Publish adoption of the independent Track A four-split pre-score review

This commit is contained in:
NewMeltAssist committed 2026-11-19 01:37:42 +08:00
1 parent 5bfef6a61a
commit c491b2d7cd
8 files changed
+442

No files matched your search

@@ -0,0 +1,115 @@
"""Verify the four-split binding read-only, using metadata and opaque hashes."""
import hashlib
import json
import math
import os
from pathlib import Path
import resource
import subprocess
import sys
from track_a_postfit_rebind_common import ORIGINAL_POSTFIT, RECOVERY, ROOT, config, terminal
def read(path):
return json.loads(path.read_text())
def sha(path):
digest = hashlib.sha256()
with Path(path).open('rb') as handle:
for chunk in iter(lambda: handle.read(1024 * 1024), b''):
digest.update(chunk)
return digest.hexdigest()
def verify_git_bytes(proof):
for path, expected in proof['files'].items():
blob = subprocess.check_output([
'git', '-C', proof['local_repository'], 'show',
f"{proof['commit']}:{proof['git_paths'][path]}",
])
assert hashlib.sha256(blob).hexdigest() == expected, path
def main():
output = Path(__file__).resolve().parent
workspace = output.parents[1]
contract, training, _ = config(ROOT)
proof = read(ROOT / 'implementation-git-freeze.json')
expected = dict(
reviewed_commit='28236c5df91329fa52ef98a584688496d2b4e272',
contract_sha256='cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63',
implementation_proof_sha256='d9ff47cb647cb5e7af6ad7f314a1ad9969b639d1063a5d135b2fd4d73b2a7804',
comparison_contract_sha256='8f0f3c1088d899b85aa3f578b7552f04cb409fd7a5dcfca3e17eaa3d7e25d0d6',
code_sha256=contract['code_sha256'],
)
binding_path = ROOT / 'expected-review-binding.json'
assert read(binding_path) == expected
assert sha(binding_path) == '96e2b0c0c7ccd262ac15f9ba08cfa29d7a01c312e4f8633a1eda56184f61c6df'
assert proof['commit'] == expected['reviewed_commit']
assert sha(ROOT / 'contract.json') == expected['contract_sha256']
assert sha(ROOT / 'implementation-git-freeze.json') == expected['implementation_proof_sha256']
assert sha(contract['comparison_contract']) == expected['comparison_contract_sha256']
assert len(contract['code_sha256']) == 33
verify_git_bytes(proof)
recovery_freezes = {}
for name in ('terminal', 'incomplete-models'):
recovery_proof = read(RECOVERY / f'{name}-git-freeze.json')
verify_git_bytes(recovery_proof)
recovery_freezes[name] = dict(commit=recovery_proof['commit'], files=len(recovery_proof['files']))
evidence = terminal(training)
assert evidence == contract['training_evidence']
assert math.isclose(evidence['training_gpu_seconds_once'], 78579.79901044304, abs_tol=1e-8, rel_tol=0)
assert [model['origin'] for model in evidence['models']] == ['ORIGINAL_HASH_REUSE'] * 3 + ['RECOVERY_FRESH_SEED0']
assert all(model['training_contract_sha256'] == contract['training_contract_sha256'] for model in evidence['models'][:3])
assert evidence['models'][3]['training_contract_sha256'] == sha(RECOVERY / 'contract.json')
original = read(ORIGINAL_POSTFIT / 'contract.json')
split_counts = {}
unique_cases = set()
for split in (0, 1, 2, 4):
job = training['jobs'][split]
references = contract['A_predictions'][str(split)]
assert references == original['A_predictions'][str(split)]
assert len(job['test']) == len(references) and set(job['test']) == set(references)
split_counts[split] = len(references)
unique_cases.update(references)
for case, ref in references.items():
part = training['parts'][case]
assert ref['nodes'] == part['nodes']
assert ref['raw_ids_sha256'] == part['output_sha256']['raw_ids.npy']
assert sha(ref['path']) == ref['sha256']
assert split_counts == {0: 304, 1: 311, 2: 308, 4: 305}
assert sum(split_counts.values()) == 1228 and len(unique_cases) == 571
assert '3' not in contract['A_predictions'] and len(contract['original_failed']) == 14
env = dict(os.environ, PYTHONDONTWRITEBYTECODE='1', CUDA_VISIBLE_DEVICES='',
PYTHONPATH=str(workspace / 'tools'), OPENBLAS_NUM_THREADS='1', OMP_NUM_THREADS='1')
before = resource.getrusage(resource.RUSAGE_CHILDREN)
with (output / 'synthetic-tests.log').open('x') as log:
subprocess.run([sys.executable, '-m', 'unittest', '-v', 'test_track_a_postfit',
'test_track_a_postfit_rebind'], cwd=workspace, env=env,
