Independent catalog audit · 12 November 2026
Two upstream architectures have completed the current AGILLM simulator training: #458 DLGM and #879 Fusion. The other 103 still need that comparison. Earlier CPU screening and zero-step operator checks are separate evidence.
No new winner or training result is claimed here. This is a coverage correction and an explicit continuation plan. The complete per-ID ledger separates prepared ports, trained ports and historical CPU screens.
The 8–9 November progress reports really did say 22 of 105, for the older small CPU Wiki/FineWeb campaign. That campaign later reached all 105 source records with 146 runs. Those AR-only byte-model results do not replace the current binary AGILLM AR/SAT/NAT experiment. Exact membership of the older 22 cannot be recovered from the later 25-entry pilot ledger.
The current records contain 89 eligible fixed-budget runs across 32 plugin/control variants, but only two mapped upstream source IDs. The 105 original-module GPU checks used zero training steps and are not quality benchmarks.
The pinned export contains 105 unique IDs and program hashes. A separate 103-total source list has not been found. Here, 103 remaining = 105 canonical records minus #458 and #879. Every ID is accounted for below; no arbitrary records were excluded.
Current per-ID campaign ledger and run evidence · Source-specific continuation plan · Pinned upstream export
Snapshot: 2026-11-12T04:53:37.029233+00:00. “Trained” means an adapted small AGILLM port, not the original full-size model. Every row also has historical AR-only CPU screening.
105 records shown
| Source | Current AGILLM status | Original program SHA256 |
|---|---|---|
| #151 AdaptiveMixTransformer | Not yet trained | ee3208a7e4db…ee3208a7e4db32832b18c3bc2104316db1f43015b57f90b709ad5f677c4d4c70 |
| #259 BAMG_MemoryGate | Prepared; not trained | 08e3b4bb618c…08e3b4bb618c64be48db8cb402ab3dc46cbc1a204741710f25afb6fedc3b699e |
| #295 DynamicMemGateNet | Not yet trained | dcadb6973fe5…dcadb6973fe58db89d70effe1c0148e26a9c1f5182e1355cd1914a8129d08245 |
| #322 AdaptiveTokenGate | Not yet trained | 5bea139d6e96…5bea139d6e96579a9bdf0f76f19391c45e50e520a7b2380ddb8c86f45f3b9453 |
| #330 AdaptiveFractalGateNet | Not yet trained | 5892c566d4c8…5892c566d4c8a20b1899cfd16db573fe411acd30f7b7c0db98d93b32cf31cb5d |
| #380 DualScaleGateNet | Not yet trained | ac1bd8484d80…ac1bd8484d8007ed65db71de778d5be2f0871c9c53efd9f7ed2fdf5a9999947a |
| #410 HybridScale-GateNet | Not yet trained | 5983491fb298…5983491fb298ff661ccdb1c64d2d7bc978890884a92ba6aca11139fa78852f25 |
| #437 FusionConv-AMG | Not yet trained | da307a09b49d…da307a09b49d35c30637a27a2d22f4435a2b1bfcab7607fef34be8afd4aa74a7 |
| #441 DualFIR-QuadFusion | Not yet trained | bef7cb97b683…bef7cb97b6833b74281446dbdec0bc3ccfd6565acfb9de61e2a7a6923d02365a |
