{ "training_seed": 42, "checkpoint_step": 6000, "checkpoint_sha256": "88a53a8f128c9f09cc053dc461ffab24ecdc008074ace4042ff212d14069c6db", "tokenizer_sha256": "ef7f54d89966a08485006e0f64eaf2b0b99217b65ad79f96979f12adbd9ca49d", "graph_sha256": "a2f35369a5ed4981b8ef75c92df285f7df576a020582f119a1372bfbedbaebee", "validation_cache_sha256": "a9fe379a68775dc5808338d69d3061c0dfbf2c1e7ae64b8a1f1f55d57e4e6075", "conditions": [ { "condition": "untrained", "score": { "bits_per_byte": 3.4640037142576174, "token_perplexity": 5263.335142199168, "nll": 280773.2686767578, "tokens": 32768, "bytes": 116937, "documents": [ { "document": "wikitext2-validation-0000", "nll": 4685.850891113281, "tokens": 547, "bytes": 1731, "bits_per_byte": 3.905403722099516 }, { "document": "wikitext2-validation-0001", "nll": 4676.654357910156, "tokens": 547, "bytes": 1826, "bits_per_byte": 3.6949540252512274 }, { "document": "wikitext2-validation-0002", "nll": 4689.857727050781, "tokens": 547, "bytes": 1916, "bits_per_byte": 3.5313332386695975 }, { "document": "wikitext2-validation-0003", "nll": 4759.583435058594, "tokens": 547, "bytes": 1715, "bits_per_byte": 4.003864383939528 }, { "document": "wikitext2-validation-0004", "nll": 4687.0784912109375, "tokens": 547, "bytes": 1862, "bits_per_byte": 3.6315923176838587 }, { "document": "wikitext2-validation-0005", "nll": 4720.221618652344, "tokens": 547, "bytes": 1844, "bits_per_byte": 3.6929719745805927 }, { "document": "wikitext2-validation-0006", "nll": 4701.788330078125, "tokens": 547, "bytes": 1907, "bits_per_byte": 3.5570250168397024 }, { "document": "wikitext2-validation-0007", "nll": 4660.239562988281, "tokens": 547, "bytes": 2019, "bits_per_byte": 3.330017091073672 }, { "document": "wikitext2-validation-0008", "nll": 4642.6434326171875, "tokens": 546, "bytes": 2154, "bits_per_byte": 3.109525838835901 }, { "document": "wikitext2-validation-0009", "nll": 4645.378112792969, "tokens": 546, "bytes": 1951, "bits_per_byte": 3.435091730589722 }, { "document": "wikitext2-validation-0010", "nll": 4657.488037109375, "tokens": 546, "bytes": 2086, "bits_per_byte": 3.221157667371701 }, { "document": "wikitext2-validation-0011", "nll": 4671.4903564453125, "tokens": 546, "bytes": 1980, "bits_per_byte": 3.4038060458607418 }, { "document": "wikitext2-validation-0012", "nll": 4640.072509765625, "tokens": 546, "bytes": 1879, "bits_per_byte": 3.5626448106461277 }, { "document": "wikitext2-validation-0013", "nll": 4687.891418457031, "tokens": 546, "bytes": 1900, "bits_per_byte": 3.559577737702046 }, { "document": "wikitext2-validation-0014", "nll": 4707.720886230469, "tokens": 546, "bytes": 1606, "bits_per_byte": 4.229019661552984 }, { "document": "wikitext2-validation-0015", "nll": 4636.122741699219, "tokens": 546, "bytes": 1986, "bits_per_byte": 3.3678304574028237 }, { "document": "wikitext2-validation-0016", "nll": 4760.514221191406, "tokens": 546, "bytes": 1881, "bits_per_byte": 3.651233524186192 }, { "document": "wikitext2-validation-0017", "nll": 4658.1572265625, "tokens": 546, "bytes": 1793, "bits_per_byte": 3.748076034825882 }, { "document": "wikitext2-validation-0018", "nll": 4659.1793212890625, "tokens": 546, "bytes": 1897, "bits_per_byte": 3.543371060324796 }, { "document": "wikitext2-validation-0019", "nll": 4643.0025634765625, "tokens": 546, "bytes": 2052, "bits_per_byte": 3.2643454060245523 }, { "document": "wikitext2-validation-0020", "nll": 4646.6201171875, "tokens": 546, "bytes": 1916, "bits_per_byte": 3.498776513549739 }, { "document": "wikitext2-validation-0021", "nll": 4651.7197265625, "tokens": 546, "bytes": 1537, "bits_per_byte": 4.3663064288334965 }, { "document": "wikitext2-validation-0022", "nll": 4671.7255859375, "tokens": 546, "bytes": 1969, "bits_per_byte": 3.4229940757877695 }, { "document": "wikitext2-validation-0023", "nll": 4667.725158691406, "tokens": 546, "bytes": 1969, "bits_per_byte": 3.4200629449856486 }, { "document": "wikitext2-validation-0024", "nll": 4673.913269042969, "tokens": 546, "bytes": 1962, "bits_per_byte": 3.436815236897765 }, { "document": "wikitext2-validation-0025", "nll": 4720.4652099609375, "tokens": 546, "bytes": 1846, "bits_per_byte": 3.6891612942034255 }, { "document": "wikitext2-validation-0026", "nll": 4644.072998046875, "tokens": 546, "bytes": 2274, "bits_per_byte": 2.946341725509485 }, { "document": "wikitext2-validation-0027", "nll": 4664.446472167969, "tokens": 546, "bytes": 1927, "bits_per_byte": 3.492150386034639 }, { "document": "wikitext2-validation-0028", "nll": 4668.872741699219, "tokens": 546, "bytes": 2288, "bits_per_byte": 2.94395085270591 }, { "document": "wikitext2-validation-0029", "nll": 4680.508850097656, "tokens": 546, "bytes": 1975, "bits_per_byte": 3.4190110920874903 }, { "document": "wikitext2-validation-0030", "nll": 4637.1549072265625, "tokens": 546, "bytes": 1938, "bits_per_byte": 3.4520125843600016 }, { "document": "wikitext2-validation-0031", "nll": 4722.5340576171875, "tokens": 546, "bytes": 1951, "bits_per_byte": 3.4921458048967455 }, { "document": "wikitext2-validation-0032", "nll": 4663.4959716796875, "tokens": 546, "bytes": 2203, "bits_per_byte": 3.054018389263706 }, { "document": "wikitext2-validation-0033", "nll": 4707.970764160156, "tokens": 