stdout=log, stderr=subprocess.STDOUT, timeout=60, check=True)
after = resource.getrusage(resource.RUSAGE_CHILDREN)
test_log = output / 'synthetic-tests.log'
assert 'Ran 21 tests' in test_log.read_text() and '\nOK\n' in test_log.read_text()
entry_state = {name: (ROOT / name).exists() for name in
('review.json', 'started.json', 'predictions-manifest.json', 'score', 'complete.json')}
result = dict(status='PASS', binding=expected, binding_file_sha256=sha(binding_path),
implementation_files=len(proof['files']), source_pins=len(contract['source_sha256']),
recovery_freezes=recovery_freezes, training_evidence=evidence,
paired_split_counts=split_counts, paired_visits=1228, unique_parts=571,
tests=21, tests_child_cpu_seconds=(after.ru_utime + after.ru_stime
- before.ru_utime - before.ru_stime),
tests_log_sha256=sha(test_log), entry_state=entry_state,
targets_decoded=False, model_values_loaded=False, GPU_query=False)
with (output / 'evidence.json').open('x') as handle:
json.dump(result, handle, indent=2)
handle.write('\n')
print({key: value for key, value in result.items() if key not in ('binding', 'training_evidence')})
if __name__ == '__main__':
main()
@@ -0,0 +1,133 @@
{
"status": "PASS",
"binding": {
"reviewed_commit": "28236c5df91329fa52ef98a584688496d2b4e272",
"contract_sha256": "cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63",
"implementation_proof_sha256": "d9ff47cb647cb5e7af6ad7f314a1ad9969b639d1063a5d135b2fd4d73b2a7804",
"comparison_contract_sha256": "8f0f3c1088d899b85aa3f578b7552f04cb409fd7a5dcfca3e17eaa3d7e25d0d6",
"code_sha256": {
"/home/frank14f/optane/MeltMind/tools/boundary_distance.py": "a45c1cbce92521a96eb814913b4f465b926c0b3883932ab0b7c37d11db113ed7",
"/home/frank14f/optane/MeltMind/tools/freeze_g1_components.py": "d0cee60d95243ac2c59ef3f2244ba818315718c4e3d1895f2f9f064f1f7f5ecd",
"/home/frank14f/optane/MeltMind/tools/gre24_features_v2.py": "7dcd5ac218ffdc1ca149edcd87e8658fb3c59e18c0fb9ce47985c7f1357d2cdf",
"/home/frank14f/optane/MeltMind/tools/gre24_models.py": "53b6fee8b700b9b982b06d29908ce2ce23e0c57512c46dbf74b2aade79e3ac3a",
"/home/frank14f/optane/MeltMind/tools/phase2_gate_outcome_review.py": "915617aeb8292d03c654e637bc34a014c254fd178a773f135d1806c4b41372a8",
"/home/frank14f/optane/MeltMind/tools/phase2_h1_worker_v2.py": "1543df85b615bb69b79bacd572074a7c77ae19dab110197601573a7e5805ed20",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_data.py": "7e03dd7a94e7bb43593277f183c222473d7cb7aaf2f60e53b45ab8d9499fcb09",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_model.py": "fab2470dd0654ca74c3b7d893f35ad1f3ae6c5c08ffe334ef6312cf6ec87892a",
"/home/frank14f/optane/MeltMind/tools/phase2_home_freeze_io.py": "bf09204b85ed2a7038acdd202257f618474d1110a9d6d772f105a3ffb334d3ff",
"/home/frank14f/optane/MeltMind/tools/phase2_local_freeze_io.py": "60adf3382577613c5234f5cf30779336b9951b66e90c075cb7cd0adab9cc42dd",
"/home/frank14f/optane/MeltMind/tools/phase2_residual_math.py": "ffbd9624dec79bdaabf95244fe87209055908f31a977e73ea4377045d6f8fe62",
"/home/frank14f/optane/MeltMind/tools/phase3_item1_postfit_data.py": "de9cbcc147677c9144c706acbe52c0f7d505dad17d741d86dcbd04d052f940e2",
"/home/frank14f/optane/MeltMind/tools/prepare_track_a_postfit_rebind.py": "e5fcb0c1ca594aa74e6c3cb70dada60f79baa0d8bcaa0622b19da191ad596536",
"/home/frank14f/optane/MeltMind/tools/r4bc_common.py": "27543fff987975ece767167f9771e6621c825035a053805604e4a696794dfcbd",
"/home/frank14f/optane/MeltMind/tools/r4bc_guard.py": "437a756ddf10b125dce450f463e76e28df52724f8c5766e0a989dfa536805628",
"/home/frank14f/optane/MeltMind/tools/s2_gpu_idle_gate.py": "e2b1613e7e6ea398a8d555220a9365a1ad998ab4ef4b2bf8ec7cc4e31f8b4b9f",
"/home/frank14f/optane/MeltMind/tools/s2_main_memory_guard.py": "e094f5443e11518efe9018215c11718a8e18aa462a977bf43e974b2d0b2aebb1",
"/home/frank14f/optane/MeltMind/tools/s2_pilot_train.py": "e9cceead51a5cc092aad23409ad98909c84a62e7e10571e750d1e9be1ad3b948",
