{ "canonical_sha256_without_self": "00db0569d2fd00599383ecdc50578c52df784a937a96dd69576f1366e88b0dcb", "environment": { "cublas_workspace_config": ":4096:8", "cuda": "12.8", "gpu": "NVIDIA GeForce RTX 5090", "torch": "2.11.0+cu128" }, "gate": "step-zero-cross-variant-byte-exact", "parent_manifest_sha256": "080afb17d1e036c0bba0a799fdb8b98ee4ad652bd42dd1b3b67110dd2ede6371", "passed": true, "protocol_id": "llm-atlas-k3-attnres-forward-training-v1", "schema_version": 1, "seed": 2026073001, "study_manifest_sha256": "49546ed5baf36bcb30885062b7c671fafe4ccff2e606624cd7dbe23717f9a712", "variants": { "learned_reference": { "exact_vs_learned_reference": { "diagnostic": true, "evaluation": true, "initial_mixer_hash": true, "initial_public_hash": true, "logits_sha256": true, "loss_nats": true }, "passed": true, "payload": { "diagnostic": { "activation_grad_rms_by_block": [ 0.0002677410375326872, 0.0002473295899108052, 0.00023034414334688336, 0.00021438254043459892, 0.00020044577831868082, 0.00018819731485564262, 0.00017973134526982903, 0.000171217747265473, 0.0001640921109355986, 0.00016061548376455903, 0.00015463090676348656, 0.00015024171443656087, 0.0001478576596127823, 0.00014403100067283958, 0.00013951722939964384, 0.00013661097909789532, 0.00013379570737015456, 0.0001311724045081064, 0.0001286817860091105, 0.00012687484559137374, 0.0001249351626029238, 0.00012369138130452484, 0.00012227990373503417, 0.00012101328320568427, 0.00011953213106608018, 0.0001181897969217971, 0.0001171672047348693, 0.00011609455395955592, 0.00011492414341773838, 0.0001140913664130494, 0.00011346775136189535, 0.00011285000800853595 ], "activation_grad_statistics": { "first_quartile_mean": 0.00021242368711682502, "first_to_last_ratio": 1.8345658968465297, "imbalance_abs_log_ratio": 0.6068078850437921, "last_quartile_mean": 0.0001157896194854402, "mean": 0.0001511171253696375, "normalized": [ 1.7717451736709768, 1.636674793183293, 1.5242755762025908, 1.4186515254985965, 1.3264266232459345, 1.2453738409548603, 1.1893512719369177, 1.1330135273991528, 1.085860457802012, 1.0628542818802846, 1.023252039669589, 0.9942070699734702, 0.9784308644775869, 0.9531083940389613, 0.9232390376562555, 0.9040072643238835, 0.8853775311228679, 0.8680181295617843, 0.8515367513400655, 0.8395795332993111, 0.8267439067367663, 0.8185133286645813, 0.8091730400239778, 0.800791325997512, 0.7909899739933552, 0.7821072339266707, 0.7753403490721149, 0.7682422073314642, 0.7604971517068638, 0.7549863467424895, 0.7508596466770359, 0.7467718018887739 ], "population_cv": 0.2714186609524326 }, "activation_output_rms_by_block": [ 0.0034885562490671873, 0.004831995815038681, 0.005952076520770788, 0.006925336085259914, 0.00385080324485898, 0.005542940925806761, 0.006729288958013058, 0.007927131839096546, 0.004081308841705322, 0.005811207927763462, 0.007534808944910765, 0.008581922389566898, 0.00499920267611742, 0.007102519739419222, 0.008542952127754688, 0.010447698645293713, 0.005041283555328846, 0.007636935915797949, 0.009387043304741383, 0.011020516976714134, 0.005807018838822842, 0.008367395959794521, 0.01024632342159748, 0.012378672137856483, 0.006453148555010557, 0.009127611294388771, 0.011037957854568958, 0.012611051090061665, 0.007010402157902718, 0.009794782847166061, 0.012445803731679916, 0.014087481424212456 ], "activation_output_statistics": { "first_quartile_mean": 0.0056560162047389895, "first_to_last_ratio": 0.5480088979804595, "imbalance_abs_log_ratio": 0.6014637549753238, "last_quartile_mean": 0.010321029869373888, "mean": 0.007962599374877755, "normalized": [ 0.4381177659237371, 0.606836484864952, 0.747504205668035, 0.8697330807712821, 0.4836113263475722, 0.6961220406611074, 0.845112084785299, 0.9955457339856744, 0.512559862622497, 0.7298129234151854, 0.9462750278110585, 1.077778999737589, 0.6278355146046978, 0.8919850673170736, 1.072884836415099, 1.3120964842531853, 0.6331203314377778, 0.9591008609440675, 1.1788918244911069, 1.384035094304029, 0.7292868277593074, 1.0508372411895934, 1.2868063479276561, 1.554601902603762, 0.8104324041933391, 1.1463105029730196, 1.3862254440922965, 1.5837857082010076, 0.8804162846646224, 1.2300986632667847, 1.5630327667805075, 1.7692063559870779 ], "population_cv": 0.3451071044649734 }, "bits_per_byte": 7.959341947457508, "branch_output_rms_by_sublayer": [ 0.0022265110164880753, 0.0027073738165199757, 0.0021613112185150385, 0.002722575329244137, 0.0025601957459002733, 0.0026850115973502398, 0.0025323182344436646, 0.0026861224323511124, 0.0027253651060163975, 0.00271838391199708, 0.002904455177485943, 0.0026645169127732515, 0.0029208383057266474, 0.0026859452482312918, 0.00299624796025455, 0.0027359756641089916, 0.0030937574338167906, 0.0026767742820084095, 0.0033804215490818024, 0.0027451682835817337, 0.0034370317589491606, 0.0026999011170119047, 0.0038969130255281925, 0.002663193503394723, 0.004255065228790045, 0.002626648638397455, 0.004119740333408117, 0.0026531771291047335, 0.003970506135374308, 0.0026517712976783514, 0.004385598469525576, 0.002676903735846281, 0.0041999719105660915, 0.002670086920261383, 0.005232699681073427, 0.0026355870068073273, 0.004664790350943804, 0.0026610493659973145, 0.004898969549685717, 0.0026388936676084995, 0.005131533369421959, 0.0026469926815479994, 0.005069608334451914, 0.00264862016774714, 0.0051318020559847355, 0.002641494618728757, 0.0058372304774820805, 0.002674198243767023, 0.0059690922498703, 0.002557571977376938, 0.00554921617731452, 0.00252733426168561, 0.0058550527319312096, 0.0025468147359788418, 0.006255472078919411, 0.0026856842450797558, 0.006502739619463682, 0.002551089273765683, 0.006122888531535864, 0.0025778315030038357, 0.007071305997669697, 0.0026001501828432083, 0.006241494789719582, 0.0025662274565547705 ], "capture": { "all_gradients_finite": true, "all_gradients_present": true, "count": 32, "dtypes": [ "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32" ], "position": "post-MLP Transformer-block output; Block AttnRes is captured before aggregation-partial reset", "shape": [ 16, 256, 192 ], "storage_unique": true }, "core_parameter_grad_rms_by_block": [ 0.0056993408498569815, 0.005586029321860134, 0.005830344684528519, 0.00540288712216694, 0.00538025811909819, 0.0052226633069546126, 0.005297605878918106, 0.0052023753769565, 0.0048107019260181325, 0.005248672753719287, 0.004962353375385338, 0.005501319898161136, 0.005588445553926757, 0.005229606380765812, 0.004884302670286647, 0.005031817046498285, 0.005000470961685071, 0.0053937604587735305, 0.004980398173225761, 0.004658922297714218, 0.00496598185096497, 0.004578385934052906, 0.004596161734371138, 0.004990442293511754, 0.005201771295872185, 0.00459813521112035, 0.004706502430387731, 0.004781878207244352, 0.0048378722082161795, 0.004473110720597102, 0.005084432926394394, 0.004685705425265166 ], "core_parameter_grad_statistics": { "first_quartile_mean": 0.005452688082542498, "first_to_last_ratio": 1.1368823875795573, "imbalance_abs_log_ratio": 0.12828976841498158, "last_quartile_mean": 0.004796176053137182, "mean": 0.005075395512328069, "normalized": [ 1.122935313319594, 1.1006096585560166, 1.1487468652180286, 1.064525337787646, 1.060066768398564, 1.0290160233362764, 1.0437818818356701, 1.025018713185997, 0.947847692723257, 1.0341406380981206, 0.9777274230809889, 