tests/golden: expected_linux.json from dispatch run 24746071833 (S18.8)
- SHA
e0e16ed241c90790a3446d117635f9f10bb193f4- Parents
-
27af978 - Tree
5ddb874
e0e16ed
e0e16ed241c90790a3446d117635f9f10bb193f427af978
5ddb874| Status | File | + | - |
|---|---|---|---|
| A |
tests/golden/expected_linux.json
|
96 | 0 |
tests/golden/expected_linux.jsonadded@@ -0,0 +1,96 @@ | |||
| 1 | +{ | ||
| 2 | + "determinism": { | ||
| 3 | + "class": "best_effort", | ||
| 4 | + "notes": [ | ||
| 5 | + "CPU-only backend: strict determinism depends on BLAS impl" | ||
| 6 | + ], | ||
| 7 | + "seed": 0 | ||
| 8 | + }, | ||
| 9 | + "null_stats": {}, | ||
| 10 | + "probes": [ | ||
| 11 | + { | ||
| 12 | + "base_value": null, | ||
| 13 | + "ci_95": [ | ||
| 14 | + 0.008465445404266456, | ||
| 15 | + 0.02977853737438285 | ||
| 16 | + ], | ||
| 17 | + "evidence": { | ||
| 18 | + "divergence_kind": "js", | ||
| 19 | + "max": 0.03615893521221293, | ||
| 20 | + "num_prompts": 4, | ||
| 21 | + "per_prompt": [ | ||
| 22 | + 0.010637343860892606, | ||
| 23 | + 0.006726111759252678, | ||
| 24 | + 0.013683446339307789, | ||
| 25 | + 0.03615893521221293 | ||
| 26 | + ], | ||
| 27 | + "raw_ci_95": [ | ||
| 28 | + 0.008465445404266456, | ||
| 29 | + 0.02977853737438285 | ||
| 30 | + ], | ||
| 31 | + "weight": 1.0, | ||
| 32 | + "z_by_rank": null | ||
| 33 | + }, | ||
| 34 | + "ft_value": null, | ||
| 35 | + "kind": "delta_kl", | ||
| 36 | + "message": "mean js=0.0168 (\u2265 0.0) (no calibration for delta_kl)", | ||
| 37 | + "name": "dk_golden", | ||
| 38 | + "raw": 0.0168014592929165, | ||
| 39 | + "score": 0.024239382001588428, | ||
| 40 | + "verdict": "pass", | ||
| 41 | + "z_score": null | ||
| 42 | + }, | ||
| 43 | + { | ||
| 44 | + "base_value": null, | ||
| 45 | + "ci_95": [ | ||
| 46 | + 0.0, | ||
| 47 | + 0.0 | ||
| 48 | + ], | ||
| 49 | + "evidence": { | ||
| 50 | + "fraction_regressed": 0.0, | ||
| 51 | + "mean_delta_nats": -0.03518710732460022, | ||
| 52 | + "raw_ci_95": [ | ||
| 53 | + 0.0, | ||
| 54 | + 0.0 | ||
| 55 | + ], | ||
| 56 | + "regressed_count": 0, | ||
| 57 | + "regression_nats_threshold": 1.0, | ||
| 58 | + "total_items": 20, | ||
| 59 | + "weight": 1.0, | ||
| 60 | + "worst_offenders": [], | ||
| 61 | + "z_by_rank": null | ||
| 62 | + }, | ||
| 63 | + "ft_value": -0.03518710732460022, | ||
| 64 | + "kind": "calibration_drift", | ||
| 65 | + "message": "0/20 items regressed >1.0 nats (frac=0.0%), mean_delta=-0.035 nats/tok (no calibration for calibration_drift)", | ||
| 66 | + "name": "cal_golden", | ||
| 67 | + "raw": 0.0, | ||
| 68 | + "score": 0.8572834380467733, | ||
| 69 | + "verdict": "pass", | ||
| 70 | + "z_score": null | ||
| 71 | + } | ||
| 72 | + ], | ||
| 73 | + "schema_version": 1, | ||
| 74 | + "score": { | ||
| 75 | + "band": "partial", | ||
| 76 | + "components": { | ||
| 77 | + "ablation": 0.0, | ||
| 78 | + "adherence": 0.024239382001588428, | ||
| 79 | + "attribution": 0.0, | ||
| 80 | + "baseline": 0.0, | ||
| 81 | + "calibration": 0.8572834380467733 | ||
| 82 | + }, | ||
| 83 | + "findings": [ | ||
| 84 | + "adherence score is 0.02 \u2014 below the noise threshold" | ||
| 85 | + ], | ||
| 86 | + "overall": 0.3574570044196624, | ||
| 87 | + "weights": { | ||
| 88 | + "ablation": 0.15, | ||
| 89 | + "adherence": 0.3, | ||
| 90 | + "attribution": 0.35, | ||
| 91 | + "baseline": 0.0, | ||
| 92 | + "calibration": 0.2 | ||
| 93 | + } | ||
| 94 | + }, | ||
| 95 | + "spec_path": "<memory>" | ||
| 96 | +} | ||