546, "bytes": 1992, "bits_per_byte": 3.409721924751045 }, { "document": "wikitext2-validation-0034", "nll": 4678.0123291015625, "tokens": 546, "bytes": 2129, "bits_per_byte": 3.170007134059302 }, { "document": "wikitext2-validation-0035", "nll": 4678.795104980469, "tokens": 546, "bytes": 1798, "bits_per_byte": 3.7542127337546605 }, { "document": "wikitext2-validation-0036", "nll": 4677.38818359375, "tokens": 546, "bytes": 1972, "bits_per_byte": 3.421929379707576 }, { "document": "wikitext2-validation-0037", "nll": 4699.4266357421875, "tokens": 546, "bytes": 1888, "bits_per_byte": 3.5910166855968004 }, { "document": "wikitext2-validation-0038", "nll": 4714.28466796875, "tokens": 546, "bytes": 2295, "bits_per_byte": 2.9635185672406923 }, { "document": "wikitext2-validation-0039", "nll": 4704.0958251953125, "tokens": 546, "bytes": 1895, "bits_per_byte": 3.5813064479555425 }, { "document": "wikitext2-validation-0040", "nll": 4642.3819580078125, "tokens": 546, "bytes": 1967, "bits_per_byte": 3.4049524294510767 }, { "document": "wikitext2-validation-0041", "nll": 4681.5931396484375, "tokens": 546, "bytes": 1907, "bits_per_byte": 3.5417468306400592 }, { "document": "wikitext2-validation-0042", "nll": 4717.453186035156, "tokens": 546, "bytes": 2009, "bits_per_byte": 3.387678604837611 }, { "document": "wikitext2-validation-0043", "nll": 4674.180725097656, "tokens": 546, "bytes": 2026, "bits_per_byte": 3.328438969554377 }, { "document": "wikitext2-validation-0044", "nll": 4674.486633300781, "tokens": 546, "bytes": 1823, "bits_per_byte": 3.6993190809461236 }, { "document": "wikitext2-validation-0045", "nll": 4676.501403808594, "tokens": 546, "bytes": 1929, "bits_per_byte": 3.49754555934937 }, { "document": "wikitext2-validation-0046", "nll": 4721.3353271484375, "tokens": 546, "bytes": 2289, "bits_per_byte": 2.9757304774359645 }, { "document": "wikitext2-validation-0047", "nll": 4660.8695068359375, "tokens": 546, "bytes": 2075, "bits_per_byte": 3.2405847343338774 }, { "document": "wikitext2-validation-0048", "nll": 4657.006896972656, "tokens": 546, "bytes": 1923, "bits_per_byte": 3.493832946254889 }, { "document": "wikitext2-validation-0049", "nll": 4637.124816894531, "tokens": 546, "bytes": 1975, "bits_per_byte": 3.3873199885148773 }, { "document": "wikitext2-validation-0050", "nll": 4640.23876953125, "tokens": 546, "bytes": 2169, "bits_per_byte": 3.08642206608734 }, { "document": "wikitext2-validation-0051", "nll": 4720.106689453125, "tokens": 546, "bytes": 1894, "bits_per_byte": 3.5953930904650715 }, { "document": "wikitext2-validation-0052", "nll": 4654.407409667969, "tokens": 546, "bytes": 2000, "bits_per_byte": 3.3574452441024123 }, { "document": "wikitext2-validation-0053", "nll": 4756.08447265625, "tokens": 546, "bytes": 1981, "bits_per_byte": 3.4636948423776746 }, { "document": "wikitext2-validation-0054", "nll": 4699.072692871094, "tokens": 546, "bytes": 2004, "bits_per_byte": 3.3828986381147077 }, { "document": "wikitext2-validation-0055", "nll": 4715.601989746094, "tokens": 546, "bytes": 1841, "bits_per_byte": 3.695369693325811 }, { "document": "wikitext2-validation-0056", "nll": 4669.879638671875, "tokens": 546, "bytes": 1944, "bits_per_byte": 3.465644133878734 }, { "document": "wikitext2-validation-0057", "nll": 4650.076965332031, "tokens": 546, "bytes": 1740, "bits_per_byte": 3.855541941170415 }, { "document": "wikitext2-validation-0058", "nll": 4678.65771484375, "tokens": 546, "bytes": 1901, "bits_per_byte": 3.550697676602824 }, { "document": "wikitext2-validation-0059", "nll": 4682.044921875, "tokens": 546, "bytes": 1831, "bits_per_byte": 3.689111409070681 } ], "seconds": 27.955396899953485, "tokenizer_sha256": "ef7f54d89966a08485006e0f64eaf2b0b99217b65ad79f96979f12adbd9ca49d", "subset": "fixed per-article token prefixes", "protocol": "Reset per article; carry native state within article; score IDs >=2; normalize by exact decoded UTF-8 byte lengths; boundary tokens excluded" } }, { "condition": "trained", "score": { "bits_per_byte": 1.9096990391099853, "token_perplexity": 112.5964072527091, "nll": 154789.7997894287, "tokens": 32768, "bytes": 116937, "documents": [ { "document": "wikitext2-validation-0000", "nll": 2494.740020751953, "tokens": 547, "bytes": 1731, "bits_per_byte": 2.079231112793803 }, { "document": "wikitext2-validation-0001", "nll": 2412.133331298828, "tokens": 547, "bytes": 1826, "bits_per_byte": 1.9057901396647285 }, { "document": "wikitext2-validation-0002", "nll": 2612.180633544922, "tokens": 547, "bytes": 1916, "bits_per_byte": 1.9668998151990864 }, { "document": "wikitext2-validation-0003", "nll": 2591.303466796875, "tokens": 547, "bytes": 1715, "bits_per_byte": 2.1798604437237494 }, { "document": "wikitext2-validation-0004", "nll": 2258.1593017578125, "tokens": 547, "bytes": 1862, "bits_per_byte": 1.749642978616155 }, { "document": "wikitext2-validation-0005", "nll": 2604.6517944335938, "tokens": 547, "bytes": 1844, "bits_per_byte": 2.0378081491713043 }, { "document": "wikitext2-validation-0006", "nll": 2773.4815368652344, "tokens": 547, "bytes": 1907, "bits_per_byte": 2.098210833367894 }, { "document": "wikitext2-validation-0007", "nll": 2572.2196044921875, "tokens": 547, "bytes": 2019, "bits_per_byte": 