1.0839194472230789, 1.1010857262951226, 1.0303840100861434, 0.9623491722808084, 0.9914137793352396, 0.9852376922229987, 1.0627271206100406, 0.9812827711906272, 0.9179427074003902, 0.9784423379227619, 0.9020747098294245, 0.9055770576316904, 0.9832617539638114, 1.024899691706223, 0.905965889742295, 0.9273173724009683, 0.9421685848185097, 0.9532010257062827, 0.8813324419214177, 1.00178063247374, 0.9232197596982638 ], "population_cv": 0.06921897980134169 }, "depth_weights": [ { "entropy_mean": 0.0, "mean_weights": [ 1.0 ], "sources": 1 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 } ], "layer_input_rms_by_sublayer": [ 0.02802751213312149, 0.014061377383768559, 0.01413465291261673, 0.014165290631353855, 0.014225807972252369, 0.014281988143920898, 0.01433455292135477, 0.014350161887705326, 0.01440478302538395, 0.00966146495193243, 0.009703479707241058, 0.009741888381540775, 0.00977881159633398, 0.009820535778999329, 0.009858808480203152, 0.00991273857653141, 0.009950309991836548, 0.0074927168898284435, 0.00752287870272994, 0.007546746172010899, 0.007571155205368996, 0.00766373984515667, 0.007693085819482803, 0.007762390188872814, 0.007788818795233965, 0.006325984839349985, 0.006353048607707024, 0.006375366821885109, 0.006389216054230928, 0.006445344537496567, 0.0064707002602517605, 0.006555322092026472, 0.00656962301582098, 0.005531130358576775, 0.005553483963012695, 0.005611324217170477, 0.0056249531917274, 0.0056777093559503555, 0.005688921548426151, 0.00574630219489336, 0.005768724251538515, 0.004946626722812653, 0.004966245498508215, 0.0050392793491482735, 0.005057469941675663, 0.005084683652967215, 0.005098348017781973, 0.005203457549214363, 0.005216081161051989, 0.004667261149734259, 0.0046758754178881645, 0.004717131145298481, 0.004727107938379049, 0.004768412560224533, 0.004777040798217058, 0.004826710093766451, 0.004837613552808762, 0.0044052014127373695, 0.0044186655431985855, 0.004504534415900707, 0.004510571248829365, 0.004574235528707504, 0.0045849340967834, 0.004644096828997135 ], "loss_nats": 5.516995429992676, "loss_scale": 1.0, "output_weights": { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, "stream_state_rms_by_sublayer": [ 0.0022265110164880753, 0.0034885562490671873, 0.0040087271481752396, 0.004831995815038681, 0.005379301495850086, 0.005952076520770788, 0.006410556845366955, 0.006925336085259914, 0.0027253651060163975, 0.00385080324485898, 0.004838948138058186, 0.005542940925806761, 0.006175771821290255, 0.006729288958013058, 0.00742313964292407, 0.007927131839096546, 0.0030937574338167906, 0.004081308841705322, 0.005150026176124811, 0.005811207927763462, 0.007019025273621082, 0.007534808944910765, 0.008143599145114422, 0.008581922389566898, 0.004255065228790045, 0.00499920267611742, 0.0066224560141563416, 0.007102519739419222, 0.008129785768687725, 0.008542952127754688, 0.010099578648805618, 0.010447698645293713, 0.0041999719105660915, 0.005041283555328846, 0.00719039561226964, 0.007636935915797949, 0.009030314162373543, 0.009387043304741383, 0.010736005380749702, 0.011020516976714134, 0.005131533369421959, 0.005807018838822842, 0.00787449348717928, 0.008367395959794521, 0.009912566281855106, 0.01024632342159748, 0.012113703414797783, 0.012378672137856483, 0.0059690922498703, 0.006453148555010557, 0.00876442901790142, 0.009127611294388771, 0.010806876234710217, 0.011037957854568958, 0.012353328987956047, 0.012611051090061665, 0.006502739619463682, 0.007010402157902718, 0.009470963850617409, 0.009794782847166061, 0.012123330496251583, 0.012445803731679916, 0.01384514570236206, 0.014087481424212456 ] }, "evaluation": { "bits_per_byte": 7.9557801867581395, "cross_entropy_nats": 5.514526605606079 }, "initial_mixer_hash": "c5a06c218c4501b16fcccee115aa8c79d5dceade78b38bbb547e4f5b2202516d", "initial_public_hash": "73bbe569e1a47981386ebaf59ca891174746bf81897fe5403676ad6d50423c58", "logits_sha256": "6b44486d93c29a13dc0b961117cc2d22f26c81dc519346ac45b1bb0fefa29939", "loss_nats": 5.516995429992676 }, "pre_reduction_uniform_weight_gate": true, "pre_reduction_uniform_weight_max_abs_error": 0.0, "selected_capture_checks": {} }, "uniform_group_6_forward": { "exact_vs_learned_reference": { "diagnostic": true, "evaluation": true, "initial_mixer_hash": true, "initial_public_hash": true, "logits_sha256": true, "loss_nats": true }, "passed": true, "payload": { "diagnostic": { "activation_grad_rms_by_block": [ 0.0002677410375326872, 0.0002473295899108052, 0.00023034414334688336, 0.00021438254043459892, 0.00020044577831868082, 0.00018819731485564262, 0.00017973134526982903, 0.000171217747265473, 0.0001640921109355986, 0.00016061548376455903, 0.00015463090676348656, 0.00015024171443656087, 0.0001478576596127823, 0.00014403100067283958, 0.00013951722939964384, 0.00013661097909789532, 0.00013379570737015456, 0.0001311724045081064, 0.0001286817860091105, 0.00012687484559137374, 0.0001249351626029238, 0.00012369138130452484, 0.00012227990373503417, 0.00012101328320568427, 0.00011953213106608018, 0.0001181897969217971, 0.0001171672047348693, 0.00011609455395955592, 0.00011492414341773838, 0.0001140913664130494, 0.00011346775136189535, 0.00011285000800853595 ], "activation_grad_statistics": { "first_quartile_mean": 0.00021242368711682502, "first_to_last_ratio": 1.8345658968465297, "imbalance_abs_log_ratio": 0.6068078850437921, "last_quartile_mean": 0.0001157896194854402, "mean": 0.0001511171253696375, "normalized": [ 1.7717451736709768, 1.636674793183293, 1.5242755762025908, 1.4186515254985965, 1.3264266232459345, 1.2453738409548603, 1.1893512719369177, 1.1330135273991528, 1.085860457802012, 1.0628542818802846, 1.023252039669589, 0.9942070699734702, 0.9784308644775869, 0.9531083940389613, 0.9232390376562555, 0.9040072643238835, 0.8853775311228679, 0.8680181295617843, 0.8515367513400655, 0.8395795332993111, 0.8267439067367663, 0.8185133286645813, 0.8091730400239778, 0.800791325997512, 0.7909899739933552, 0.7821072339266707, 0.7753403490721149, 0.7682422073314642, 0.7604971517068638, 0.7549863467424895, 0.7508596466770359, 0.7467718018887739 ], "population_cv": 0.2714186609524326 }, "activation_output_rms_by_block": [ 0.0034885562490671873, 0.004831995815038681, 0.005952076520770788, 0.006925336085259914, 0.00385080324485898, 0.005542940925806761, 0.006729288958013058, 0.007927131839096546, 0.004081308841705322, 0.005811207927763462, 0.007534808944910765, 0.008581922389566898, 0.00499920267611742, 0.007102519739419222, 0.008542952127754688, 0.010447698645293713, 0.005041283555328846, 0.007636935915797949, 0.009387043304741383, 0.011020516976714134, 0.005807018838822842, 0.008367395959794521, 0.01024632342159748, 0.012378672137856483, 0.006453148555010557, 0.009127611294388771, 0.011037957854568958, 0.012611051090061665, 0.007010402157902718, 0.009794782847166061, 0.012445803731679916, 0.014087481424212456 ], "activation_output_statistics": { "first_quartile_mean": 0.0056560162047389895, "first_to_last_ratio": 0.5480088979804595, "imbalance_abs_log_ratio": 0.6014637549753238, "last_quartile_mean": 0.010321029869373888, "mean": 0.007962599374877755, "normalized": [ 0.4381177659237371, 0.606836484864952, 0.747504205668035, 0.8697330807712821, 0.4836113263475722, 0.6961220406611074, 0.845112084785299, 0.9955457339856744, 0.512559862622497, 