File size: 6,250 Bytes
dca5b2d | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 | {
"package_version": "3.5.0.post7",
"parent_revision": "213eed554008e633b1150977049a8c66b7717852",
"default_merge_policy": "alternatives",
"default_schedule": "pipelined",
"chart_rules_enabled_by_default": true,
"regression_tests": {
"passed": 86,
"failed": 0,
"source_tested": true,
"installed_wheel_tested": true
},
"wheel": {
"filename": "nemotron_parse-3.5.0.post7-py3-none-any.whl",
"sha256": "e403d294fc9c0563f6d38028ce0e0a639977fdba2807fa6981692aabbf269077",
"all_runtime_python_identical_to_post6": true,
"RECORD_verified": true,
"repeat_build_identical": true
},
"runtime_validation": {
"new_gpu_runs_for_post7": 0,
"new_official_score_runs_for_post7": 0,
"note": "Notice/version/header-only release. Runtime Python byte parity with post6 is verified; no new model-inference or scoring result claimed.",
"previous_version_validation": {
"package_version": "3.5.0.post6",
"parent_revision": "6cbbc638b8fdb1d94d76d5b9830d43e55da290ad",
"default_merge_policy": "alternatives",
"default_schedule": "pipelined",
"chart_rules_enabled_by_default": true,
"regression_tests": {
"passed": 86,
"failed": 0,
"source_tested": true,
"installed_wheel_tested": true,
"timing_checks": [
"dependent crop preparation counted",
"final render excluded",
"overlapping spans counted as wall time"
],
"existing_coverage": [
"default retains differing alternatives",
"matching alternatives deduplicated",
"explicit guarded replacement",
"both merge policies preserve staged/pipelined output equivalence for deterministic responses",
"early crops and full-page/detection overlap",
"shared HTTP window",
"partial output accounting",
"HTTP default 128 and custom-backend default 4",
"crop geometry, native tables, charts and text preservation"
]
},
"wheel": {
"filename": "nemotron_parse-3.5.0.post6-py3-none-any.whl",
"sha256": "25203cdedc40fc7c362f7de60b3b858bbcf227efb4d473ae198242c5b9bb8dff",
"source_byte_parity": true,
"source_files_checked": 33,
"entry_points_verified": true
},
"runtime_validation": {
"new_gpu_runs_for_post6": 0,
"new_official_score_runs_for_post6": 0,
"scope": "Timing-only instrumentation. Existing post5 model/scoring evidence retained in ACCURACY_RESTORATION_RESULTS.json; existing single-pass timing from post4. No claim that post6 wheel was GPU benchmarked.",
"runtime_files_changed_from_post5": [
"nemotron_parse/batch.py",
"nemotron_parse/SOURCE_MANIFEST.json"
],
"model_settings_and_prompts_unchanged": true,
"crop_selection_encoding_merge_and_rendering_sources_unchanged": true,
"server_plugin_unchanged": true,
"timing_receipt": "inference_benchmarks.json",
"prior_validation": {
"package_version": "3.5.0.post5",
"parent_revision": "0fad7099302dc11313b7e3157eb8fd6ae633176e",
"default_merge_policy": "alternatives",
"optional_merge_policy": "guarded",
"default_schedule": "pipelined",
"chart_rules_enabled_by_default": true,
"regression_tests": {
"passed": 83,
"failed": 0,
"source_tested": true,
"installed_wheel_tested": true,
"coverage": [
"default retains differing alternatives",
"matching alternatives deduplicated",
"explicit guarded replacement",
"both merge policies preserve staged/pipelined output equivalence for deterministic responses",
"early crops and full-page/detection overlap",
"shared HTTP window",
"partial output accounting",
"HTTP default 128 and custom-backend default 4",
"crop geometry, native tables, charts and text preservation"
]
},
"wheel": {
"filename": "nemotron_parse-3.5.0.post5-py3-none-any.whl",
"sha256": "f4efe1f21f73dac8a4ed27ecc5b05b83bd1cd33d5194dfd1569e188a5a4d16d5",
"source_byte_parity": true,
"source_files_checked": 33,
"installed_wheel_tests_passed": 83,
"failed": 0,
"entry_points_verified": true,
"vllm_plugin_source_unchanged": true
},
"runtime_validation": {
"fresh_real_model_full_runs": 2,
"pages_per_run": 2078,
"one_h100_per_profile": true,
"frozen_wheel_matches_release": true,
"default_merge_policy_tested": true,
"official_score_runs": 6,
"fresh_official_page_evaluations": 12468,
"official_errors": 0,
"official_scorer_revision": "9af4102e26b51f234ab5f037e19fec05c0696ec6",
"verified_scorer_source_files": 241,
"sampling_settings_unchanged": true,
"benchmark_receipt": "accuracy_restoration.json",
"prior_scheduler_comparison_receipt": "batch_benchmarks.json",
"pinned_reproduction_unchanged": true,
"partial_generation_policy": "Benchmarks explicitly allow partial responses; incomplete pages and capped crops remain accounted for. Strict handling remains the API default."
},
"interpretation": "The restored merge recovers the reference multiple-hypothesis output contract. A higher score over alternatives is not proof of an equally accurate unique transcription. The guarded single-region policy is available explicitly.",
"format_validation": {
"markdown_fences_balanced": true,
"python_examples_parsed": 12,
"bash_examples_syntax_checked": true
}
}
},
"format_validation": {
"markdown_fences_balanced": true,
"examples_syntax_checked": {
"python": 13,
"bash": 9
}
}
}
},
"compliance_checks": "compliance.json",
"build_provenance": "../package/build_provenance.json"
}
|