"/home/frank14f/optane/MeltMind/tools/s2_precision_continuation_guard.py": "01235d897c7c03e5ffa109c58b14f5d24d128e8f1ba873e1f76cfb14e02f7a7f",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_gpu_supervisor.py": "689c0f3179e4fcd362e546d90214829a9e4f8739abdbdf74b560763e5a392f15",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_monitoring.py": "442c2a9c75e8c00b30a8f2e91cb52d87c466b19a80aea9128b10c021e61b922e",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_supervisor.py": "f563ca3c2665f5baf0c04f296616e4640423a59b815fbb00479af0f29e57a919",
"/home/frank14f/optane/MeltMind/tools/test_track_a_postfit_rebind.py": "aefdab5d2bcb0e4cf63b7ebe4e736307726f2bd8f98dc2985cc85b0d814315b3",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_common.py": "79918e7de887d1f0a2de5523156b27d00dc2d85ac930a07efc34d817d904fe65",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_prepare.py": "f657be036dc48fc5936f56c4b8510a4b295d4787af6877cc5ccaad663107be17",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_common.py": "8b7d425ab702eb033d4027d24d2812042b66411943b004d6be08a13d4ffcd1c9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind.py": "54389c854f8ee43bdf66834675c22b5e705446cab363ebf55d340d8e3c79381b",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_common.py": "b0f283577da25142201c6d2c7fbfdc486e0f21a75a554a6dd8c2a1da58d29778",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_guard.py": "0aaacc9518041235eecb7f078cebba6d0bc62dadfbd4951f0066bd8271db14a8",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_statistics.py": "07b1d26dbdb898d3ffe26bec497204a59226749d299d35ad55b074a4e0ad03e9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_worker.py": "e34dbcd50bd91b9f84367b61f7568e0cdac471e9e1262a082bb9cfba80b79751",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_statistics.py": "b7d346d1762d7737eb1ebd723989a71f8c65b597d4cdb8c5d512b48a0fd984ce",
"/home/frank14f/optane/MeltMind/tools/track_a_recovery_common.py": "a932d069340909465a5a1e66be1504e28c8df159a5ce304b0586295003d2e734"
}
},
"binding_file_sha256": "96e2b0c0c7ccd262ac15f9ba08cfa29d7a01c312e4f8633a1eda56184f61c6df",
"implementation_files": 57,
"source_pins": 36,
"recovery_freezes": {
"terminal": {
"commit": "73fcb4213ec50afe8cb9e0b1558d507725ba8715",
"files": 4
},
"incomplete-models": {
"commit": "cf27dde1cd3c952534e6999ac34a14427027079c",
"files": 11
}
},
"training_evidence": {
"models": [
{
"job": 0,
"path": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/0/worker/last-epoch200.pt",
"sha256": "767c186bdd0c99d9bcd195cadd5e7b15374c4c07bcb53aa03c157219c74b326d",
"standardization": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/0/worker/standardization.json",
"standardization_sha256": "6fab06dd34d30d10409198bc177627293d4d8da559be3b33585c410964e6a15c",
"origin": "ORIGINAL_HASH_REUSE",
"training_contract_sha256": "be4ac1f6d5a5d6c0183aa50d2e046b22d7e5fa77c7714fef7eab6d616e68bee7",
"guard_complete_sha256": "024bb110576da7c351ad8e98227cace2f40f4247a8f349c01ea4ebfde954891c"
},
{
"job": 1,
"path": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/1/worker/last-epoch200.pt",
"sha256": "9b1356403017f5c351729ce3d33494c48033b0addff4b7b9ca6c1aad6bf90b93",
"standardization": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/1/worker/standardization.json",
"standardization_sha256": "6734f28dc7ca6ffd3a39a43ad2964ffb747c77e55d24d0705c3b5b41f56f397b",
"origin": "ORIGINAL_HASH_REUSE",
"training_contract_sha256": "be4ac1f6d5a5d6c0183aa50d2e046b22d7e5fa77c7714fef7eab6d616e68bee7",
"guard_complete_sha256": "9259261f44a5813451e9f6f7878fd69d65db6e509889f7c78adcb6f417e00b3d"
},
{
"job": 2,
"path": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/2/worker/last-epoch200.pt",
"sha256": "0f71453359e8c6df759434b00759b5721e0cd3131cfb474a17a3b47f0c3c2e18",
"standardization": "/home/frank14f/MeltMind-runs/track-a-h2-2026-11-18/jobs/2/worker/standardization.json",
"standardization_sha256": "e1f6c14617ed019379998142d49bf9359474db41e1614e2b4e20834ccc37b09d",
"origin": "ORIGINAL_HASH_REUSE",
"training_contract_sha256": "be4ac1f6d5a5d6c0183aa50d2e046b22d7e5fa77c7714fef7eab6d616e68bee7",
"guard_complete_sha256": "b60b2575b9efdbb28eabcc2682dc04ce3519c9e7b2f6954cc4dcea0e29b1ab2f"
},
{
"job": 4,
"path": "/home/frank14f/MeltMind-runs/track-a-recovery-2026-11-18/jobs/4/worker/last-epoch200.pt",
"sha256": "d6aa7dc1b572c827a09cce9b9c23e9c744f1b283ee352686b3c507588677eb10",