0.7298129234151854, 0.9462750278110585, 1.077778999737589, 0.6278355146046978, 0.8919850673170736, 1.072884836415099, 1.3120964842531853, 0.6331203314377778, 0.9591008609440675, 1.1788918244911069, 1.384035094304029, 0.7292868277593074, 1.0508372411895934, 1.2868063479276561, 1.554601902603762, 0.8104324041933391, 1.1463105029730196, 1.3862254440922965, 1.5837857082010076, 0.8804162846646224, 1.2300986632667847, 1.5630327667805075, 1.7692063559870779 ], "population_cv": 0.3451071044649734 }, "bits_per_byte": 7.959341947457508, "branch_output_rms_by_sublayer": [ 0.0022265110164880753, 0.0027073738165199757, 0.0021613112185150385, 0.002722575329244137, 0.0025601957459002733, 0.0026850115973502398, 0.0025323182344436646, 0.0026861224323511124, 0.0027253651060163975, 0.00271838391199708, 0.002904455177485943, 0.0026645169127732515, 0.0029208383057266474, 0.0026859452482312918, 0.00299624796025455, 0.0027359756641089916, 0.0030937574338167906, 0.0026767742820084095, 0.0033804215490818024, 0.0027451682835817337, 0.0034370317589491606, 0.0026999011170119047, 0.0038969130255281925, 0.002663193503394723, 0.004255065228790045, 0.002626648638397455, 0.004119740333408117, 0.0026531771291047335, 0.003970506135374308, 0.0026517712976783514, 0.004385598469525576, 0.002676903735846281, 0.0041999719105660915, 0.002670086920261383, 0.005232699681073427, 0.0026355870068073273, 0.004664790350943804, 0.0026610493659973145, 0.004898969549685717, 0.0026388936676084995, 0.005131533369421959, 0.0026469926815479994, 0.005069608334451914, 0.00264862016774714, 0.0051318020559847355, 0.002641494618728757, 0.0058372304774820805, 0.002674198243767023, 0.0059690922498703, 0.002557571977376938, 0.00554921617731452, 0.00252733426168561, 0.0058550527319312096, 0.0025468147359788418, 0.006255472078919411, 0.0026856842450797558, 0.006502739619463682, 0.002551089273765683, 0.006122888531535864, 0.0025778315030038357, 0.007071305997669697, 0.0026001501828432083, 0.006241494789719582, 0.0025662274565547705 ], "capture": { "all_gradients_finite": true, "all_gradients_present": true, "count": 32, "dtypes": [ "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32" ], "position": "post-MLP Transformer-block output; Block AttnRes is captured before aggregation-partial reset", "shape": [ 16, 256, 192 ], "storage_unique": true }, "core_parameter_grad_rms_by_block": [ 0.0056993408498569815, 0.005586029321860134, 0.005830344684528519, 0.00540288712216694, 0.00538025811909819, 0.0052226633069546126, 0.005297605878918106, 0.0052023753769565, 0.0048107019260181325, 0.005248672753719287, 0.004962353375385338, 0.005501319898161136, 0.005588445553926757, 0.005229606380765812, 0.004884302670286647, 0.005031817046498285, 0.005000470961685071, 0.0053937604587735305, 0.004980398173225761, 0.004658922297714218, 0.00496598185096497, 0.004578385934052906, 0.004596161734371138, 0.004990442293511754, 0.005201771295872185, 0.00459813521112035, 0.004706502430387731, 0.004781878207244352, 0.0048378722082161795, 0.004473110720597102, 0.005084432926394394, 0.004685705425265166 ], "core_parameter_grad_statistics": { "first_quartile_mean": 0.005452688082542498, "first_to_last_ratio": 1.1368823875795573, "imbalance_abs_log_ratio": 0.12828976841498158, "last_quartile_mean": 0.004796176053137182, "mean": 0.005075395512328069, "normalized": [ 1.122935313319594, 1.1006096585560166, 1.1487468652180286, 1.064525337787646, 1.060066768398564, 1.0290160233362764, 1.0437818818356701, 1.025018713185997, 0.947847692723257, 1.0341406380981206, 0.9777274230809889, 1.0839194472230789, 1.1010857262951226, 1.0303840100861434, 0.9623491722808084, 0.9914137793352396, 0.9852376922229987, 1.0627271206100406, 0.9812827711906272, 0.9179427074003902, 0.9784423379227619, 0.9020747098294245, 0.9055770576316904, 0.9832617539638114, 1.024899691706223, 0.905965889742295, 0.9273173724009683, 0.9421685848185097, 0.9532010257062827, 0.8813324419214177, 1.00178063247374, 0.9232197596982638 ], "population_cv": 0.06921897980134169 }, "depth_weights": [ { "entropy_mean": 0.0, "mean_weights": [ 1.0 ], "sources": 1 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 } ], "layer_input_rms_by_sublayer": [ 0.02802751213312149, 0.014061377383768559, 0.01413465291261673, 0.014165290631353855, 0.014225807972252369, 0.014281988143920898, 0.01433455292135477, 0.014350161887705326, 0.01440478302538395, 0.00966146495193243, 0.009703479707241058, 0.009741888381540775, 0.00977881159633398, 0.009820535778999329, 0.009858808480203152, 0.00991273857653141, 0.009950309991836548, 0.0074927168898284435, 0.00752287870272994, 0.007546746172010899, 0.007571155205368996, 0.00766373984515667, 0.007693085819482803, 0.007762390188872814, 0.007788818795233965, 0.006325984839349985, 0.006353048607707024, 0.006375366821885109, 0.006389216054230928, 0.006445344537496567, 0.0064707002602517605, 0.006555322092026472, 0.00656962301582098, 0.005531130358576775, 0.005553483963012695, 0.005611324217170477, 0.0056249531917274, 0.0056777093559503555, 0.005688921548426151, 0.00574630219489336, 0.005768724251538515, 0.004946626722812653, 0.004966245498508215, 0.0050392793491482735, 0.005057469941675663, 0.005084683652967215, 0.005098348017781973, 0.005203457549214363, 0.005216081161051989, 0.004667261149734259, 0.0046758754178881645, 0.004717131145298481, 0.004727107938379049, 0.004768412560224533, 0.004777040798217058, 0.004826710093766451, 0.004837613552808762, 0.0044052014127373695, 0.0044186655431985855, 0.004504534415900707, 0.004510571248829365, 0.004574235528707504, 0.0045849340967834, 0.004644096828997135 ], "loss_nats": 5.516995429992676, "loss_scale": 1.0, "output_weights": { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, "stream_state_rms_by_sublayer": [ 0.0022265110164880753, 0.0034885562490671873, 0.0040087271481752396, 0.004831995815038681, 0.005379301495850086, 0.005952076520770788, 0.006410556845366955, 0.006925336085259914, 0.0027253651060163975, 0.00385080324485898, 0.004838948138058186, 0.005542940925806761, 0.006175771821290255, 0.006729288958013058, 0.00742313964292407, 0.007927131839096546, 0.0030937574338167906, 0.004081308841705322, 0.005150026176124811, 0.005811207927763462, 0.007019025273621082, 0.007534808944910765, 0.008143599145114422, 0.008581922389566898, 0.004255065228790045, 0.00499920267611742, 0.0066224560141563416, 0.007102519739419222, 0.008129785768687725, 0.008542952127754688, 0.010099578648805618, 0.010447698645293713, 0.0041999719105660915, 0.005041283555328846, 0.00719039561226964, 0.007636935915797949, 0.009030314162373543, 0.009387043304741383, 0.010736005380749702, 0.011020516976714134, 0.005131533369421959, 0.005807018838822842, 0.00787449348717928, 0.008367395959794521, 0.009912566281855106, 0.01024632342159748, 0.012113703414797783, 0.012378672137856483, 0.0059690922498703, 0.006453148555010557, 0.00876442901790142, 0.009127611294388771, 0.010806876234710217, 0.011037957854568958, 0.012353328987956047, 0.012611051090061665, 0.006502739619463682, 0.007010402157902718, 0.009470963850617409, 0.009794782847166061, 0.012123330496251583, 0.012445803731679916, 0.01384514570236206, 0.014087481424212456 ] }, "evaluation": { "bits_per_byte": 7.9557801867581395, "cross_entropy_nats": 5.514526605606079 }, "initial_mixer_hash": "c5a06c218c4501b16fcccee115aa8c79d5dceade78b38bbb547e4f5b2202516d", "initial_public_hash": "73bbe569e1a47981386ebaf59ca891174746bf81897fe5403676ad6d50423c58", "logits_sha256": "6b44486d93c29a13dc0b961117cc2d22f26c81dc519346ac45b1bb0fefa29939", "loss_nats": 5.516995429992676 }, "pre_reduction_uniform_weight_gate": true, "pre_reduction_uniform_weight_max_abs_error": 0.0, "selected_capture_checks": { "40": { "capture_summary_exact_vs_learned": true, "expected_source_count": 6, "passed": true, "source_count": 6 }, "41": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "42": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "43": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "44": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "45": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "46": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "47": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 } } }, "uniform_group_7_forward": { "exact_vs_learned_reference": { "diagnostic": true, "evaluation": true, "initial_mixer_hash": true, "initial_public_hash": true, "logits_sha256": true, "loss_nats": true }, "passed": true, "payload": { "diagnostic": { "activation_grad_rms_by_block": [ 0.0002677410375326872, 0.0002473295899108052, 0.00023034414334688336, 0.00021438254043459892, 0.00020044577831868082, 0.00018819731485564262, 0.00017973134526982903, 0.000171217747265473, 0.0001640921109355986, 0.00016061548376455903, 0.00015463090676348656, 0.00015024171443656087, 0.0001478576596127823, 0.00014403100067283958, 0.00013951722939964384, 0.00013661097909789532, 0.00013379570737015456, 0.0001311724045081064, 0.0001286817860091105, 0.00012687484559137374, 0.0001249351626029238, 0.00012369138130452484, 0.00012227990373503417, 0.00012101328320568427, 0.00011953213106608018, 0.0001181897969217971, 0.0001171672047348693, 0.00011609455395955592, 0.00011492414341773838, 0.0001140913664130494, 0.00011346775136189535, 0.00011285000800853595 ], "activation_grad_statistics": { "first_quartile_mean": 0.00021242368711682502, "first_to_last_ratio": 1.8345658968465297, "imbalance_abs_log_ratio": 0.6068078850437921, "last_quartile_mean": 0.0001157896194854402, "mean": 0.0001511171253696375, "normalized": [ 1.7717451736709768, 1.636674793183293, 1.5242755762025908, 1.4186515254985965, 1.3264266232459345, 1.2453738409548603, 1.1893512719369177, 1.1330135273991528, 1.085860457802012, 1.0628542818802846, 1.023252039669589, 0.9942070699734702, 0.9784308644775869, 0.9531083940389613, 0.9232390376562555, 0.9040072643238835, 0.8853775311228679, 0.8680181295617843, 0.8515367513400655, 0.8395795332993111, 0.8267439067367663, 0.8185133286645813, 0.8091730400239778, 0.800791325997512, 0.7909899739933552, 0.7821072339266707, 0.7753403490721149, 0.7682422073314642, 0.7604971517068638, 0.7549863467424895, 0.7508596466770359, 0.7467718018887739 ], "population_cv": 0.2714186609524326 }, "activation_output_rms_by_block": [ 0.0034885562490671873, 0.004831995815038681, 0.005952076520770788, 0.006925336085259914, 0.00385080324485898, 0.005542940925806761, 0.006729288958013058, 0.007927131839096546, 0.004081308841705322, 0.005811207927763462, 0.007534808944910765, 0.008581922389566898, 0.00499920267611742, 0.007102519739419222, 0.008542952127754688, 0.010447698645293713, 0.005041283555328846, 0.007636935915797949, 0.009387043304741383, 0.011020516976714134, 0.005807018838822842, 0.008367395959794521, 0.01024632342159748, 0.012378672137856483, 0.006453148555010557, 0.009127611294388771, 0.011037957854568958, 0.012611051090061665, 0.007010402157902718, 0.009794782847166061, 0.012445803731679916, 0.014087481424212456 ], "activation_output_statistics": { "first_quartile_mean": 0.0056560162047389895, "first_to_last_ratio": 0.5480088979804595, "imbalance_abs_log_ratio": 0.6014637549753238, "last_quartile_mean": 0.010321029869373888, "mean": 0.007962599374877755, "normalized": [ 0.4381177659237371, 0.606836484864952, 0.747504205668035, 0.8697330807712821, 0.4836113263475722, 0.6961220406611074, 0.845112084785299, 0.9955457339856744, 0.512559862622497, 0.7298129234151854, 0.9462750278110585, 1.077778999737589, 0.6278355146046978, 0.8919850673170736, 1.072884836415099, 1.3120964842531853, 0.6331203314377778, 0.9591008609440675, 1.1788918244911069, 1.384035094304029, 0.7292868277593074, 1.0508372411895934, 1.2868063479276561, 1.554601902603762, 0.8104324041933391, 1.1463105029730196, 1.3862254440922965, 1.5837857082010076, 0.8804162846646224, 1.2300986632667847, 1.5630327667805075, 1.7692063559870779 ], "population_cv": 0.3451071044649734 }, "bits_per_byte": 7.959341947457508, "branch_output_rms_by_sublayer": [ 0.0022265110164880753, 0.0027073738165199757, 0.0021613112185150385, 0.002722575329244137, 0.0025601957459002733, 0.0026850115973502398, 0.0025323182344436646, 0.0026861224323511124, 0.0027253651060163975, 0.00271838391199708, 0.002904455177485943, 0.0026645169127732515, 0.0029208383057266474, 0.0026859452482312918, 0.00299624796025455, 0.0027359756641089916, 0.0030937574338167906, 0.0026767742820084095, 0.0033804215490818024, 0.0027451682835817337, 0.0034370317589491606, 0.0026999011170119047, 0.0038969130255281925, 0.002663193503394723, 0.004255065228790045, 0.002626648638397455, 0.004119740333408117, 0.0026531771291047335, 0.003970506135374308, 0.0026517712976783514, 0.004385598469525576, 0.002676903735846281, 0.0041999719105660915, 0.002670086920261383, 0.005232699681073427, 0.0026355870068073273, 0.004664790350943804, 0.0026610493659973145, 0.004898969549685717, 0.0026388936676084995, 0.005131533369421959, 0.0026469926815479994, 0.005069608334451914, 0.00264862016774714, 0.0051318020559847355, 0.002641494618728757, 0.0058372304774820805, 0.002674198243767023, 0.0059690922498703, 0.002557571977376938, 0.00554921617731452, 0.00252733426168561, 0.0058550527319312096, 0.0025468147359788418, 0.006255472078919411, 0.0026856842450797558, 0.006502739619463682, 0.002551089273765683, 0.006122888531535864, 0.0025778315030038357, 0.007071305997669697, 0.0026001501828432083, 0.006241494789719582, 0.0025662274565547705 ], "capture": { "all_gradients_finite": true, "all_gradients_present": true, "count": 32, "dtypes": [ "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32" ], "position": "post-MLP Transformer-block output; Block AttnRes is captured before aggregation-partial reset", "shape": [ 16, 256, 192 ], "storage_unique": true }, "core_parameter_grad_rms_by_block": [ 0.0056993408498569815, 0.005586029321860134, 0.005830344684528519, 0.00540288712216694, 0.00538025811909819, 0.0052226633069546126, 0.005297605878918106, 0.0052023753769565, 0.0048107019260181325, 0.005248672753719287, 0.004962353375385338, 0.005501319898161136, 0.005588445553926757, 0.005229606380765812, 0.004884302670286647, 0.005031817046498285, 0.005000470961685071, 0.0053937604587735305, 0.004980398173225761, 0.004658922297714218, 0.00496598185096497, 0.004578385934052906, 0.004596161734371138, 0.004990442293511754, 0.005201771295872185, 0.00459813521112035, 0.004706502430387731, 0.004781878207244352, 0.0048378722082161795, 0.004473110720597102, 0.005084432926394394, 0.004685705425265166 ], "core_parameter_grad_statistics": { "first_quartile_mean": 0.005452688082542498, "first_to_last_ratio": 1.1368823875795573, "imbalance_abs_log_ratio": 