1.8380032033077018 }, { "document": "wikitext2-validation-0008", "nll": 2382.6592712402344, "tokens": 546, "bytes": 2154, "bits_per_byte": 1.5958452714700084 }, { "document": "wikitext2-validation-0009", "nll": 2603.7994384765625, "tokens": 546, "bytes": 1951, "bits_per_byte": 1.9254169848075882 }, { "document": "wikitext2-validation-0010", "nll": 2407.3597564697266, "tokens": 546, "bytes": 2086, "bits_per_byte": 1.6649501353281577 }, { "document": "wikitext2-validation-0011", "nll": 2495.4834899902344, "tokens": 546, "bytes": 1980, "bits_per_byte": 1.8182937654692901 }, { "document": "wikitext2-validation-0012", "nll": 2743.381134033203, "tokens": 546, "bytes": 1879, "bits_per_byte": 2.1063663423832053 }, { "document": "wikitext2-validation-0013", "nll": 2636.346893310547, "tokens": 546, "bytes": 1900, "bits_per_byte": 2.0018129416011323 }, { "document": "wikitext2-validation-0014", "nll": 2496.657196044922, "tokens": 546, "bytes": 1606, "bits_per_byte": 2.2427863982152885 }, { "document": "wikitext2-validation-0015", "nll": 2455.4090270996094, "tokens": 546, "bytes": 1986, "bits_per_byte": 1.7836890366317226 }, { "document": "wikitext2-validation-0016", "nll": 2400.9591064453125, "tokens": 546, "bytes": 1881, "bits_per_byte": 1.8414948411727003 }, { "document": "wikitext2-validation-0017", "nll": 2992.892364501953, "tokens": 546, "bytes": 1793, "bits_per_byte": 2.4081600514118304 }, { "document": "wikitext2-validation-0018", "nll": 2725.791046142578, "tokens": 546, "bytes": 1897, "bits_per_byte": 2.0730022270793027 }, { "document": "wikitext2-validation-0019", "nll": 2684.322540283203, "tokens": 546, "bytes": 2052, "bits_per_byte": 1.8872606320726324 }, { "document": "wikitext2-validation-0020", "nll": 2571.9927673339844, "tokens": 546, "bytes": 1916, "bits_per_byte": 1.9366394627531422 }, { "document": "wikitext2-validation-0021", "nll": 2466.680419921875, "tokens": 546, "bytes": 1537, "bits_per_byte": 2.315333512868702 }, { "document": "wikitext2-validation-0022", "nll": 2516.4673461914062, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.843826795789412 }, { "document": "wikitext2-validation-0023", "nll": 2373.2543334960938, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.7388939856287742 }, { "document": "wikitext2-validation-0024", "nll": 2710.406036376953, "tokens": 546, "bytes": 1962, "bits_per_byte": 1.9930118998351385 }, { "document": "wikitext2-validation-0025", "nll": 2605.2749633789062, "tokens": 546, "bytes": 1846, "bits_per_byte": 2.0360873617653974 }, { "document": "wikitext2-validation-0026", "nll": 2248.5079345703125, "tokens": 546, "bytes": 2274, "bits_per_byte": 1.4265220961319593 }, { "document": "wikitext2-validation-0027", "nll": 2534.2969970703125, "tokens": 546, "bytes": 1927, "bits_per_byte": 1.897362589420411 }, { "document": "wikitext2-validation-0028", "nll": 2543.7784423828125, "tokens": 546, "bytes": 2288, "bits_per_byte": 1.6039757622141326 }, { "document": "wikitext2-validation-0029", "nll": 2473.7238159179688, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.807002066710773 }, { "document": "wikitext2-validation-0030", "nll": 2578.061309814453, "tokens": 546, "bytes": 1938, "bits_per_byte": 1.9191724802770989 }, { "document": "wikitext2-validation-0031", "nll": 2650.1937255859375, "tokens": 546, "bytes": 1951, "bits_per_byte": 1.959723908404858 }, { "document": "wikitext2-validation-0032", "nll": 2581.355194091797, "tokens": 546, "bytes": 2203, "bits_per_byte": 1.6904713287740367 }, { "document": "wikitext2-validation-0033", "nll": 2693.183563232422, "tokens": 546, "bytes": 1992, "bits_per_byte": 1.950523378955363 }, { "document": "wikitext2-validation-0034", "nll": 2418.191925048828, "tokens": 546, "bytes": 2129, "bits_per_byte": 1.6386629864657962 }, { "document": "wikitext2-validation-0035", "nll": 2673.490936279297, "tokens": 546, "bytes": 1798, "bits_per_byte": 2.145179152186726 }, { "document": "wikitext2-validation-0036", "nll": 2624.0311584472656, "tokens": 546, "bytes": 1972, "bits_per_byte": 1.919714370907704 }, { "document": "wikitext2-validation-0037", "nll": 2401.4349670410156, "tokens": 546, "bytes": 1888, "bits_per_byte": 1.8350308887539326 }, { "document": "wikitext2-validation-0038", "nll": 2519.8751525878906, "tokens": 546, "bytes": 2295, "bits_per_byte": 1.5840572489315337 }, { "document": "wikitext2-validation-0039", "nll": 2599.6799926757812, "tokens": 546, "bytes": 1895, "bits_per_byte": 1.9791796481960982 }, { "document": "wikitext2-validation-0040", "nll": 2589.8274841308594, "tokens": 546, "bytes": 1967, "bits_per_byte": 1.8995075079377384 }, { "document": "wikitext2-validation-0041", "nll": 2681.9029235839844, "tokens": 546, "bytes": 1907, "bits_per_byte": 2.0289292333509317 }, { "document": "wikitext2-validation-0042", "nll": 2489.5740966796875, "tokens": 546, "bytes": 2009, "bits_per_byte": 1.7878029881560011 }, { "document": "wikitext2-validation-0043", "nll": 2463.123046875, "tokens": 546, "bytes": 2026, "bits_per_byte": 1.7539661425596624 }, { "document": "wikitext2-validation-0044", "nll": 2886.5573120117188, "tokens": 546, "bytes": 1823, "bits_per_byte": 2.2843784527049276 }, { "document": "wikitext2-validation-0045", "nll": 2770.3184814453125, "tokens": 