| #458 DualScaleMemoryRouter | Trained adapted port | 76412d2f7aff…76412d2f7aff30f99ce7089e0f57f8ef26add7f2d9f965adf8650a00538bd1f5 |
| #486 DepthwiseConvPointMixer | Not yet trained | 6d7f4a8a297a…6d7f4a8a297a0131567dc1d6aac04b7c43bca80584bb0f12e7477643ebd80ddd |
| #511 FusionGateMemoryNet | Not yet trained | b2e21ef4e7c6…b2e21ef4e7c62ebb3ea655cb0957f2a89d3759000ea4ca268176202bef2c1e40 |
| #515 FusionGate-MS3 | Not yet trained | c99cd7cac8d3…c99cd7cac8d3552283675ed13e6cda623786c0fc7fbcfe788c4cf72cb617f832 |
| #518 TriScale-GatedFusion | Not yet trained | 6f588e4282cf…6f588e4282cf72bf47befd95ddd21727e042f6868c1a0ae4343347571f077b7b |
| #519 SparseGateDelta | Not yet trained | c353803f2ab6…c353803f2ab6ca5a8c767d1522785f359c5f894618e4a2a9ba86a64dd9bdd92a |
| #537 FusionGateNet_v3 | Not yet trained | d8ce9221a182…d8ce9221a182be08dc86f1b48ad9d3a62d92bddca276e0d079e497e31eda784a |
| #556 PathAwareMemoryRouter | Not yet trained | 65eae8f2ca90…65eae8f2ca90ff4ad30f002e8ff061da3aaaa8ee578c6e4ae3c4ea748bdc6573 |
| #559 OutputAwareMultiScaleRouter | Not yet trained | c3b0ba6d8dc0…c3b0ba6d8dc01d1119ff3a9523bde6b8d53330f5f04ea7508ae763174ecfff19 |
| #562 SparseTempGateNet | Not yet trained | 9037a14645ed…9037a14645ede30ea7f9f9d5b8a1ee10523cd5eccf033b60657b8f0a6d9764ee |
| #565 FusionGate-X | Not yet trained | b649fc8e348d…b649fc8e348d600f25dc1d28e51ad8ccdded8ffb20be291e02bfb071492309eb |
| #629 ConvFusionWide31 | Not yet trained | 38b568fbc9f5…38b568fbc9f52b479b9cc9d2c2313682c93808593421070ad8638c82064ab6d8 |
| #641 HeadWiseGateNet | Not yet trained | c970077bad63…c970077bad63f5829568cbcbe1019f91013416b2f10d634e363e421bc4077691 |
| #645 AdaptiveHierGateNet | Not yet trained | a4f6820d3042…a4f6820d304293f7880590a95597b3406c11aad003c34e77bc2d146195a05352 |
| #657 QuotaGatedStatNet | Not yet trained | 7444d5b8ce5a…7444d5b8ce5aa1d64609105f555db8961c1e9f54cdc7c6aa088b989bb1b2c75f |
| #665 HyperRouteFusion | Not yet trained | 0d0a99322e56…0d0a99322e56e347c72b0a4f2fd16efe033f78efe49f93f98d338046678ba262 |
| #675 FusionGate-MS3E-Hybrid | Not yet trained | 748409c3c30d…748409c3c30d5705f4bb8355814ce48b4a70812fe9f9873f595d96c8129ec75a |
| #676 StatGateRouter | Not yet trained | 99e617af33ee…99e617af33ee81d813ef0c26e1562fe19ee3c7cf382b4085dfc21e001183df5a |
| #700 FusionGate-XR | Not yet trained | 9478e7f544ab…9478e7f544ab042799bf23c32cf458059129b9d992c385a5594ce9cb4c3e2144 |
| #702 BlockStateFusionNet | Not yet trained | 631c0694bd58…631c0694bd588a5427d40e5ad9da1fd2dd5f282f71ac8d8b1124cac20058b99e |
| #703 ContentSharpRouter | Not yet trained | 65fad929731e…65fad929731e9eaac7a67a42088249de95c66edb3391071e55e144f5e1b22017 |
| #709 HierGate-MEM | Not yet trained | ef224861172a…ef224861172acb1272a822bac2ceef713f2a2c13c0de2af4d798055fbf4df311 |
| #712 AdaptiveGateNet-AFP | Not yet trained | 62d59f5bf7f1…62d59f5bf7f19ba502343f2c1aeba41311659f981660cd0feb37a58940451b7e |