"standardization": "/home/frank14f/MeltMind-runs/track-a-recovery-2026-11-18/jobs/4/worker/standardization.json",
"standardization_sha256": "e8a85e95ad3829856c897be1257e0557363a9293be10e0e25edc5a8c6be7bd0e",
"origin": "RECOVERY_FRESH_SEED0",
"training_contract_sha256": "4b21dedde07534bbe059e9886d7d5fbd7f6d3c33a5b08f4564c9bb360ef511aa",
"guard_complete_sha256": "6c7fb910b9fc928596a080e5ca254dc9c6af6a06d2eea6f16c89183434df4b4f"
}
],
"excluded_split": {
"split": 3,
"status": "NOT_TRAINED_TO_COMPLETION_BUDGET_STOP",
"reason": "OWN_PROJECTED_CAP",
"recovery_completed_epochs": 1
},
"training_gpu_seconds_once": 78579.79901044304,
"original_gpu_seconds_once": 47943.94719455298,
"recovery_gpu_seconds_once": 30635.851815890055,
"original_terminal_sha256": "299a73b8435877cf6f35af7e2340f27a833122588b3e1a7740f2a014606499ad",
"recovery_terminal_sha256": "2e9568c68ba03acfa099039b801c878474a9a5f227da0f2b425c1799b12d673b"
},
"paired_split_counts": {
"0": 304,
"1": 311,
"2": 308,
"4": 305
},
"paired_visits": 1228,
"unique_parts": 571,
"tests": 21,
"tests_child_cpu_seconds": 5.394119000000001,
"tests_log_sha256": "0004d12baa5819b5cbed21339d9cf033d91a2142df5ffc6ce581dfc9fe05140a",
"entry_state": {
"review.json": false,
"started.json": false,
"predictions-manifest.json": false,
"score": false,
"complete.json": false
},
"targets_decoded": false,
"model_values_loaded": false,
"GPU_query": false
}
@@ -0,0 +1,63 @@
{
"reviewed_commit": "28236c5df91329fa52ef98a584688496d2b4e272",
"contract_sha256": "cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63",
"implementation_proof_sha256": "d9ff47cb647cb5e7af6ad7f314a1ad9969b639d1063a5d135b2fd4d73b2a7804",
"comparison_contract_sha256": "8f0f3c1088d899b85aa3f578b7552f04cb409fd7a5dcfca3e17eaa3d7e25d0d6",
"code_sha256": {
"/home/frank14f/optane/MeltMind/tools/boundary_distance.py": "a45c1cbce92521a96eb814913b4f465b926c0b3883932ab0b7c37d11db113ed7",
"/home/frank14f/optane/MeltMind/tools/freeze_g1_components.py": "d0cee60d95243ac2c59ef3f2244ba818315718c4e3d1895f2f9f064f1f7f5ecd",
"/home/frank14f/optane/MeltMind/tools/gre24_features_v2.py": "7dcd5ac218ffdc1ca149edcd87e8658fb3c59e18c0fb9ce47985c7f1357d2cdf",
"/home/frank14f/optane/MeltMind/tools/gre24_models.py": "53b6fee8b700b9b982b06d29908ce2ce23e0c57512c46dbf74b2aade79e3ac3a",
"/home/frank14f/optane/MeltMind/tools/phase2_gate_outcome_review.py": "915617aeb8292d03c654e637bc34a014c254fd178a773f135d1806c4b41372a8",
"/home/frank14f/optane/MeltMind/tools/phase2_h1_worker_v2.py": "1543df85b615bb69b79bacd572074a7c77ae19dab110197601573a7e5805ed20",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_data.py": "7e03dd7a94e7bb43593277f183c222473d7cb7aaf2f60e53b45ab8d9499fcb09",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_model.py": "fab2470dd0654ca74c3b7d893f35ad1f3ae6c5c08ffe334ef6312cf6ec87892a",
"/home/frank14f/optane/MeltMind/tools/phase2_home_freeze_io.py": "bf09204b85ed2a7038acdd202257f618474d1110a9d6d772f105a3ffb334d3ff",
"/home/frank14f/optane/MeltMind/tools/phase2_local_freeze_io.py": "60adf3382577613c5234f5cf30779336b9951b66e90c075cb7cd0adab9cc42dd",
"/home/frank14f/optane/MeltMind/tools/phase2_residual_math.py": "ffbd9624dec79bdaabf95244fe87209055908f31a977e73ea4377045d6f8fe62",
"/home/frank14f/optane/MeltMind/tools/phase3_item1_postfit_data.py": "de9cbcc147677c9144c706acbe52c0f7d505dad17d741d86dcbd04d052f940e2",
"/home/frank14f/optane/MeltMind/tools/prepare_track_a_postfit_rebind.py": "e5fcb0c1ca594aa74e6c3cb70dada60f79baa0d8bcaa0622b19da191ad596536",
"/home/frank14f/optane/MeltMind/tools/r4bc_common.py": "27543fff987975ece767167f9771e6621c825035a053805604e4a696794dfcbd",
"/home/frank14f/optane/MeltMind/tools/r4bc_guard.py": "437a756ddf10b125dce450f463e76e28df52724f8c5766e0a989dfa536805628",
"/home/frank14f/optane/MeltMind/tools/s2_gpu_idle_gate.py": "e2b1613e7e6ea398a8d555220a9365a1ad998ab4ef4b2bf8ec7cc4e31f8b4b9f",
"/home/frank14f/optane/MeltMind/tools/s2_main_memory_guard.py": "e094f5443e11518efe9018215c11718a8e18aa462a977bf43e974b2d0b2aebb1",
"/home/frank14f/optane/MeltMind/tools/s2_pilot_train.py": "e9cceead51a5cc092aad23409ad98909c84a62e7e10571e750d1e9be1ad3b948",