546, "bytes": 1929, "bits_per_byte": 2.071915362811921 }, { "document": "wikitext2-validation-0046", "nll": 2477.9560546875, "tokens": 546, "bytes": 2289, "bits_per_byte": 1.5617889522229957 }, { "document": "wikitext2-validation-0047", "nll": 2619.4644470214844, "tokens": 546, "bytes": 2075, "bits_per_byte": 1.8212474060254682 }, { "document": "wikitext2-validation-0048", "nll": 2740.855438232422, "tokens": 546, "bytes": 1923, "bits_per_byte": 2.056275896272211 }, { "document": "wikitext2-validation-0049", "nll": 2318.9285583496094, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.6939274588895064 }, { "document": "wikitext2-validation-0050", "nll": 2480.8716735839844, "tokens": 546, "bytes": 2169, "bits_per_byte": 1.6501342833386434 }, { "document": "wikitext2-validation-0051", "nll": 2705.882080078125, "tokens": 546, "bytes": 1894, "bits_per_byte": 2.06112072764468 }, { "document": "wikitext2-validation-0052", "nll": 2753.831329345703, "tokens": 546, "bytes": 2000, "bits_per_byte": 1.986469401145854 }, { "document": "wikitext2-validation-0053", "nll": 2778.3965759277344, "tokens": 546, "bytes": 1981, "bits_per_byte": 2.0234118938484698 }, { "document": "wikitext2-validation-0054", "nll": 2575.4087524414062, "tokens": 546, "bytes": 2004, "bits_per_byte": 1.854056604495633 }, { "document": "wikitext2-validation-0055", "nll": 2832.7428283691406, "tokens": 546, "bytes": 1841, "bits_per_byte": 2.219871825422018 }, { "document": "wikitext2-validation-0056", "nll": 2531.077667236328, "tokens": 546, "bytes": 1944, "bits_per_byte": 1.878381275013712 }, { "document": "wikitext2-validation-0057", "nll": 2733.337127685547, "tokens": 546, "bytes": 1740, "bits_per_byte": 2.2663057006836906 }, { "document": "wikitext2-validation-0058", "nll": 2535.987060546875, "tokens": 546, "bytes": 1901, "bits_per_byte": 1.9245954529245428 }, { "document": "wikitext2-validation-0059", "nll": 2695.9449157714844, "tokens": 546, "bytes": 1831, "bits_per_byte": 2.1242088260476977 } ], "seconds": 26.750106199877337, "tokenizer_sha256": "ef7f54d89966a08485006e0f64eaf2b0b99217b65ad79f96979f12adbd9ca49d", "subset": "fixed per-article token prefixes", "protocol": "Reset per article; carry native state within article; score IDs >=2; normalize by exact decoded UTF-8 byte lengths; boundary tokens excluded" } }, { "condition": "recurrence_off", "score": { "bits_per_byte": 2.06313332807702, "token_perplexity": 164.5711696535346, "nll": 167226.34732055664, "tokens": 32768, "bytes": 116937, "documents": [ { "document": "wikitext2-validation-0000", "nll": 2729.6156005859375, "tokens": 547, "bytes": 1731, "bits_per_byte": 2.2749872273243685 }, { "document": "wikitext2-validation-0001", "nll": 2665.3170166015625, "tokens": 547, "bytes": 1826, "bits_per_byte": 2.105826748219081 }, { "document": "wikitext2-validation-0002", "nll": 2880.15673828125, "tokens": 547, "bytes": 1916, "bits_per_byte": 2.1686784150841816 }, { "document": "wikitext2-validation-0003", "nll": 2848.782196044922, "tokens": 547, "bytes": 1715, "bits_per_byte": 2.3964571118406877 }, { "document": "wikitext2-validation-0004", "nll": 2439.7390747070312, "tokens": 547, "bytes": 1862, "bits_per_byte": 1.8903326875095932 }, { "document": "wikitext2-validation-0005", "nll": 2761.7628173828125, "tokens": 547, "bytes": 1844, "bits_per_byte": 2.160727505829564 }, { "document": "wikitext2-validation-0006", "nll": 2979.4857177734375, "tokens": 547, "bytes": 1907, "bits_per_byte": 2.2540583478926233 }, { "document": "wikitext2-validation-0007", "nll": 2764.718505859375, "tokens": 547, "bytes": 2019, "bits_per_byte": 1.9755550658034993 }, { "document": "wikitext2-validation-0008", "nll": 2602.847381591797, "tokens": 546, "bytes": 2154, "bits_per_byte": 1.7433217314825018 }, { "document": "wikitext2-validation-0009", "nll": 2768.9053344726562, "tokens": 546, "bytes": 1951, "bits_per_byte": 2.047506916829676 }, { "document": "wikitext2-validation-0010", "nll": 2612.596954345703, "tokens": 546, "bytes": 2086, "bits_per_byte": 1.8068938973519444 }, { "document": "wikitext2-validation-0011", "nll": 2700.6493530273438, "tokens": 546, "bytes": 1980, "bits_per_byte": 1.9677845599962305 }, { "document": "wikitext2-validation-0012", "nll": 2930.0675048828125, "tokens": 546, "bytes": 1879, "bits_per_byte": 2.249704022759091 }, { "document": "wikitext2-validation-0013", "nll": 2842.897216796875, "tokens": 546, "bytes": 1900, "bits_per_byte": 2.1586493244367833 }, { "document": "wikitext2-validation-0014", "nll": 2700.480926513672, "tokens": 546, "bytes": 1606, "bits_per_byte": 2.4258844587151356 }, { "document": "wikitext2-validation-0015", "nll": 2647.1298217773438, "tokens": 546, "bytes": 1986, "bits_per_byte": 1.9229612620682073 }, { "document": "wikitext2-validation-0016", "nll": 2605.3638610839844, "tokens": 546, "bytes": 1881, "bits_per_byte": 1.998269815043693 }, { "document": "wikitext2-validation-0017", "nll": 3268.0674743652344, "tokens": 546, "bytes": 1793, "bits_per_byte": 2.629573194956633 }, { "document": "wikitext2-validation-0018", "nll": 2914.5210571289062, "tokens": 546, "bytes": 1897, "bits_per_byte": 2.2165340409522045 }, { "document": "wikitext2-validation-0019", "nll": 2916.764404296875, "tokens": 