| #714 ResGate_MS_FusionNet | Not yet trained | 7fd95751c2c7…7fd95751c2c727ef5ee232556dc9f95bf30317d5b3141011a7af2856673ee7e9 |
| #720 HybridSparseGateMemoryNet | Not yet trained | 711e07f46db8…711e07f46db86c1fc8058041c69adfcca79299492337ee07c86ba095c8c6516a |
| #750 AdaptiveSpanGateConv | Not yet trained | 58ca4270916f…58ca4270916f4e4505df4704218c09a01a7e9ee7c2a9462ce368e5cd43f6b275 |
| #804 PerHeadSimplexRouter | Not yet trained | 0351c824749a…0351c824749a39ec9f76f16c4eaa7540038c4ce5f81179f1c7687b851827a584 |
| #814 GateDivergeTransformer | Not yet trained | d2f18b906008…d2f18b9060087d151bdd407f4bb339d41c943ec0319b23668929c05778c4a8e9 |
| #828 FusionGate AdaptiveNet | Not yet trained | 978892861788…978892861788996b13c5ef7072bfe15e41485e1bc25217cfca667a3df07ac59d |
| #830 LocalGlobalBlendNet | Not yet trained | 22f641d51ee2…22f641d51ee2c83e070a5c4a4d1206a76223b283e53a3295ebd25728d612e608 |
| #843 DualStagePathGateNet | Not yet trained | 8beebf88634c…8beebf88634cf51fb6b0e3b35f342c5b75cc64341f2f8f40103708576c12ac84 |
| #865 DynFuseFlexGate | Not yet trained | c4a395c2285f…c4a395c2285f3245545a95fc50c1a76be75f19316882e4e80f0441eede1c5e4f |
| #879 FusionGatedFIRNet | Trained adapted port | 771d5c670fee…771d5c670feeb0c70df9d9817ec01b3f72c85fefd4483fff5b10c4e1978a5f80 |
| #887 EntropyFlowGateNet | Not yet trained | 5bcba408a97d…5bcba408a97dd3481d46a726511ff748f800bdddfca29401b5e318be1fb68937 |
| #890 FusionGATE-HMSR | Not yet trained | b4f6dc85edf2…b4f6dc85edf28f82fe0624e81aaf95ae527d06366e64e272419e1b11eff52605 |
| #902 EntropyEnhancedMultiScaleGateNet | Not yet trained | e161833544d1…e161833544d18c36987b4541978160da6604e3897cfc8e08226a74d1cabe484a |
| #908 AdaptiveFusionNet | Not yet trained | 00c192fe4a64…00c192fe4a64811033bd0934fdbbcccc374b1c48066ad96b5963126591f653be |
| #922 ParallelFusionTransformer | Not yet trained | fd8824211f93…fd8824211f93a162d47cb68bd88bb9a146c4a8cf0b90b8e204b59b5e27ce01dd |
| #927 AdaptiveFusionRNet | Not yet trained | cdbc51b0175e…cdbc51b0175ed3b73390fd335c621ab6f38e7211f27baa28de862a5ae9252700 |
| #933 FusionGateX | Not yet trained | f19b9e402e3d…f19b9e402e3d91fcbdc2f5ebdb556c04b26602a3e62a32f602a15d1afe154eca |
| #939 EntropyKLAdaptiveGateNet | Not yet trained | a2dbfcb4f2f4…a2dbfcb4f2f4eaf018161c1279fcc56b3a44ed6f017c37445ef9348246630636 |
| #947 GroupTempMLP | Not yet trained | 336a6ad01332…336a6ad0133210c4fdcf6f7f95bf311a4fa236befdbf2af3142a997d076dd602 |
| #948 AdaptiveFloorGate | Not yet trained | 9b9249aa5842…9b9249aa584228c828aff4c0edb0de4999878d74e5625b3298a4a2f7b8cebdac |
| #954 AdaptiveGateNet | Not yet trained | 944982086a0c…944982086a0c1a70ff992dc377bfecdf235b640eaa49d2d4c91580b00de69837 |
| #957 FusionBalanceTransformer | Not yet trained | ee330e9e2190…ee330e9e2190c7aefa356d6780c820bc16494c7eb5eeb8911759fd843adfed01 |
| #965 GateFlooredResNet | Not yet trained | 3fc13e8a45e7…3fc13e8a45e7fc4138ccf3f211c1651ad7e0bdf271f3610741b478fb14141320 |