0.12828976841498158, "last_quartile_mean": 0.004796176053137182, "mean": 0.005075395512328069, "normalized": [ 1.122935313319594, 1.1006096585560166, 1.1487468652180286, 1.064525337787646, 1.060066768398564, 1.0290160233362764, 1.0437818818356701, 1.025018713185997, 0.947847692723257, 1.0341406380981206, 0.9777274230809889, 1.0839194472230789, 1.1010857262951226, 1.0303840100861434, 0.9623491722808084, 0.9914137793352396, 0.9852376922229987, 1.0627271206100406, 0.9812827711906272, 0.9179427074003902, 0.9784423379227619, 0.9020747098294245, 0.9055770576316904, 0.9832617539638114, 1.024899691706223, 0.905965889742295, 0.9273173724009683, 0.9421685848185097, 0.9532010257062827, 0.8813324419214177, 1.00178063247374, 0.9232197596982638 ], "population_cv": 0.06921897980134169 }, "depth_weights": [ { "entropy_mean": 0.0, "mean_weights": [ 1.0 ], "sources": 1 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 } ], "layer_input_rms_by_sublayer": [ 0.02802751213312149, 0.014061377383768559, 0.01413465291261673, 0.014165290631353855, 0.014225807972252369, 0.014281988143920898, 0.01433455292135477, 0.014350161887705326, 0.01440478302538395, 0.00966146495193243, 0.009703479707241058, 0.009741888381540775, 0.00977881159633398, 0.009820535778999329, 0.009858808480203152, 0.00991273857653141, 0.009950309991836548, 0.0074927168898284435, 0.00752287870272994, 0.007546746172010899, 0.007571155205368996, 0.00766373984515667, 0.007693085819482803, 0.007762390188872814, 0.007788818795233965, 0.006325984839349985, 0.006353048607707024, 0.006375366821885109, 0.006389216054230928, 0.006445344537496567, 0.0064707002602517605, 0.006555322092026472, 0.00656962301582098, 0.005531130358576775, 0.005553483963012695, 0.005611324217170477, 0.0056249531917274, 0.0056777093559503555, 0.005688921548426151, 0.00574630219489336, 0.005768724251538515, 0.004946626722812653, 0.004966245498508215, 0.0050392793491482735, 0.005057469941675663, 0.005084683652967215, 0.005098348017781973, 0.005203457549214363, 0.005216081161051989, 0.004667261149734259, 0.0046758754178881645, 0.004717131145298481, 0.004727107938379049, 0.004768412560224533, 0.004777040798217058, 0.004826710093766451, 0.004837613552808762, 0.0044052014127373695, 0.0044186655431985855, 0.004504534415900707, 0.004510571248829365, 0.004574235528707504, 0.0045849340967834, 0.004644096828997135 ], "loss_nats": 5.516995429992676, "loss_scale": 1.0, "output_weights": { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, "stream_state_rms_by_sublayer": [ 0.0022265110164880753, 0.0034885562490671873, 0.0040087271481752396, 0.004831995815038681, 0.005379301495850086, 0.005952076520770788, 0.006410556845366955, 0.006925336085259914, 0.0027253651060163975, 0.00385080324485898, 0.004838948138058186, 0.005542940925806761, 0.006175771821290255, 0.006729288958013058, 0.00742313964292407, 0.007927131839096546, 0.0030937574338167906, 0.004081308841705322, 0.005150026176124811, 0.005811207927763462, 0.007019025273621082, 0.007534808944910765, 0.008143599145114422, 0.008581922389566898, 0.004255065228790045, 0.00499920267611742, 0.0066224560141563416, 0.007102519739419222, 0.008129785768687725, 0.008542952127754688, 0.010099578648805618, 0.010447698645293713, 0.0041999719105660915, 0.005041283555328846, 0.00719039561226964, 0.007636935915797949, 0.009030314162373543, 0.009387043304741383, 0.010736005380749702, 0.011020516976714134, 0.005131533369421959, 0.005807018838822842, 0.00787449348717928, 0.008367395959794521, 0.009912566281855106, 0.01024632342159748, 0.012113703414797783, 0.012378672137856483, 0.0059690922498703, 0.006453148555010557, 0.00876442901790142, 0.009127611294388771, 0.010806876234710217, 0.011037957854568958, 0.012353328987956047, 0.012611051090061665, 0.006502739619463682, 0.007010402157902718, 0.009470963850617409, 0.009794782847166061, 0.012123330496251583, 0.012445803731679916, 0.01384514570236206, 0.014087481424212456 ] }, "evaluation": { "bits_per_byte": 7.9557801867581395, "cross_entropy_nats": 5.514526605606079 }, "initial_mixer_hash": "c5a06c218c4501b16fcccee115aa8c79d5dceade78b38bbb547e4f5b2202516d", "initial_public_hash": "73bbe569e1a47981386ebaf59ca891174746bf81897fe5403676ad6d50423c58", "logits_sha256": "6b44486d93c29a13dc0b961117cc2d22f26c81dc519346ac45b1bb0fefa29939", "loss_nats": 5.516995429992676 }, "pre_reduction_uniform_weight_gate": true, "pre_reduction_uniform_weight_max_abs_error": 0.0, "selected_capture_checks": { "48": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "49": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "50": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "51": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "52": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "53": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "54": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "55": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 } } }, "uniform_group_7_mlp_forward": { "exact_vs_learned_reference": { "diagnostic": true, "evaluation": true, "initial_mixer_hash": true, "initial_public_hash": true, "logits_sha256": true, "loss_nats": true }, "passed": true, "payload": { "diagnostic": { "activation_grad_rms_by_block": [ 0.0002677410375326872, 0.0002473295899108052, 0.00023034414334688336, 0.00021438254043459892, 0.00020044577831868082, 0.00018819731485564262, 0.00017973134526982903, 0.000171217747265473, 0.0001640921109355986, 0.00016061548376455903, 0.00015463090676348656, 0.00015024171443656087, 0.0001478576596127823, 0.00014403100067283958, 0.00013951722939964384, 0.00013661097909789532, 0.00013379570737015456, 0.0001311724045081064, 0.0001286817860091105, 0.00012687484559137374, 0.0001249351626029238, 0.00012369138130452484, 0.00012227990373503417, 0.00012101328320568427, 0.00011953213106608018, 0.0001181897969217971, 0.0001171672047348693, 0.00011609455395955592, 0.00011492414341773838, 0.0001140913664130494, 0.00011346775136189535, 0.00011285000800853595 ], "activation_grad_statistics": { "first_quartile_mean": 0.00021242368711682502, "first_to_last_ratio": 1.8345658968465297, "imbalance_abs_log_ratio": 0.6068078850437921, "last_quartile_mean": 0.0001157896194854402, "mean": 0.0001511171253696375, "normalized": [ 1.7717451736709768, 1.636674793183293, 1.5242755762025908, 1.4186515254985965, 1.3264266232459345, 1.2453738409548603, 1.1893512719369177, 1.1330135273991528, 1.085860457802012, 1.0628542818802846, 1.023252039669589, 0.9942070699734702, 0.9784308644775869, 0.9531083940389613, 0.9232390376562555, 0.9040072643238835, 0.8853775311228679, 0.8680181295617843, 0.8515367513400655, 0.8395795332993111, 0.8267439067367663, 0.8185133286645813, 0.8091730400239778, 0.800791325997512, 0.7909899739933552, 0.7821072339266707, 0.7753403490721149, 0.7682422073314642, 0.7604971517068638, 0.7549863467424895, 0.7508596466770359, 0.7467718018887739 ], "population_cv": 0.2714186609524326 }, "activation_output_rms_by_block": [ 0.0034885562490671873, 0.004831995815038681, 0.005952076520770788, 0.006925336085259914, 0.00385080324485898, 0.005542940925806761, 0.006729288958013058, 0.007927131839096546, 0.004081308841705322, 0.005811207927763462, 0.007534808944910765, 0.008581922389566898, 0.00499920267611742, 0.007102519739419222, 0.008542952127754688, 0.010447698645293713, 0.005041283555328846, 