File size: 5,186 Bytes
1219ee2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 | {
"schema": "meta3.process_model.v1",
"status": "active_process_contract",
"principle": "Do not scale until the smaller loop proves evidence quality, validity, and owner usefulness.",
"global_stop_conditions": [
"unresolved source-artifact mismatch",
"missing run manifest or batch plan",
"network dry run not performed before real network calls",
"error rate or empty-output rate exceeds the accepted threshold",
"answers lack evidence or counterevidence fields",
"owner cannot tell what changed since Meta2.0"
],
"control_metrics": [
"source_coverage_pct",
"source_family_spread",
"time_range_spread",
"claim_count",
"claims_with_evidence_pct",
"claims_with_counterevidence_pct",
"empty_or_low_confidence_output_pct",
"actual_api_calls",
"cost_estimate",
"owner_ready_answer_count"
],
"phases": [
{
"id": "p00_requirements_lock",
"kind": "manual_gate",
"purpose": "Lock the owner question, allowed sources, stop conditions, budget, and morning success definition.",
"inputs": ["GOAL", "requirements/PENDING.md", "config/questions.json"],
"outputs": ["requirements/LOCKED.md"],
"entry_gate": "Samuel's requirements have arrived.",
"exit_gate": "Questions, constraints, and success criteria are reconciled."
},
{
"id": "p01_baseline_audit",
"kind": "measurement_gate",
"purpose": "Measure the Meta2.0 baseline before changing scale.",
"inputs": [
"/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
"/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json",
"/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl"
],
"outputs": ["runs/<run_id>/manifests/baseline_snapshot.json"],
"entry_gate": "Meta2.0 artifacts exist.",
"exit_gate": "Coverage, unique digest count, raw row count, failures, and source spread are recorded."
},
{
"id": "p02_calibration_slice",
"kind": "small_model_run",
"purpose": "Answer a small representative slice before broad overnight execution.",
"inputs": ["baseline_snapshot.json", "config/questions.json", "config/experiment_design.json"],
"outputs": ["runs/<run_id>/outputs/calibration_answers.jsonl"],
"entry_gate": "Baseline audit is valid.",
"exit_gate": "At least one answer shows evidence, uncertainty, counterevidence status, and owner-ready conclusion."
},
{
"id": "p03_stratified_backfill",
"kind": "meta2_layer1_backfill",
"purpose": "Increase digest coverage while preserving source and time diversity.",
"inputs": ["baseline_snapshot.json"],
"outputs": ["updated Meta2.0 layer1 audit"],
"entry_gate": "Calibration did not reveal a method flaw.",
"exit_gate": "New audit shows no unresolved failures and improved coverage."
},
{
"id": "p04_claim_ledger_generation",
"kind": "structured_answer_run",
"purpose": "Generate structured claims per question instead of freeform essays.",
"inputs": ["session_digests.jsonl", "config/questions.json", "config/experiment_design.json"],
"outputs": ["runs/<run_id>/outputs/claim_ledger.jsonl"],
"entry_gate": "Coverage gate and requirements gate pass.",
"exit_gate": "Claims cite evidence and mark uncertainty."
},
{
"id": "p05_counterevidence_pass",
"kind": "adversarial_review",
"purpose": "Search for sessions that weaken, contradict, or narrow the generated claims.",
"inputs": ["claim_ledger.jsonl", "session_digests.jsonl"],
"outputs": ["runs/<run_id>/outputs/counterevidence_ledger.jsonl"],
"entry_gate": "Claim ledger exists.",
"exit_gate": "Each main claim has counterevidence status."
},
{
"id": "p06_validity_review",
"kind": "quality_gate",
"purpose": "Separate useful answers from invalid measurement.",
"inputs": ["claim_ledger.jsonl", "counterevidence_ledger.jsonl", "run_manifest.json", "batch_plan.json"],
"outputs": ["runs/<run_id>/reports/validity_review.md"],
"entry_gate": "Claim and counterevidence ledgers exist.",
"exit_gate": "Invalid, weak, and owner-ready outputs are explicitly separated."
},
{
"id": "p07_owner_synthesis",
"kind": "owner_report",
"purpose": "Compress validated claims into canonical owner-facing answers.",
"inputs": ["validity_review.md", "claim_ledger.jsonl", "counterevidence_ledger.jsonl"],
"outputs": ["runs/<run_id>/reports/OWNER_SYNTHESIS.md"],
"entry_gate": "Validity review passes.",
"exit_gate": "Each question has one thesis, evidence, limits, and next action."
},
{
"id": "p08_morning_closeout",
"kind": "closure",
"purpose": "Make the overnight run inspectable and stop hidden cycle debt.",
"inputs": ["run_manifest.json", "batch_plan.json", "logs", "reports"],
"outputs": ["runs/<run_id>/reports/MORNING_REPORT.md"],
"entry_gate": "Overnight run ended or was stopped.",
"exit_gate": "Actual calls, errors, outputs, missing evidence, and next decision are visible."
}
]
}
|