| #974 AdaptiveFusionNet | Not yet trained | bcdeb59778d3…bcdeb59778d369db113d1e1fa185da32f10762050f02014de9fd1dc701a3f0fe |
| #988 LexiFuse-Percept | Not yet trained | a0ea996022cb…a0ea996022cbb1987334b0c598d52862143f655cddf8c51cb12d9a21433303c5 |
| #1017 FusionLogicNet | Not yet trained | a4e6619d848c…a4e6619d848c3a044b91afbc6a99564326ca51a6c9e5399ad3a12ac8fac66c69 |
| #1031 FusionGate-X | Not yet trained | d3211189e71f…d3211189e71f914c4005378f653e7bf0559435b94df41b7bec21bcf7ffdee5d1 |
| #1041 ResConvGate | Not yet trained | 48c552937a51…48c552937a51ad86b59300481795ba2bd6a9d766f4b58c7c9d50f65cbd460158 |
| #1045 AdaptiveHybridGateNet | Not yet trained | 3d83473fd95a…3d83473fd95a7acbf6766ae99b54ef8c6d39b085957b5ab33b7408c18be8134d |
| #1051 ResiFuse-CausalGater | Not yet trained | f57fbefd781a…f57fbefd781a1040f1762ce42df83d776facf1bdf74e45a43a48575a6623f64f |
| #1081 EntropyFusionNormX | Not yet trained | 95769712260f…95769712260f1da987c5139b257410dfcb1df9d4e81d0f2515480585d3505e72 |
| #1132 DualScaleStatFusionNet | Not yet trained | d1688bccaa32…d1688bccaa32d7e60de5bd870a53365ead51c95c953b8d28b4e714bfa844a9ac |
| #1155 TriScaleFusionNet | Not yet trained | 5894f5c19d25…5894f5c19d258dff4976ead226be58cabe8f2ff163dc30bd81028c64c965c527 |
| #1167 GatedFusionTransformer | Not yet trained | a7110317b35b…a7110317b35bd0813715f948882c258d8febdb08c2433841be8d7609235cd631 |
| #1175 AnnealedPathFusionNet | Not yet trained | 8502465628e5…8502465628e52acb9aff5ff92278195ffa63cd89360c8bfb36a6fc482d8bc3b7 |
| #1195 BoundedTempAnnealNet | Not yet trained | c389bf967371…c389bf967371a9762bb2dd9bd1193ec27d6a439016e31c84ad23fb372c009ff7 |
| #1221 FusionGateX | Not yet trained | 4914648a6662…4914648a666289b4bf2fdafa1084a4065df68d3bfa4f64afb117b8ec8ce4f08a |
| #1235 MinFloorRouter | Not yet trained | 5dd7fd063d5f…5dd7fd063d5f37e7fc22e4639ac82c559d31613747d9df13d4d99d5bfde2f2e1 |
| #1252 AdaptivePathRouter | Not yet trained | 0cf729c24611…0cf729c24611666aafc732b0a27829afc41d4c8cbb178c034aa57a4f7180d1d7 |
| #1283 HybridFusionFloor | Not yet trained | 5b3238b839b0…5b3238b839b074b4983f0983c0031fed479bcf43e444b8d9b698111bbdfdb109 |
| #1298 AdaptiveEntropyRouterNet | Not yet trained | bd894f9e443b…bd894f9e443bf26fd0089bbb3e5038a4fbd5187612065b36b2534ccfcf4b49a5 |
| #1311 HybridFlowNet | Not yet trained | 8ed3f3858b70…8ed3f3858b70f781601dc71c23a4645312404c1ff6f5dab0728d8a6ae02b48fc |
| #1338 ConvexBlendFloorNet | Not yet trained | 722bb5685a42…722bb5685a42fa76be7a0735009c8eff85fd1b23ab07eba44c37ef5957fe7b89 |
| #1349 FusionFeedback-MixNormNet | Not yet trained | f123e8718439…f123e8718439a62fb38a38e919f7487cd719eddb10b5c00a37c925664c01fcea |
| #1350 HybridCausalRouter | Not yet trained | fd539786a131…fd539786a1316df385a14ceb96b99431b61ae3f83828d6293f3369a25491b1e5 |