546, "bytes": 2052, "bits_per_byte": 2.050683012436917 }, { "document": "wikitext2-validation-0020", "nll": 2841.1431274414062, "tokens": 546, "bytes": 1916, "bits_per_byte": 2.1393022444757186 }, { "document": "wikitext2-validation-0021", "nll": 2690.5043334960938, "tokens": 546, "bytes": 1537, "bits_per_byte": 2.525424371779493 }, { "document": "wikitext2-validation-0022", "nll": 2720.117462158203, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.9930421398126523 }, { "document": "wikitext2-validation-0023", "nll": 2652.8833618164062, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.9437794160230428 }, { "document": "wikitext2-validation-0024", "nll": 2876.2102661132812, "tokens": 546, "bytes": 1962, "bits_per_byte": 2.114930829498245 }, { "document": "wikitext2-validation-0025", "nll": 2792.56005859375, "tokens": 546, "bytes": 1846, "bits_per_byte": 2.1824553347333677 }, { "document": "wikitext2-validation-0026", "nll": 2380.476837158203, "tokens": 546, "bytes": 2274, "bits_per_byte": 1.510247197853643 }, { "document": "wikitext2-validation-0027", "nll": 2725.3926391601562, "tokens": 546, "bytes": 1927, "bits_per_byte": 2.0404309522530575 }, { "document": "wikitext2-validation-0028", "nll": 2726.573516845703, "tokens": 546, "bytes": 2288, "bits_per_byte": 1.7192369280474111 }, { "document": "wikitext2-validation-0029", "nll": 2677.0089111328125, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.9554974584844924 }, { "document": "wikitext2-validation-0030", "nll": 2800.838104248047, "tokens": 546, "bytes": 1938, "bits_per_byte": 2.085013025454852 }, { "document": "wikitext2-validation-0031", "nll": 2885.8955078125, "tokens": 546, "bytes": 1951, "bits_per_byte": 2.134016984953783 }, { "document": "wikitext2-validation-0032", "nll": 2743.6390380859375, "tokens": 546, "bytes": 2203, "bits_per_byte": 1.7967473600707888 }, { "document": "wikitext2-validation-0033", "nll": 2877.3960876464844, "tokens": 546, "bytes": 1992, "bits_per_byte": 2.083938286305667 }, { "document": "wikitext2-validation-0034", "nll": 2633.3711547851562, "tokens": 546, "bytes": 2129, "bits_per_byte": 1.7844769872374768 }, { "document": "wikitext2-validation-0035", "nll": 2861.5111083984375, "tokens": 546, "bytes": 1798, "bits_per_byte": 2.296044430219748 }, { "document": "wikitext2-validation-0036", "nll": 2796.8141479492188, "tokens": 546, "bytes": 1972, "bits_per_byte": 2.0461206397233416 }, { "document": "wikitext2-validation-0037", "nll": 2607.7124938964844, "tokens": 546, "bytes": 1888, "bits_per_byte": 1.9926556583732253 }, { "document": "wikitext2-validation-0038", "nll": 2700.273681640625, "tokens": 546, "bytes": 2295, "bits_per_byte": 1.6974603265995265 }, { "document": "wikitext2-validation-0039", "nll": 2832.528564453125, "tokens": 546, "bytes": 1895, "bits_per_byte": 2.156451141484358 }, { "document": "wikitext2-validation-0040", "nll": 2809.834259033203, "tokens": 546, "bytes": 1967, "bits_per_byte": 2.0608713529370197 }, { "document": "wikitext2-validation-0041", "nll": 2883.7970275878906, "tokens": 546, "bytes": 1907, "bits_per_byte": 2.181667367924165 }, { "document": "wikitext2-validation-0042", "nll": 2690.1819763183594, "tokens": 546, "bytes": 2009, "bits_per_byte": 1.9318627159399542 }, { "document": "wikitext2-validation-0043", "nll": 2671.88134765625, "tokens": 546, "bytes": 2026, "bits_per_byte": 1.9026209131823257 }, { "document": "wikitext2-validation-0044", "nll": 3150.864227294922, "tokens": 546, "bytes": 1823, "bits_per_byte": 2.4935470078073614 }, { "document": "wikitext2-validation-0045", "nll": 2964.4891662597656, "tokens": 546, "bytes": 1929, "bits_per_byte": 2.2171352093997 }, { "document": "wikitext2-validation-0046", "nll": 2614.318603515625, "tokens": 546, "bytes": 2289, "bits_per_byte": 1.6477345935324388 }, { "document": "wikitext2-validation-0047", "nll": 2809.9388427734375, "tokens": 546, "bytes": 2075, "bits_per_byte": 1.9536794379134994 }, { "document": "wikitext2-validation-0048", "nll": 2998.9278564453125, "tokens": 546, "bytes": 1923, "bits_per_byte": 2.24988993576569 }, { "document": "wikitext2-validation-0049", "nll": 2539.5824584960938, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.8551103892662868 }, { "document": "wikitext2-validation-0050", "nll": 2653.758270263672, "tokens": 546, "bytes": 2169, "bits_per_byte": 1.7651285828619054 }, { "document": "wikitext2-validation-0051", "nll": 2911.8966674804688, "tokens": 546, "bytes": 1894, "bits_per_byte": 2.2180458720988234 }, { "document": "wikitext2-validation-0052", "nll": 2909.857635498047, "tokens": 546, "bytes": 2000, "bits_per_byte": 2.0990185902129586 }, { "document": "wikitext2-validation-0053", "nll": 3010.3087463378906, "tokens": 546, "bytes": 1981, "bits_per_byte": 2.192305653652876 }, { "document": "wikitext2-validation-0054", "nll": 2759.4067993164062, "tokens": 546, "bytes": 2004, "bits_per_byte": 1.9865182161522286 }, { "document": "wikitext2-validation-0055", "nll": 3105.1803283691406, "tokens": 546, "bytes": 1841, "bits_per_byte": 2.4333667902249445 }, { "document": "wikitext2-validation-0056", "nll": 2727.6742248535156, "tokens": 546, "bytes": 1944, "bits_per_byte": 2.024280903990131 }, { "document": "wikitext2-validation-0057", "nll": 