0.007636935915797949, 0.009387043304741383, 0.011020516976714134, 0.005807018838822842, 0.008367395959794521, 0.01024632342159748, 0.012378672137856483, 0.006453148555010557, 0.009127611294388771, 0.011037957854568958, 0.012611051090061665, 0.007010402157902718, 0.009794782847166061, 0.012445803731679916, 0.014087481424212456 ], "activation_output_statistics": { "first_quartile_mean": 0.0056560162047389895, "first_to_last_ratio": 0.5480088979804595, "imbalance_abs_log_ratio": 0.6014637549753238, "last_quartile_mean": 0.010321029869373888, "mean": 0.007962599374877755, "normalized": [ 0.4381177659237371, 0.606836484864952, 0.747504205668035, 0.8697330807712821, 0.4836113263475722, 0.6961220406611074, 0.845112084785299, 0.9955457339856744, 0.512559862622497, 0.7298129234151854, 0.9462750278110585, 1.077778999737589, 0.6278355146046978, 0.8919850673170736, 1.072884836415099, 1.3120964842531853, 0.6331203314377778, 0.9591008609440675, 1.1788918244911069, 1.384035094304029, 0.7292868277593074, 1.0508372411895934, 1.2868063479276561, 1.554601902603762, 0.8104324041933391, 1.1463105029730196, 1.3862254440922965, 1.5837857082010076, 0.8804162846646224, 1.2300986632667847, 1.5630327667805075, 1.7692063559870779 ], "population_cv": 0.3451071044649734 }, "bits_per_byte": 7.959341947457508, "branch_output_rms_by_sublayer": [ 0.0022265110164880753, 0.0027073738165199757, 0.0021613112185150385, 0.002722575329244137, 0.0025601957459002733, 0.0026850115973502398, 0.0025323182344436646, 0.0026861224323511124, 0.0027253651060163975, 0.00271838391199708, 0.002904455177485943, 0.0026645169127732515, 0.0029208383057266474, 0.0026859452482312918, 0.00299624796025455, 0.0027359756641089916, 0.0030937574338167906, 0.0026767742820084095, 0.0033804215490818024, 0.0027451682835817337, 0.0034370317589491606, 0.0026999011170119047, 0.0038969130255281925, 0.002663193503394723, 0.004255065228790045, 0.002626648638397455, 0.004119740333408117, 0.0026531771291047335, 0.003970506135374308, 0.0026517712976783514, 0.004385598469525576, 0.002676903735846281, 0.0041999719105660915, 0.002670086920261383, 0.005232699681073427, 0.0026355870068073273, 0.004664790350943804, 0.0026610493659973145, 0.004898969549685717, 0.0026388936676084995, 0.005131533369421959, 0.0026469926815479994, 0.005069608334451914, 0.00264862016774714, 0.0051318020559847355, 0.002641494618728757, 0.0058372304774820805, 0.002674198243767023, 0.0059690922498703, 0.002557571977376938, 0.00554921617731452, 0.00252733426168561, 0.0058550527319312096, 0.0025468147359788418, 0.006255472078919411, 0.0026856842450797558, 0.006502739619463682, 0.002551089273765683, 0.006122888531535864, 0.0025778315030038357, 0.007071305997669697, 0.0026001501828432083, 0.006241494789719582, 0.0025662274565547705 ], "capture": { "all_gradients_finite": true, "all_gradients_present": true, "count": 32, "dtypes": [ "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32" ], "position": "post-MLP Transformer-block output; Block AttnRes is captured before aggregation-partial reset", "shape": [ 16, 256, 192 ], "storage_unique": true }, "core_parameter_grad_rms_by_block": [ 0.0056993408498569815, 0.005586029321860134, 0.005830344684528519, 0.00540288712216694, 0.00538025811909819, 0.0052226633069546126, 0.005297605878918106, 0.0052023753769565, 0.0048107019260181325, 0.005248672753719287, 0.004962353375385338, 0.005501319898161136, 0.005588445553926757, 0.005229606380765812, 0.004884302670286647, 0.005031817046498285, 0.005000470961685071, 0.0053937604587735305, 0.004980398173225761, 0.004658922297714218, 0.00496598185096497, 0.004578385934052906, 0.004596161734371138, 0.004990442293511754, 0.005201771295872185, 0.00459813521112035, 0.004706502430387731, 0.004781878207244352, 0.0048378722082161795, 0.004473110720597102, 0.005084432926394394, 0.004685705425265166 ], "core_parameter_grad_statistics": { "first_quartile_mean": 0.005452688082542498, "first_to_last_ratio": 1.1368823875795573, "imbalance_abs_log_ratio": 0.12828976841498158, "last_quartile_mean": 0.004796176053137182, "mean": 0.005075395512328069, "normalized": [ 1.122935313319594, 1.1006096585560166, 1.1487468652180286, 1.064525337787646, 1.060066768398564, 1.0290160233362764, 1.0437818818356701, 1.025018713185997, 0.947847692723257, 1.0341406380981206, 0.9777274230809889, 1.0839194472230789, 1.1010857262951226, 1.0303840100861434, 0.9623491722808084, 0.9914137793352396, 0.9852376922229987, 1.0627271206100406, 0.9812827711906272, 0.9179427074003902, 0.9784423379227619, 0.9020747098294245, 0.9055770576316904, 0.9832617539638114, 1.024899691706223, 0.905965889742295, 0.9273173724009683, 0.9421685848185097, 0.9532010257062827, 0.8813324419214177, 1.00178063247374, 0.9232197596982638 ], "population_cv": 0.06921897980134169 }, "depth_weights": [ { "entropy_mean": 0.0, "mean_weights": [ 1.0 ], "sources": 1 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 } ], "layer_input_rms_by_sublayer": [ 0.02802751213312149, 0.014061377383768559, 0.01413465291261673, 0.014165290631353855, 0.014225807972252369, 0.014281988143920898, 0.01433455292135477, 0.014350161887705326, 0.01440478302538395, 0.00966146495193243, 0.009703479707241058, 0.009741888381540775, 0.00977881159633398, 0.009820535778999329, 0.009858808480203152, 0.00991273857653141, 0.009950309991836548, 0.0074927168898284435, 0.00752287870272994, 0.007546746172010899, 0.007571155205368996, 0.00766373984515667, 0.007693085819482803, 0.007762390188872814, 0.007788818795233965, 0.006325984839349985, 0.006353048607707024, 0.006375366821885109, 0.006389216054230928, 0.006445344537496567, 0.0064707002602517605, 0.006555322092026472, 0.00656962301582098, 0.005531130358576775, 0.005553483963012695, 0.005611324217170477, 0.0056249531917274, 0.0056777093559503555, 0.005688921548426151, 0.00574630219489336, 0.005768724251538515, 0.004946626722812653, 0.004966245498508215, 0.0050392793491482735, 0.005057469941675663, 0.005084683652967215, 0.005098348017781973, 0.005203457549214363, 0.005216081161051989, 0.004667261149734259, 0.0046758754178881645, 0.004717131145298481, 0.004727107938379049, 0.004768412560224533, 0.004777040798217058, 0.004826710093766451, 0.004837613552808762, 0.0044052014127373695, 0.0044186655431985855, 0.004504534415900707, 0.004510571248829365, 0.004574235528707504, 0.0045849340967834, 0.004644096828997135 ], "loss_nats": 5.516995429992676, "loss_scale": 1.0, "output_weights": { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, "stream_state_rms_by_sublayer": [ 0.0022265110164880753, 0.0034885562490671873, 0.0040087271481752396, 0.004831995815038681, 0.005379301495850086, 0.005952076520770788, 0.006410556845366955, 0.006925336085259914, 0.0027253651060163975, 0.00385080324485898, 0.004838948138058186, 0.005542940925806761, 0.006175771821290255, 0.006729288958013058, 0.00742313964292407, 0.007927131839096546, 0.0030937574338167906, 0.004081308841705322, 0.005150026176124811, 0.005811207927763462, 0.007019025273621082, 0.007534808944910765, 0.008143599145114422, 0.008581922389566898, 0.004255065228790045, 0.00499920267611742, 0.0066224560141563416, 0.007102519739419222, 0.008129785768687725, 0.008542952127754688, 0.010099578648805618, 0.010447698645293713, 0.0041999719105660915, 0.005041283555328846, 0.00719039561226964, 