| #1354 FusionGate-MS | Not yet trained | 9acc1dd8c77c…9acc1dd8c77ca4e2b336721434ceb7eb4afe323cf08fb8018dba604f216a98fd |
| #1360 AdaptiveFloorNet-HAF | Not yet trained | 4e74bdb28884…4e74bdb288844d204b41f13c6ade1d4ac94e49b17428a9ea366ea5486c7f5b5e |
| #1370 AdaptiveTokenRouter | Not yet trained | 15a316f442c3…15a316f442c3365477da3bde4381a4fa9e69ebac0b595fe5821b73122acfd99d |
| #1393 SynerFuse-LGX | Not yet trained | d96ab556464e…d96ab556464ea98c7944e34b72a4d33421e069713d6c1e4d54477547df0ab573 |
| #1396 FusionGate-CAGT | Not yet trained | 870af837f43f…870af837f43fd2032ffda8bdbbb104600bb8b196024b7e2826b3c719501f52cc |
| #1408 AdaptiveEntropyGateNet | Not yet trained | c586702e1fe6…c586702e1fe6190abd9933baffdb68b5801946148aaaadb8da57f963a5ec7eb7 |
| #1434 HierarchiMix-Gate | Not yet trained | 376c42e93d75…376c42e93d7596a3473b20232f2b3f666b3a36a51d98600b2722fe388e535797 |
| #1442 DynMemGate | Not yet trained | 8e613f93c3eb…8e613f93c3eb9bff150f34b418f1180433733263959a97be1c68ce6a92bb60c9 |
| #1444 FusionGate-XL | Not yet trained | dd5268d7c13a…dd5268d7c13a01e9878c8ef28a79573fcb49ecbbb39a14ea7e3689f1e635470b |
| #1448 PathGatedFusionNet | Not yet trained | 2c40b7b09f16…2c40b7b09f16b75ae1dc90d24461c8c05637ad5f6df0743d9083befe53742d3d |
| #1454 TokenPruneRouter | Not yet trained | 5639a72f1be3…5639a72f1be3bd8df342ac4fb9b234215bc6efe90fc7cff91f7944a10ad7ca2a |
| #1475 TempMixAnnealRouter | Not yet trained | c7a242b1cb18…c7a242b1cb183067e65d59dd5127c207cd17b75fa23784d31b4b96a649d67d9e |
| #1490 AdaptiveContextFusionNet | Not yet trained | 0da7cd59cb73…0da7cd59cb7366dc72e77e7cb1c0ec8163a8017ddfab040687c253d5910640ea |
| #1507 TokenScaleRouter | Not yet trained | e710fcbad020…e710fcbad020ca414268cfd63c5ddac21e34ad938c4e59fe09e03c60201d5b44 |
| #1529 HyenaMAFR | Not yet trained | dd0e62ef7008…dd0e62ef7008baa37a32c60db66aa5b3d2d8dc007b2d01ebcaf294104b79388a |
| #1543 FusionGateBR | Not yet trained | 9dbd01c41cec…9dbd01c41cec03ff0434928316b1efcbb1934159b4ff22db365d90b68522669d |
| #1567 StreamAwareRouter | Not yet trained | 952e06f72f6f…952e06f72f6f5bf7b13abc6e859355beebff1ab5ed647b6cbbc447bdf4ae5767 |
| #1584 AdaptiveFusionNet | Not yet trained | cc5a6301a7b2…cc5a6301a7b2ddbe109105214b7cdd1644fd644c6cda631d538cd0983a2a8cf6 |
| #1595 GateFusionNet | Not yet trained | a5b080a24136…a5b080a241365b3bd9dba9084ed9558d82c698f5ae872e51bb3990cc77159a4d |
| #1621 HybridGateTransformer | Not yet trained | 723df3537e6b…723df3537e6b1491852db1fe966c65b839860ed714e90ff75216a662e927ff78 |
| #1693 AdaptiveMixGateNet | Not yet trained | d460870af142…d460870af142c43c82028dfe55d8f7ed7174a6366b6d1de2ccd2f1abbd9bc461 |
| #1730 AdaptiveEntropyRouter | Not yet trained | a4b7710b084f…a4b7710b084fb99a5d0a74c482de2cb5424ad938428f22689e4d6d7f01e718a1 |
| #1733 FusionGate-XL | Not yet trained | b2fcd4ab98f0…b2fcd4ab98f0b744662dbfef654b374e82319d5186f25297377a5102f0ef0b7e |