3007.5564880371094, "tokens": 546, "bytes": 1740, "bits_per_byte": 2.4936705922325086 }, { "document": "wikitext2-validation-0058", "nll": 2714.0936279296875, "tokens": 546, "bytes": 1901, "bits_per_byte": 2.0597629760770624 }, { "document": "wikitext2-validation-0059", "nll": 2890.0794067382812, "tokens": 546, "bytes": 1831, "bits_per_byte": 2.2771727077425643 } ], "seconds": 14.292471599997953, "tokenizer_sha256": "ef7f54d89966a08485006e0f64eaf2b0b99217b65ad79f96979f12adbd9ca49d", "subset": "fixed per-article token prefixes", "protocol": "Reset per article; carry native state within article; score IDs >=2; normalize by exact decoded UTF-8 byte lengths; boundary tokens excluded" } }, { "condition": "slow_state_off", "score": { "bits_per_byte": 1.9951740385172911, "token_perplexity": 139.1064953987976, "nll": 161717.93756103516, "tokens": 32768, "bytes": 116937, "documents": [ { "document": "wikitext2-validation-0000", "nll": 2680.208709716797, "tokens": 547, "bytes": 1731, "bits_per_byte": 2.23380925133208 }, { "document": "wikitext2-validation-0001", "nll": 2504.954620361328, "tokens": 547, "bytes": 1826, "bits_per_byte": 1.9791268392372312 }, { "document": "wikitext2-validation-0002", "nll": 2763.5161743164062, "tokens": 547, "bytes": 1916, "bits_per_byte": 2.0808512944168682 }, { "document": "wikitext2-validation-0003", "nll": 2724.4888305664062, "tokens": 547, "bytes": 1715, "bits_per_byte": 2.291898848288936 }, { "document": "wikitext2-validation-0004", "nll": 2387.646026611328, "tokens": 547, "bytes": 1862, "bits_per_byte": 1.8499705059024711 }, { "document": "wikitext2-validation-0005", "nll": 2710.185760498047, "tokens": 547, "bytes": 1844, "bits_per_byte": 2.120375030671592 }, { "document": "wikitext2-validation-0006", "nll": 2888.9236755371094, "tokens": 547, "bytes": 1907, "bits_per_byte": 2.1855458102800758 }, { "document": "wikitext2-validation-0007", "nll": 2700.948272705078, "tokens": 547, "bytes": 2019, "bits_per_byte": 1.9299874585087806 }, { "document": "wikitext2-validation-0008", "nll": 2514.80712890625, "tokens": 546, "bytes": 2154, "bits_per_byte": 1.684354583874308 }, { "document": "wikitext2-validation-0009", "nll": 2660.9430541992188, "tokens": 546, "bytes": 1951, "bits_per_byte": 1.9676726542189364 }, { "document": "wikitext2-validation-0010", "nll": 2534.457550048828, "tokens": 546, "bytes": 2086, "bits_per_byte": 1.7528520320225485 }, { "document": "wikitext2-validation-0011", "nll": 2617.3624877929688, "tokens": 546, "bytes": 1980, "bits_per_byte": 1.907098929973594 }, { "document": "wikitext2-validation-0012", "nll": 2803.4515075683594, "tokens": 546, "bytes": 1879, "bits_per_byte": 2.1524883381274935 }, { "document": "wikitext2-validation-0013", "nll": 2728.761993408203, "tokens": 546, "bytes": 1900, "bits_per_byte": 2.0719849450822614 }, { "document": "wikitext2-validation-0014", "nll": 2706.4403076171875, "tokens": 546, "bytes": 1606, "bits_per_byte": 2.4312378644217416 }, { "document": "wikitext2-validation-0015", "nll": 2524.0362548828125, "tokens": 546, "bytes": 1986, "bits_per_byte": 1.8335420885918357 }, { "document": "wikitext2-validation-0016", "nll": 2599.767822265625, "tokens": 546, "bytes": 1881, "bits_per_byte": 1.9939777483494512 }, { "document": "wikitext2-validation-0017", "nll": 3096.6900939941406, "tokens": 546, "bytes": 1793, "bits_per_byte": 2.4916784393615865 }, { "document": "wikitext2-validation-0018", "nll": 2849.0580444335938, "tokens": 546, "bytes": 1897, "bits_per_byte": 2.1667485039057213 }, { "document": "wikitext2-validation-0019", "nll": 2742.161163330078, "tokens": 546, "bytes": 2052, "bits_per_byte": 1.9279251031455238 }, { "document": "wikitext2-validation-0020", "nll": 2745.7763061523438, "tokens": 546, "bytes": 1916, "bits_per_byte": 2.0674936640273502 }, { "document": "wikitext2-validation-0021", "nll": 2653.8081970214844, "tokens": 546, "bytes": 1537, "bits_per_byte": 2.4909797822468294 }, { "document": "wikitext2-validation-0022", "nll": 2643.781524658203, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.9371104595318702 }, { "document": "wikitext2-validation-0023", "nll": 2530.02001953125, "tokens": 546, "bytes": 1969, "bits_per_byte": 1.8537568997092597 }, { "document": "wikitext2-validation-0024", "nll": 2779.01416015625, "tokens": 546, "bytes": 1962, "bits_per_byte": 2.043460727531921 }, { "document": "wikitext2-validation-0025", "nll": 2707.035186767578, "tokens": 546, "bytes": 1846, "bits_per_byte": 2.1156155143345146 }, { "document": "wikitext2-validation-0026", "nll": 2448.3475646972656, "tokens": 546, "bytes": 2274, "bits_per_byte": 1.5533064599653985 }, { "document": "wikitext2-validation-0027", "nll": 2663.779510498047, "tokens": 546, "bytes": 1927, "bits_per_byte": 1.9943027969990466 }, { "document": "wikitext2-validation-0028", "nll": 2628.3040161132812, "tokens": 546, "bytes": 2288, "bits_per_byte": 1.6572732386342557 }, { "document": "wikitext2-validation-0029", "nll": 2534.7482299804688, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.8515790892632407 }, { "document": "wikitext2-validation-0030", "nll": 2707.8090209960938, "tokens": 546, "bytes": 1938, "bits_per_byte": 2.015759879393944 }, { "document": "wikitext2-validation-0031", "nll": 