"/home/frank14f/optane/MeltMind/tools/s2_precision_continuation_guard.py": "01235d897c7c03e5ffa109c58b14f5d24d128e8f1ba873e1f76cfb14e02f7a7f",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_gpu_supervisor.py": "689c0f3179e4fcd362e546d90214829a9e4f8739abdbdf74b560763e5a392f15",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_monitoring.py": "442c2a9c75e8c00b30a8f2e91cb52d87c466b19a80aea9128b10c021e61b922e",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_supervisor.py": "f563ca3c2665f5baf0c04f296616e4640423a59b815fbb00479af0f29e57a919",
"/home/frank14f/optane/MeltMind/tools/test_track_a_postfit_rebind.py": "aefdab5d2bcb0e4cf63b7ebe4e736307726f2bd8f98dc2985cc85b0d814315b3",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_common.py": "79918e7de887d1f0a2de5523156b27d00dc2d85ac930a07efc34d817d904fe65",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_prepare.py": "f657be036dc48fc5936f56c4b8510a4b295d4787af6877cc5ccaad663107be17",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_common.py": "8b7d425ab702eb033d4027d24d2812042b66411943b004d6be08a13d4ffcd1c9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind.py": "54389c854f8ee43bdf66834675c22b5e705446cab363ebf55d340d8e3c79381b",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_common.py": "b0f283577da25142201c6d2c7fbfdc486e0f21a75a554a6dd8c2a1da58d29778",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_guard.py": "0aaacc9518041235eecb7f078cebba6d0bc62dadfbd4951f0066bd8271db14a8",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_statistics.py": "07b1d26dbdb898d3ffe26bec497204a59226749d299d35ad55b074a4e0ad03e9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_worker.py": "e34dbcd50bd91b9f84367b61f7568e0cdac471e9e1262a082bb9cfba80b79751",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_statistics.py": "b7d346d1762d7737eb1ebd723989a71f8c65b597d4cdb8c5d512b48a0fd984ce",
"/home/frank14f/optane/MeltMind/tools/track_a_recovery_common.py": "a932d069340909465a5a1e66be1504e28c8df159a5ce304b0586295003d2e734"
},
"verdict": "PASS",
"reviewer": "MeltAssist",
"scope": "Independent pre-score review of the fixed Track A four-of-five rebind under Claude 1795022093.308479.",
"evidence_sha256": "b1d574d7a7067abd7106e09d398cad320e046d29df1c70ba6b8d7608a3a1fcdd",
"audit_source_sha256": "6c61a054c942dec06842b57d7646c13d474abd8a1137857d5fb27461d5d098c6",
"findings": [],
"passed_checks": [
"Verified all five binding fields, 33 code pins, 36 source pins and 57 local Git objects. Independently checked the recovery terminal and incomplete-models freeze bytes.",
"Track A training processes have exited. Model and learning normalization hashes match the three original splits and recovery split 4. Checkpoint admission checks seed 0, last epoch 200, exact optimizer steps and the appropriate source training contract.",
"Split 3 stopped at its first-epoch budget gate and has no completed worker result. The explicit omission is carried through prediction manifest, score and terminal summaries.",
"All 1,228 original A prediction hashes, native node identities and four original test lists match. The test lists cover 571 unique parts and retain disclosure of the 14 original preparation failures.",
"Inference loader has no target branch. Scoring first validates the actual new review, frozen model adoption and frozen complete prediction manifest, and checks the full paired support. Each unique truth array is logged before reading and used across its visits.",
"Per-visit full-node RMSE/max(full truth) is unchanged. Mean visit difference within part and equal weighting across parts use the original 2,000 replicate percentile estimator with seed 2026111408. Per-split summaries are averaged across four splits, with no significance gate.",
"Training cost 78579.79901044304 GPU seconds reconciles the original and recovery ledgers once. GPU2 idle gate and original 900 GPU-second/300 CPU-second per-prediction limits, total 7200 GPU/3600 CPU postfit limits and Track A total 162000 GPU seconds remain in the entry.",
"Independently ran all 21 CPU synthetic checks, including exact support, mixed source identities and missing review or prediction freeze rejection."