| #1735 HybridGateFlow | Not yet trained | 648b335c651d…648b335c651d5449928c1c1147471333d15a4223f1b5013d1a69b927dd57a32e |
| #1763 AdaptiveGatedRouter-Hybrid | Not yet trained | 8431a58e9cd6…8431a58e9cd6437833a02cebfcb79e42f77d07aa0d6023a9ba230f5910984595 |
| #1771 AdaptiveGateRouter_X | Not yet trained | e70adafd02da…e70adafd02da750f717122d631e963d3070c4d2b118a4854fbadba757fa16f42 |
| #1786 AdaptiveFusionNet-DSI | Not yet trained | 72cfc0626100…72cfc06261003c2707fc59f0867fff3ce2f675353d4705beb847cb6822ccef33 |
| #1792 FusionGate-HierarchicalRouter | Not yet trained | a0f4faeaf821…a0f4faeaf821f64915ee841cb6ed416f2b8f45a1f771a7a7038f965f486be969 |
The original within-machine results remain intact. This is a portability finding, not a new quality winner.
Full checkpoint verification results · Data hash verification
Preparation update, 05:42 UTC: #259 BAMG_MemoryGate, #641 HeadWiseGateNet and #712 AdaptiveGateNet-AFP now have published CPU correctness bundles. None has a qualifying AGILLM training run. #259 remains first in line for a safe pilot; preparation continues across the remaining source IDs. A shared harness will support bounded batches across all remaining IDs, with fidelity, causality, masking, finite-gradient and memory gates. Failures and unsupported cases will stay visible.
Plan: 1,200 updates and 2,457,600 token targets per seed, fixed AR/SAT/NAT objective and a three-seed pilot per eligible source. Freeze source/config/data hashes and shared initialization before each new cohort. Historical runs remain separate. A decisive claim needs a fixed confirmation cohort and adjusted paired comparisons.
No new GPU run was started for this coverage update. Existing queued experiments and live serving are preserved. The next run needs fail-closed admission and verified capacity. Tiny correctness checks do not count as training, and current ports have no validated cached decoding.
#259 is a cheap prepared hypothesis, not the paper favorite. #641 has stronger paper ranking, but neither has current AGILLM quality results. Reused evaluation arrays will not be called a fresh holdout. Full-scale quality, memory and serving claims remain unverified.
Complete coverage reconciliation, reproducible inputs and continuation protocol
Tested #259, #641 and #712 ports, source references and correctness reports. #641 native-random and practical-Dirac initialization variants are kept separate. #712 passed 41 CPU checks with zero optimizer updates. Correctness preparation adds zero trained architectures.
Versioned code, results and all six checkpoints on GitHub
Full report, reproduction commands and next-round protocol · Repair pull request
Hugging Face: complete gallery package with all six checkpoints. The key source and report files and every checkpoint were downloaded from this commit and SHA-256 checked against the published package.