deac5b specialists: switch training to PyTorch, add capacity sweep infrastructure
- train_specialist_pt.py: pure PyTorch trainer replacing Fabricate binary.
  No SQLite/WAL — saves checkpoint.pt + history.json only. Fabricate was
  causing D-state IO blocking on the HDD due to continuous WAL writes.
- eval_specialist_pt.py: PyTorch eval with confusion matrix, threshold
  analysis, guardrail assessment, and deployment verdict.
- capacity_sweep.py: updated to call PyTorch scripts; TIERS now include
  heads field (4/6/8) since PyTorch supports multi-head unlike Fabricate.
- prereq_op_binary_split.py: evaluate_combined rewritten in PyTorch.
- Performance fix in train_specialist_pt.py: load training data onto GPU
  once and sample via torch.randint instead of DataLoader (eliminates
  Python/CPU overhead on tiny datasets). Fix applied, not yet timed.

Next session: time the fix, then run sweep_all_gates.sh --pilot-only.
See HANDOFF-2026-03-31.md for full context.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-31 22:55:51 -06:00

2.9 KiB

120`whetstone_enqueue_pair_upgrade` MCP tool
230Deprecation/succession lifecycle framework
330Security scan orchestrator for generated targets
410Phase 12c Integration Tests
520Reviewer quorum policy model for critical approvals
610Phase 38a Integration
710Phase 23b Integration + Sprint Summary
800Reprojection safety policy
920Stable corpus certification suite
1000Timing regression runner integration
1100`whetstone_get_tool_permissions` MCP tool
1230Cross-language rollout state snapshot model
1310Compatibility policy profiles by interface criticality
1430`whetstone_publish_roadmap_epoch` MCP tool
1520Security Report Generation
1620Compatibility/deprecation report artifact
1710Human Review Interface via MCP
1820HiveMind job publisher from editor
1910Parser/AST adapter interface schema
2000Sprint 76 integration summary + regression
2110Scenario — Multi-Model Orchestration
2200Autonomous loop controller with “green or explicit blocked” contract
2320Security-Aware Routing
2400Verifier trust weighting policy engine
2500Define verification and readiness surface | Build distributed rate limiter with fixed/sliding window modes and policy APIs.
2610Align docs with bootstrap architecture | Build KYC workflow with document checks, status transitions, and reviewer assignment queues.
2730Align docs with bootstrap architecture | Build document search service with indexing, permissions filtering, and ranking features.
2830Plan Introduce deny-by-default RBAC policy and propagate through backend | Introduce deny-by-default RBAC policy and propagate through backend auth middleware, endpoint guards, frontend route gua
2910Align docs with bootstrap architecture | Build KYC workflow with document checks, status transitions, and reviewer assignment queues.
3020Plan Add normalized address schema migrate legacy rows with | Add normalized address schema, migrate legacy rows with backfill job, dual-read/write transition, and rollback script.
3110Define verification and readiness surface | Refactor billing pipeline to introduce idempotency keys and rollback checkpoints across ledger service, queue worker, re
3220Plan Replace search ranking pipeline behind feature flag with | Replace search ranking pipeline behind feature flag with staged rollout, telemetry checks, and automatic abort trigger.
3300Define verification and readiness surface | Build a JWT auth service with register, login, refresh token rotation, and password hashing.
3430Align docs with bootstrap architecture | Build document search service with indexing, permissions filtering, and ranking features.
3530Align docs with bootstrap architecture | Build document search service with indexing, permissions filtering, and ranking features.
3630Plan Introduce deny-by-default RBAC policy and propagate through backend | Introduce deny-by-default RBAC policy and propagate through backend auth middleware, endpoint guards, frontend route gua