],
"limitations": [
"Four completed splits only. Four-split scores would be descriptive on fixed available support; split 3 absence reflects its budget stop.",
"Prediction and score runtime feasibility, actual resource receipts and all future target-access gates remain to be verified by execution.",
"Checkpoint and baseline prediction contents were hash-checked as opaque files. No targets, model values or prediction arrays were decoded. No GPU query, prediction, scoring or training was executed.",
"Old five-split PASS belongs to its previous binding. This receipt is confined to the new exact binding. The local immutable Git bytes were independently verified; publication 482c490a remote readback was reported by the executor."
]
}
@@ -0,0 +1,26 @@
test_all_node_proxy_not_relative_l2 (test_track_a_postfit.PostfitTests) ... ok
test_bootstrap_order_invariant_and_deterministic (test_track_a_postfit.PostfitTests) ... ok
test_bootstrap_zero_difference_degenerate_interval (test_track_a_postfit.PostfitTests) ... ok
test_fixed_last_epoch_identity (test_track_a_postfit.PostfitTests) ... ok
test_missing_visit_not_silently_averaged (test_track_a_postfit.PostfitTests) ... ok
test_no_target_io_before_actual_review (test_track_a_postfit.PostfitTests) ... ok
test_no_target_io_before_predictions_frozen (test_track_a_postfit.PostfitTests) ... ok
test_postfit_reserve_and_training_charged_once (test_track_a_postfit.PostfitTests) ... ok
test_prediction_manifest_native_id_and_duplicate_rejection (test_track_a_postfit.PostfitTests) ... ok
test_same_paired_support_five_summary_means (test_track_a_postfit.PostfitTests) ... ok
test_test_loader_rejects_foreign_case_before_io (test_track_a_postfit.PostfitTests) ... ok
test_undefined_candidates_and_visit_counts_retained (test_track_a_postfit.PostfitTests) ... ok
test_undefined_pairs_excluded_symmetrically (test_track_a_postfit.PostfitTests) ... ok
test_unique_part_bootstrap_not_visit_weighted (test_track_a_postfit.PostfitTests) ... ok
test_zero_truth_is_explicitly_undefined (test_track_a_postfit.PostfitTests) ... ok
test_exact_split_ids_and_visit_counts (test_track_a_postfit_rebind.RebindTests) ... ok
test_four_split_summary_uses_original_part_bootstrap (test_track_a_postfit_rebind.RebindTests) ... ok
test_incomplete_duplicate_or_extra_visits_rejected (test_track_a_postfit_rebind.RebindTests) ... ok
test_mixed_source_checkpoint_identity (test_track_a_postfit_rebind.RebindTests) ... ok
test_scoring_requires_new_review_before_target_access (test_track_a_postfit_rebind.RebindTests) ... ok
test_scoring_requires_prediction_freeze_before_target_access (test_track_a_postfit_rebind.RebindTests) ... ok
----------------------------------------------------------------------
Ran 21 tests in 0.025s
OK
@@ -0,0 +1,4 @@
{
"contract_sha256": "cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63",
"proof_sha256": "6e53122402b31c506db058f08e6a125b69ad94d1df021854855896fd382e60e1"
}
@@ -0,0 +1,23 @@
{
"commit": "958abd8c916614c5366b5f0216958d46e00b8504",
"files": {
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/contract.json": "cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63",
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/expected-review-binding.json": "96e2b0c0c7ccd262ac15f9ba08cfa29d7a01c312e4f8633a1eda56184f61c6df",
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/review.json": "a5fa4a6e040493ff187b084e07cba0d95e3cae8f0fc0fa0d571f4af9efd30fce",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/review.json": "a5fa4a6e040493ff187b084e07cba0d95e3cae8f0fc0fa0d571f4af9efd30fce",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/evidence.json": "b1d574d7a7067abd7106e09d398cad320e046d29df1c70ba6b8d7608a3a1fcdd",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/audit.py": "6c61a054c942dec06842b57d7646c13d474abd8a1137857d5fb27461d5d098c6",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/synthetic-tests.log": "0004d12baa5819b5cbed21339d9cf033d91a2142df5ffc6ce581dfc9fe05140a"
},
"git_paths": {
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/contract.json": "home-runs/track-a-postfit-four-split-2026-11-19/contract.json",
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/expected-review-binding.json": "home-runs/track-a-postfit-four-split-2026-11-19/expected-review-binding.json",