0.007636935915797949, 0.009030314162373543, 0.009387043304741383, 0.010736005380749702, 0.011020516976714134, 0.005131533369421959, 0.005807018838822842, 0.00787449348717928, 0.008367395959794521, 0.009912566281855106, 0.01024632342159748, 0.012113703414797783, 0.012378672137856483, 0.0059690922498703, 0.006453148555010557, 0.00876442901790142, 0.009127611294388771, 0.010806876234710217, 0.011037957854568958, 0.012353328987956047, 0.012611051090061665, 0.006502739619463682, 0.007010402157902718, 0.009470963850617409, 0.009794782847166061, 0.012123330496251583, 0.012445803731679916, 0.01384514570236206, 0.014087481424212456 ] }, "evaluation": { "bits_per_byte": 7.9557801867581395, "cross_entropy_nats": 5.514526605606079 }, "initial_mixer_hash": "c5a06c218c4501b16fcccee115aa8c79d5dceade78b38bbb547e4f5b2202516d", "initial_public_hash": "73bbe569e1a47981386ebaf59ca891174746bf81897fe5403676ad6d50423c58", "logits_sha256": "6b44486d93c29a13dc0b961117cc2d22f26c81dc519346ac45b1bb0fefa29939", "loss_nats": 5.516995429992676 }, "pre_reduction_uniform_weight_gate": true, "pre_reduction_uniform_weight_max_abs_error": 0.0, "selected_capture_checks": { "49": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "51": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "53": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "55": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 } } }, "uniform_groups_6_7_forward": { "exact_vs_learned_reference": { "diagnostic": true, "evaluation": true, "initial_mixer_hash": true, "initial_public_hash": true, "logits_sha256": true, "loss_nats": true }, "passed": true, "payload": { "diagnostic": { "activation_grad_rms_by_block": [ 0.0002677410375326872, 0.0002473295899108052, 0.00023034414334688336, 0.00021438254043459892, 0.00020044577831868082, 0.00018819731485564262, 0.00017973134526982903, 0.000171217747265473, 0.0001640921109355986, 0.00016061548376455903, 0.00015463090676348656, 0.00015024171443656087, 0.0001478576596127823, 0.00014403100067283958, 0.00013951722939964384, 0.00013661097909789532, 0.00013379570737015456, 0.0001311724045081064, 0.0001286817860091105, 0.00012687484559137374, 0.0001249351626029238, 0.00012369138130452484, 0.00012227990373503417, 0.00012101328320568427, 0.00011953213106608018, 0.0001181897969217971, 0.0001171672047348693, 0.00011609455395955592, 0.00011492414341773838, 0.0001140913664130494, 0.00011346775136189535, 0.00011285000800853595 ], "activation_grad_statistics": { "first_quartile_mean": 0.00021242368711682502, "first_to_last_ratio": 1.8345658968465297, "imbalance_abs_log_ratio": 0.6068078850437921, "last_quartile_mean": 0.0001157896194854402, "mean": 0.0001511171253696375, "normalized": [ 1.7717451736709768, 1.636674793183293, 1.5242755762025908, 1.4186515254985965, 1.3264266232459345, 1.2453738409548603, 1.1893512719369177, 1.1330135273991528, 1.085860457802012, 1.0628542818802846, 1.023252039669589, 0.9942070699734702, 0.9784308644775869, 0.9531083940389613, 0.9232390376562555, 0.9040072643238835, 0.8853775311228679, 0.8680181295617843, 0.8515367513400655, 0.8395795332993111, 0.8267439067367663, 0.8185133286645813, 0.8091730400239778, 0.800791325997512, 0.7909899739933552, 0.7821072339266707, 0.7753403490721149, 0.7682422073314642, 0.7604971517068638, 0.7549863467424895, 0.7508596466770359, 0.7467718018887739 ], "population_cv": 0.2714186609524326 }, "activation_output_rms_by_block": [ 0.0034885562490671873, 0.004831995815038681, 0.005952076520770788, 0.006925336085259914, 0.00385080324485898, 0.005542940925806761, 0.006729288958013058, 0.007927131839096546, 0.004081308841705322, 0.005811207927763462, 0.007534808944910765, 0.008581922389566898, 0.00499920267611742, 0.007102519739419222, 0.008542952127754688, 0.010447698645293713, 0.005041283555328846, 0.007636935915797949, 0.009387043304741383, 0.011020516976714134, 0.005807018838822842, 0.008367395959794521, 0.01024632342159748, 0.012378672137856483, 0.006453148555010557, 0.009127611294388771, 0.011037957854568958, 0.012611051090061665, 0.007010402157902718, 0.009794782847166061, 0.012445803731679916, 0.014087481424212456 ], "activation_output_statistics": { "first_quartile_mean": 0.0056560162047389895, "first_to_last_ratio": 0.5480088979804595, "imbalance_abs_log_ratio": 0.6014637549753238, "last_quartile_mean": 0.010321029869373888, "mean": 0.007962599374877755, "normalized": [ 0.4381177659237371, 0.606836484864952, 0.747504205668035, 0.8697330807712821, 0.4836113263475722, 0.6961220406611074, 0.845112084785299, 0.9955457339856744, 0.512559862622497, 0.7298129234151854, 0.9462750278110585, 1.077778999737589, 0.6278355146046978, 0.8919850673170736, 1.072884836415099, 1.3120964842531853, 0.6331203314377778, 0.9591008609440675, 1.1788918244911069, 1.384035094304029, 0.7292868277593074, 1.0508372411895934, 1.2868063479276561, 1.554601902603762, 0.8104324041933391, 1.1463105029730196, 1.3862254440922965, 1.5837857082010076, 0.8804162846646224, 1.2300986632667847, 1.5630327667805075, 1.7692063559870779 ], "population_cv": 0.3451071044649734 }, "bits_per_byte": 7.959341947457508, "branch_output_rms_by_sublayer": [ 0.0022265110164880753, 0.0027073738165199757, 0.0021613112185150385, 0.002722575329244137, 0.0025601957459002733, 0.0026850115973502398, 0.0025323182344436646, 0.0026861224323511124, 0.0027253651060163975, 0.00271838391199708, 0.002904455177485943, 0.0026645169127732515, 0.0029208383057266474, 0.0026859452482312918, 0.00299624796025455, 0.0027359756641089916, 0.0030937574338167906, 0.0026767742820084095, 0.0033804215490818024, 0.0027451682835817337, 0.0034370317589491606, 0.0026999011170119047, 0.0038969130255281925, 0.002663193503394723, 0.004255065228790045, 0.002626648638397455, 0.004119740333408117, 0.0026531771291047335, 0.003970506135374308, 0.0026517712976783514, 0.004385598469525576, 0.002676903735846281, 0.0041999719105660915, 0.002670086920261383, 0.005232699681073427, 0.0026355870068073273, 0.004664790350943804, 0.0026610493659973145, 0.004898969549685717, 0.0026388936676084995, 0.005131533369421959, 0.0026469926815479994, 0.005069608334451914, 0.00264862016774714, 0.0051318020559847355, 0.002641494618728757, 0.0058372304774820805, 0.002674198243767023, 0.0059690922498703, 0.002557571977376938, 0.00554921617731452, 0.00252733426168561, 0.0058550527319312096, 0.0025468147359788418, 0.006255472078919411, 0.0026856842450797558, 0.006502739619463682, 0.002551089273765683, 0.006122888531535864, 0.0025778315030038357, 0.007071305997669697, 0.0026001501828432083, 0.006241494789719582, 0.0025662274565547705 ], "capture": { "all_gradients_finite": true, "all_gradients_present": true, "count": 32, "dtypes": [ "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32", "torch.float32" ], "position": "post-MLP Transformer-block output; Block AttnRes is captured before aggregation-partial reset", "shape": [ 16, 256, 192 ], "storage_unique": true }, "core_parameter_grad_rms_by_block": [ 0.0056993408498569815, 0.005586029321860134, 0.005830344684528519, 0.00540288712216694, 0.00538025811909819, 0.0052226633069546126, 0.005297605878918106, 0.0052023753769565, 0.0048107019260181325, 0.005248672753719287, 0.004962353375385338, 0.005501319898161136, 0.005588445553926757, 0.005229606380765812, 0.004884302670286647, 0.005031817046498285, 