2760.73974609375, "tokens": 546, "bytes": 1951, "bits_per_byte": 2.0414687549331156 }, { "document": "wikitext2-validation-0032", "nll": 2688.324005126953, "tokens": 546, "bytes": 2203, "bits_per_byte": 1.7605227918744493 }, { "document": "wikitext2-validation-0033", "nll": 2789.6294860839844, "tokens": 546, "bytes": 1992, "bits_per_byte": 2.0203738079774056 }, { "document": "wikitext2-validation-0034", "nll": 2507.3457641601562, "tokens": 546, "bytes": 2129, "bits_per_byte": 1.699077172262943 }, { "document": "wikitext2-validation-0035", "nll": 2791.3067016601562, "tokens": 546, "bytes": 1798, "bits_per_byte": 2.239713201382223 }, { "document": "wikitext2-validation-0036", "nll": 2700.775115966797, "tokens": 546, "bytes": 1972, "bits_per_byte": 1.9758594657006152 }, { "document": "wikitext2-validation-0037", "nll": 2529.344024658203, "tokens": 546, "bytes": 1888, "bits_per_byte": 1.932771229383751 }, { "document": "wikitext2-validation-0038", "nll": 2690.9923706054688, "tokens": 546, "bytes": 2295, "bits_per_byte": 1.6916258597571006 }, { "document": "wikitext2-validation-0039", "nll": 2709.904510498047, "tokens": 546, "bytes": 1895, "bits_per_byte": 2.063095408220668 }, { "document": "wikitext2-validation-0040", "nll": 2641.3834228515625, "tokens": 546, "bytes": 1967, "bits_per_byte": 1.937321182122148 }, { "document": "wikitext2-validation-0041", "nll": 2740.0431518554688, "tokens": 546, "bytes": 1907, "bits_per_byte": 2.07291382643086 }, { "document": "wikitext2-validation-0042", "nll": 2608.16552734375, "tokens": 546, "bytes": 2009, "bits_per_byte": 1.8729653917951103 }, { "document": "wikitext2-validation-0043", "nll": 2572.489776611328, "tokens": 546, "bytes": 2026, "bits_per_byte": 1.831845134972715 }, { "document": "wikitext2-validation-0044", "nll": 3016.1297302246094, "tokens": 546, "bytes": 1823, "bits_per_byte": 2.3869201341046686 }, { "document": "wikitext2-validation-0045", "nll": 2858.697052001953, "tokens": 546, "bytes": 1929, "bits_per_byte": 2.13801350976004 }, { "document": "wikitext2-validation-0046", "nll": 2606.5748901367188, "tokens": 546, "bytes": 2289, "bits_per_byte": 1.6428539395831974 }, { "document": "wikitext2-validation-0047", "nll": 2685.7316284179688, "tokens": 546, "bytes": 2075, "bits_per_byte": 1.8673213019167443 }, { "document": "wikitext2-validation-0048", "nll": 2841.2259826660156, "tokens": 546, "bytes": 1923, "bits_per_byte": 2.1315770334046453 }, { "document": "wikitext2-validation-0049", "nll": 2403.561737060547, "tokens": 546, "bytes": 1975, "bits_per_byte": 1.7557501764697283 }, { "document": "wikitext2-validation-0050", "nll": 2600.902374267578, "tokens": 546, "bytes": 2169, "bits_per_byte": 1.7299718567045486 }, { "document": "wikitext2-validation-0051", "nll": 2846.621307373047, "tokens": 546, "bytes": 1894, "bits_per_byte": 2.168324415752879 }, { "document": "wikitext2-validation-0052", "nll": 2873.136016845703, "tokens": 546, "bytes": 2000, "bits_per_byte": 2.0725295416513827 }, { "document": "wikitext2-validation-0053", "nll": 2914.9937133789062, "tokens": 546, "bytes": 1981, "bits_per_byte": 2.122890951294423 }, { "document": "wikitext2-validation-0054", "nll": 2694.8048095703125, "tokens": 546, "bytes": 2004, "bits_per_byte": 1.9400107459734617 }, { "document": "wikitext2-validation-0055", "nll": 2989.4195556640625, "tokens": 546, "bytes": 1841, "bits_per_byte": 2.3426511505122387 }, { "document": "wikitext2-validation-0056", "nll": 2633.774658203125, "tokens": 546, "bytes": 1944, "bits_per_byte": 1.9545954929056961 }, { "document": "wikitext2-validation-0057", "nll": 2839.046600341797, "tokens": 546, "bytes": 1740, "bits_per_byte": 2.3539531328538974 }, { "document": "wikitext2-validation-0058", "nll": 2584.8514709472656, "tokens": 546, "bytes": 1901, "bits_per_byte": 1.9616793259180234 }, { "document": "wikitext2-validation-0059", "nll": 2816.7892150878906, "tokens": 546, "bytes": 1831, "bits_per_byte": 2.219425358731194 } ], "seconds": 16.940452500013635, "tokenizer_sha256": "ef7f54d89966a08485006e0f64eaf2b0b99217b65ad79f96979f12adbd9ca49d", "subset": "fixed per-article token prefixes", "protocol": "Reset per article; carry native state within article; score IDs >=2; normalize by exact decoded UTF-8 byte lengths; boundary tokens excluded" } } ], "response": { "stimulus": "A small bird lives in a garden. Each morning it leaves the tree and returns to the same branch. The gardener watches from a window. Later, rain falls on the leaves and the bird shelters beneath them. When the sky clears, it crosses the garden again.", "tokens": 77, "records": [ { "position": 0, "token": 0, "piece": "", "fast": { "initial_mean_absolute": 0.023114493116736412, "trained_mean_absolute": 0.12406399846076965, "paired_rms_distance": 0.16715337336063385 }, "slow": { "initial_mean_absolute": 0.0008488452294841409, "trained_mean_absolute": 0.005012591369450092, "paired_rms_distance": 0.008295652456581593 } }, { "position": 1, "token": 34, "piece": "A", "fast": { "initial_mean_absolute": 0.024180753156542778, "trained_mean_absolute": 0.10710226744413376, "paired_rms_distance": 0.13976050913333893 }, "slow": { "initial_mean_absolute": 0.0014030586462467909, "trained_mean_absolute": 