"/home/frank14f/MeltMind-runs/track-a-postfit-four-split-2026-11-19/review.json": "home-runs/track-a-postfit-four-split-2026-11-19/review.json",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/review.json": "artifacts/meltassist-track-a-four-split-review-2026-11-19/review.json",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/evidence.json": "artifacts/meltassist-track-a-four-split-review-2026-11-19/evidence.json",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/audit.py": "artifacts/meltassist-track-a-four-split-review-2026-11-19/audit.py",
"/home/frank14f/optane/MeltMind/artifacts/meltassist-track-a-four-split-review-2026-11-19/synthetic-tests.log": "artifacts/meltassist-track-a-four-split-review-2026-11-19/synthetic-tests.log"
},
"remote_published": false,
"local_repository": "/tmp/meltmind-phase2-local-fixed-20261117"
}
@@ -0,0 +1,15 @@
{
"status": "PASS",
"independent_reviewer": "MeltAssist",
"adopted_by": "NewMeltAssist",
"review_sha256": "a5fa4a6e040493ff187b084e07cba0d95e3cae8f0fc0fa0d571f4af9efd30fce",
"binding_sha256": "96e2b0c0c7ccd262ac15f9ba08cfa29d7a01c312e4f8633a1eda56184f61c6df",
"review_freeze": "958abd8c916614c5366b5f0216958d46e00b8504",
"implementation": "28236c5df91329fa52ef98a584688496d2b4e272",
"claude_acceptance": "1795023230.034749",
"reviewer_message": "1795023210.064739",
"native_receipt": true,
"launched": false,
"test_targets_read": false,
"unix": 1795023363.5219464
}
@@ -0,0 +1,63 @@
{
"reviewed_commit": "28236c5df91329fa52ef98a584688496d2b4e272",
"contract_sha256": "cc00b599f584d722d246c94dd5b7db9ecc082adf10581c4b32514e55c72c6e63",
"implementation_proof_sha256": "d9ff47cb647cb5e7af6ad7f314a1ad9969b639d1063a5d135b2fd4d73b2a7804",
"comparison_contract_sha256": "8f0f3c1088d899b85aa3f578b7552f04cb409fd7a5dcfca3e17eaa3d7e25d0d6",
"code_sha256": {
"/home/frank14f/optane/MeltMind/tools/boundary_distance.py": "a45c1cbce92521a96eb814913b4f465b926c0b3883932ab0b7c37d11db113ed7",
"/home/frank14f/optane/MeltMind/tools/freeze_g1_components.py": "d0cee60d95243ac2c59ef3f2244ba818315718c4e3d1895f2f9f064f1f7f5ecd",
"/home/frank14f/optane/MeltMind/tools/gre24_features_v2.py": "7dcd5ac218ffdc1ca149edcd87e8658fb3c59e18c0fb9ce47985c7f1357d2cdf",
"/home/frank14f/optane/MeltMind/tools/gre24_models.py": "53b6fee8b700b9b982b06d29908ce2ce23e0c57512c46dbf74b2aade79e3ac3a",
"/home/frank14f/optane/MeltMind/tools/phase2_gate_outcome_review.py": "915617aeb8292d03c654e637bc34a014c254fd178a773f135d1806c4b41372a8",
"/home/frank14f/optane/MeltMind/tools/phase2_h1_worker_v2.py": "1543df85b615bb69b79bacd572074a7c77ae19dab110197601573a7e5805ed20",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_data.py": "7e03dd7a94e7bb43593277f183c222473d7cb7aaf2f60e53b45ab8d9499fcb09",
"/home/frank14f/optane/MeltMind/tools/phase2_h2_model.py": "fab2470dd0654ca74c3b7d893f35ad1f3ae6c5c08ffe334ef6312cf6ec87892a",
"/home/frank14f/optane/MeltMind/tools/phase2_home_freeze_io.py": "bf09204b85ed2a7038acdd202257f618474d1110a9d6d772f105a3ffb334d3ff",
"/home/frank14f/optane/MeltMind/tools/phase2_local_freeze_io.py": "60adf3382577613c5234f5cf30779336b9951b66e90c075cb7cd0adab9cc42dd",
"/home/frank14f/optane/MeltMind/tools/phase2_residual_math.py": "ffbd9624dec79bdaabf95244fe87209055908f31a977e73ea4377045d6f8fe62",
"/home/frank14f/optane/MeltMind/tools/phase3_item1_postfit_data.py": "de9cbcc147677c9144c706acbe52c0f7d505dad17d741d86dcbd04d052f940e2",
"/home/frank14f/optane/MeltMind/tools/prepare_track_a_postfit_rebind.py": "e5fcb0c1ca594aa74e6c3cb70dada60f79baa0d8bcaa0622b19da191ad596536",
"/home/frank14f/optane/MeltMind/tools/r4bc_common.py": "27543fff987975ece767167f9771e6621c825035a053805604e4a696794dfcbd",
"/home/frank14f/optane/MeltMind/tools/r4bc_guard.py": "437a756ddf10b125dce450f463e76e28df52724f8c5766e0a989dfa536805628",
"/home/frank14f/optane/MeltMind/tools/s2_gpu_idle_gate.py": "e2b1613e7e6ea398a8d555220a9365a1ad998ab4ef4b2bf8ec7cc4e31f8b4b9f",
"/home/frank14f/optane/MeltMind/tools/s2_main_memory_guard.py": "e094f5443e11518efe9018215c11718a8e18aa462a977bf43e974b2d0b2aebb1",
"/home/frank14f/optane/MeltMind/tools/s2_pilot_train.py": "e9cceead51a5cc092aad23409ad98909c84a62e7e10571e750d1e9be1ad3b948",
"/home/frank14f/optane/MeltMind/tools/s2_precision_continuation_guard.py": "01235d897c7c03e5ffa109c58b14f5d24d128e8f1ba873e1f76cfb14e02f7a7f",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_gpu_supervisor.py": "689c0f3179e4fcd362e546d90214829a9e4f8739abdbdf74b560763e5a392f15",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_monitoring.py": "442c2a9c75e8c00b30a8f2e91cb52d87c466b19a80aea9128b10c021e61b922e",