0.005000470961685071, 0.0053937604587735305, 0.004980398173225761, 0.004658922297714218, 0.00496598185096497, 0.004578385934052906, 0.004596161734371138, 0.004990442293511754, 0.005201771295872185, 0.00459813521112035, 0.004706502430387731, 0.004781878207244352, 0.0048378722082161795, 0.004473110720597102, 0.005084432926394394, 0.004685705425265166 ], "core_parameter_grad_statistics": { "first_quartile_mean": 0.005452688082542498, "first_to_last_ratio": 1.1368823875795573, "imbalance_abs_log_ratio": 0.12828976841498158, "last_quartile_mean": 0.004796176053137182, "mean": 0.005075395512328069, "normalized": [ 1.122935313319594, 1.1006096585560166, 1.1487468652180286, 1.064525337787646, 1.060066768398564, 1.0290160233362764, 1.0437818818356701, 1.025018713185997, 0.947847692723257, 1.0341406380981206, 0.9777274230809889, 1.0839194472230789, 1.1010857262951226, 1.0303840100861434, 0.9623491722808084, 0.9914137793352396, 0.9852376922229987, 1.0627271206100406, 0.9812827711906272, 0.9179427074003902, 0.9784423379227619, 0.9020747098294245, 0.9055770576316904, 0.9832617539638114, 1.024899691706223, 0.905965889742295, 0.9273173724009683, 0.9421685848185097, 0.9532010257062827, 0.8813324419214177, 1.00178063247374, 0.9232197596982638 ], "population_cv": 0.06921897980134169 }, "depth_weights": [ { "entropy_mean": 0.0, "mean_weights": [ 1.0 ], "sources": 1 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 0.6931471824645996, "mean_weights": [ 0.5, 0.5 ], "sources": 2 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.0986123085021973, "mean_weights": [ 0.3333333432674408, 0.3333333432674408, 0.3333333432674408 ], "sources": 3 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.3862943649291992, "mean_weights": [ 0.25, 0.25, 0.25, 0.25 ], "sources": 4 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.6094379425048828, "mean_weights": [ 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343, 0.20000001788139343 ], "sources": 5 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.791759729385376, "mean_weights": [ 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592, 0.1666666567325592 ], "sources": 6 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 1.9459102153778076, "mean_weights": [ 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548, 0.1428571492433548 ], "sources": 7 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.079441547393799, "mean_weights": [ 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125, 0.125 ], "sources": 8 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 } ], "layer_input_rms_by_sublayer": [ 0.02802751213312149, 0.014061377383768559, 0.01413465291261673, 0.014165290631353855, 0.014225807972252369, 0.014281988143920898, 0.01433455292135477, 0.014350161887705326, 0.01440478302538395, 0.00966146495193243, 0.009703479707241058, 0.009741888381540775, 0.00977881159633398, 0.009820535778999329, 0.009858808480203152, 0.00991273857653141, 0.009950309991836548, 0.0074927168898284435, 0.00752287870272994, 0.007546746172010899, 0.007571155205368996, 0.00766373984515667, 0.007693085819482803, 0.007762390188872814, 0.007788818795233965, 0.006325984839349985, 0.006353048607707024, 0.006375366821885109, 0.006389216054230928, 0.006445344537496567, 0.0064707002602517605, 0.006555322092026472, 0.00656962301582098, 0.005531130358576775, 0.005553483963012695, 0.005611324217170477, 0.0056249531917274, 0.0056777093559503555, 0.005688921548426151, 0.00574630219489336, 0.005768724251538515, 0.004946626722812653, 0.004966245498508215, 0.0050392793491482735, 0.005057469941675663, 0.005084683652967215, 0.005098348017781973, 0.005203457549214363, 0.005216081161051989, 0.004667261149734259, 0.0046758754178881645, 0.004717131145298481, 0.004727107938379049, 0.004768412560224533, 0.004777040798217058, 0.004826710093766451, 0.004837613552808762, 0.0044052014127373695, 0.0044186655431985855, 0.004504534415900707, 0.004510571248829365, 0.004574235528707504, 0.0045849340967834, 0.004644096828997135 ], "loss_nats": 5.516995429992676, "loss_scale": 1.0, "output_weights": { "entropy_mean": 2.1972246170043945, "mean_weights": [ 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519, 0.11111113429069519 ], "sources": 9 }, "stream_state_rms_by_sublayer": [ 0.0022265110164880753, 0.0034885562490671873, 0.0040087271481752396, 0.004831995815038681, 0.005379301495850086, 0.005952076520770788, 0.006410556845366955, 0.006925336085259914, 0.0027253651060163975, 0.00385080324485898, 0.004838948138058186, 0.005542940925806761, 0.006175771821290255, 0.006729288958013058, 0.00742313964292407, 0.007927131839096546, 0.0030937574338167906, 0.004081308841705322, 0.005150026176124811, 0.005811207927763462, 0.007019025273621082, 0.007534808944910765, 0.008143599145114422, 0.008581922389566898, 0.004255065228790045, 0.00499920267611742, 0.0066224560141563416, 0.007102519739419222, 0.008129785768687725, 0.008542952127754688, 0.010099578648805618, 0.010447698645293713, 0.0041999719105660915, 0.005041283555328846, 0.00719039561226964, 0.007636935915797949, 0.009030314162373543, 0.009387043304741383, 0.010736005380749702, 0.011020516976714134, 0.005131533369421959, 0.005807018838822842, 0.00787449348717928, 0.008367395959794521, 0.009912566281855106, 0.01024632342159748, 0.012113703414797783, 0.012378672137856483, 0.0059690922498703, 0.006453148555010557, 0.00876442901790142, 0.009127611294388771, 0.010806876234710217, 0.011037957854568958, 0.012353328987956047, 0.012611051090061665, 0.006502739619463682, 0.007010402157902718, 0.009470963850617409, 0.009794782847166061, 0.012123330496251583, 0.012445803731679916, 0.01384514570236206, 0.014087481424212456 ] }, "evaluation": { "bits_per_byte": 7.9557801867581395, "cross_entropy_nats": 5.514526605606079 }, "initial_mixer_hash": "c5a06c218c4501b16fcccee115aa8c79d5dceade78b38bbb547e4f5b2202516d", "initial_public_hash": "73bbe569e1a47981386ebaf59ca891174746bf81897fe5403676ad6d50423c58", "logits_sha256": "6b44486d93c29a13dc0b961117cc2d22f26c81dc519346ac45b1bb0fefa29939", "loss_nats": 5.516995429992676 }, "pre_reduction_uniform_weight_gate": true, "pre_reduction_uniform_weight_max_abs_error": 0.0, "selected_capture_checks": { "40": { "capture_summary_exact_vs_learned": true, "expected_source_count": 6, "passed": true, "source_count": 6 }, "41": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "42": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "43": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "44": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "45": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "46": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "47": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "48": { "capture_summary_exact_vs_learned": true, "expected_source_count": 7, "passed": true, "source_count": 7 }, "49": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "50": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "51": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "52": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "53": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "54": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 }, "55": { "capture_summary_exact_vs_learned": true, "expected_source_count": 8, "passed": true, "source_count": 8 } } } } }