"/home/frank14f/optane/MeltMind/tools/s2_recovery_supervisor.py": "f563ca3c2665f5baf0c04f296616e4640423a59b815fbb00479af0f29e57a919",
"/home/frank14f/optane/MeltMind/tools/test_track_a_postfit_rebind.py": "aefdab5d2bcb0e4cf63b7ebe4e736307726f2bd8f98dc2985cc85b0d814315b3",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_common.py": "79918e7de887d1f0a2de5523156b27d00dc2d85ac930a07efc34d817d904fe65",
"/home/frank14f/optane/MeltMind/tools/track_a_h2_prepare.py": "f657be036dc48fc5936f56c4b8510a4b295d4787af6877cc5ccaad663107be17",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_common.py": "8b7d425ab702eb033d4027d24d2812042b66411943b004d6be08a13d4ffcd1c9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind.py": "54389c854f8ee43bdf66834675c22b5e705446cab363ebf55d340d8e3c79381b",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_common.py": "b0f283577da25142201c6d2c7fbfdc486e0f21a75a554a6dd8c2a1da58d29778",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_guard.py": "0aaacc9518041235eecb7f078cebba6d0bc62dadfbd4951f0066bd8271db14a8",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_statistics.py": "07b1d26dbdb898d3ffe26bec497204a59226749d299d35ad55b074a4e0ad03e9",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_rebind_worker.py": "e34dbcd50bd91b9f84367b61f7568e0cdac471e9e1262a082bb9cfba80b79751",
"/home/frank14f/optane/MeltMind/tools/track_a_postfit_statistics.py": "b7d346d1762d7737eb1ebd723989a71f8c65b597d4cdb8c5d512b48a0fd984ce",
"/home/frank14f/optane/MeltMind/tools/track_a_recovery_common.py": "a932d069340909465a5a1e66be1504e28c8df159a5ce304b0586295003d2e734"
},
"verdict": "PASS",
"reviewer": "MeltAssist",
"scope": "Independent pre-score review of the fixed Track A four-of-five rebind under Claude 1795022093.308479.",
"evidence_sha256": "b1d574d7a7067abd7106e09d398cad320e046d29df1c70ba6b8d7608a3a1fcdd",
"audit_source_sha256": "6c61a054c942dec06842b57d7646c13d474abd8a1137857d5fb27461d5d098c6",
"findings": [],
"passed_checks": [
"Verified all five binding fields, 33 code pins, 36 source pins and 57 local Git objects. Independently checked the recovery terminal and incomplete-models freeze bytes.",
"Track A training processes have exited. Model and learning normalization hashes match the three original splits and recovery split 4. Checkpoint admission checks seed 0, last epoch 200, exact optimizer steps and the appropriate source training contract.",
"Split 3 stopped at its first-epoch budget gate and has no completed worker result. The explicit omission is carried through prediction manifest, score and terminal summaries.",
"All 1,228 original A prediction hashes, native node identities and four original test lists match. The test lists cover 571 unique parts and retain disclosure of the 14 original preparation failures.",
"Inference loader has no target branch. Scoring first validates the actual new review, frozen model adoption and frozen complete prediction manifest, and checks the full paired support. Each unique truth array is logged before reading and used across its visits.",
"Per-visit full-node RMSE/max(full truth) is unchanged. Mean visit difference within part and equal weighting across parts use the original 2,000 replicate percentile estimator with seed 2026111408. Per-split summaries are averaged across four splits, with no significance gate.",
"Training cost 78579.79901044304 GPU seconds reconciles the original and recovery ledgers once. GPU2 idle gate and original 900 GPU-second/300 CPU-second per-prediction limits, total 7200 GPU/3600 CPU postfit limits and Track A total 162000 GPU seconds remain in the entry.",
"Independently ran all 21 CPU synthetic checks, including exact support, mixed source identities and missing review or prediction freeze rejection."
],
"limitations": [
"Four completed splits only. Four-split scores would be descriptive on fixed available support; split 3 absence reflects its budget stop.",
"Prediction and score runtime feasibility, actual resource receipts and all future target-access gates remain to be verified by execution.",
"Checkpoint and baseline prediction contents were hash-checked as opaque files. No targets, model values or prediction arrays were decoded. No GPU query, prediction, scoring or training was executed.",
"Old five-split PASS belongs to its previous binding. This receipt is confined to the new exact binding. The local immutable Git bytes were independently verified; publication 482c490a remote readback was reported by the executor."
]
}