0.007566943299025297, "paired_rms_distance": 0.01224948838353157 } }, { "position": 2, "token": 1183, "piece": " small", "fast": { "initial_mean_absolute": 0.025006456300616264, "trained_mean_absolute": 0.10264086723327637, "paired_rms_distance": 0.13461017608642578 }, "slow": { "initial_mean_absolute": 0.0018371419282630086, "trained_mean_absolute": 0.009649325162172318, "paired_rms_distance": 0.01467194315046072 } }, { "position": 3, "token": 2154, "piece": " bird", "fast": { "initial_mean_absolute": 0.024417364969849586, "trained_mean_absolute": 0.12249694764614105, "paired_rms_distance": 0.15798237919807434 }, "slow": { "initial_mean_absolute": 0.002205837285146117, "trained_mean_absolute": 0.012434160336852074, "paired_rms_distance": 0.019363639876246452 } }, { "position": 4, "token": 314, "piece": " l", "fast": { "initial_mean_absolute": 0.025101836770772934, "trained_mean_absolute": 0.10493816435337067, "paired_rms_distance": 0.13537096977233887 }, "slow": { "initial_mean_absolute": 0.002462224569171667, "trained_mean_absolute": 0.012538920156657696, "paired_rms_distance": 0.01886901818215847 } }, { "position": 5, "token": 1263, "piece": "ives", "fast": { "initial_mean_absolute": 0.02633177675306797, "trained_mean_absolute": 0.14378900825977325, "paired_rms_distance": 0.20670075714588165 }, "slow": { "initial_mean_absolute": 0.0026766061782836914, "trained_mean_absolute": 0.016581077128648758, "paired_rms_distance": 0.02732216939330101 } }, { "position": 6, "token": 285, "piece": " in", "fast": { "initial_mean_absolute": 0.026247425004839897, "trained_mean_absolute": 0.1347522735595703, "paired_rms_distance": 0.20231463015079498 }, "slow": { "initial_mean_absolute": 0.0029052195604890585, "trained_mean_absolute": 0.0180592592805624, "paired_rms_distance": 0.032972484827041626 } }, { "position": 7, "token": 260, "piece": " a", "fast": { "initial_mean_absolute": 0.028729788959026337, "trained_mean_absolute": 0.14817650616168976, "paired_rms_distance": 0.21444252133369446 }, "slow": { "initial_mean_absolute": 0.0032526892609894276, "trained_mean_absolute": 0.01996541954576969, "paired_rms_distance": 0.04033540561795235 } }, { "position": 8, "token": 331, "piece": " g", "fast": { "initial_mean_absolute": 0.027257947251200676, "trained_mean_absolute": 0.13183803856372833, "paired_rms_distance": 0.18123741447925568 }, "slow": { "initial_mean_absolute": 0.0034839690197259188, "trained_mean_absolute": 0.02152818813920021, "paired_rms_distance": 0.04441894590854645 } }, { "position": 9, "token": 428, "piece": "ard", "fast": { "initial_mean_absolute": 0.031105462461709976, "trained_mean_absolute": 0.1281193196773529, "paired_rms_distance": 0.1870080679655075 }, "slow": { "initial_mean_absolute": 0.0038337474688887596, "trained_mean_absolute": 0.023998964577913284, "paired_rms_distance": 0.05031291022896767 } }, { "position": 10, "token": 277, "piece": "en", "fast": { "initial_mean_absolute": 0.02796049416065216, "trained_mean_absolute": 0.11521711200475693, "paired_rms_distance": 0.1695956438779831 }, "slow": { "initial_mean_absolute": 0.003968462813645601, "trained_mean_absolute": 0.0258084274828434, "paired_rms_distance": 0.0540631003677845 } }, { "position": 11, "token": 15, "piece": ".", "fast": { "initial_mean_absolute": 0.02803495153784752, "trained_mean_absolute": 0.14442116022109985, "paired_rms_distance": 0.19719548523426056 }, "slow": { "initial_mean_absolute": 0.0039836266078054905, "trained_mean_absolute": 0.02824549563229084, "paired_rms_distance": 0.05951095372438431 } }, { "position": 12, "token": 402, "piece": " E", "fast": { "initial_mean_absolute": 0.02781280316412449, "trained_mean_absolute": 0.1452343612909317, "paired_rms_distance": 0.19767366349697113 }, "slow": { "initial_mean_absolute": 0.004227504134178162, "trained_mean_absolute": 0.0297610592097044, "paired_rms_distance": 0.06323803216218948 } }, { "position": 13, "token": 596, "piece": "ach", "fast": { "initial_mean_absolute": 0.024144500494003296, "trained_mean_absolute": 0.13367411494255066, "paired_rms_distance": 0.19519980251789093 }, "slow": { "initial_mean_absolute": 0.004301833920180798, "trained_mean_absolute": 0.032039593905210495, "paired_rms_distance": 0.06801848113536835 } }, { "position": 14, "token": 3937, "piece": " morning", "fast": { "initial_mean_absolute": 0.02867206372320652, "trained_mean_absolute": 0.1408279687166214, "paired_rms_distance": 0.2052178680896759 }, "slow": { "initial_mean_absolute": 0.004429489839822054, "trained_mean_absolute": 0.034493010491132736, "paired_rms_distance": 0.07262562215328217 } }, { "position": 15, "token": 386, "piece": " it", "fast": { "initial_mean_absolute": 0.03273334726691246, "trained_mean_absolute": 0.1287066638469696, "paired_rms_distance": 0.18415942788124084 }, "slow": { "initial_mean_absolute": 0.004711623303592205, "trained_mean_absolute": 0.036515720188617706, "paired_rms_distance": 0.07701054215431213 } }, { "position": 16, "token": 473, "piece": " le", "fast": { "initial_mean_absolute": 0.02877967432141304, "trained_mean_absolute": 0.12119467556476593, "paired_rms_distance": 0.15838809311389923 }, "slow": { "initial_mean_absolute": 0.004919454921036959, "trained_mean_absolute": 0.03672357276082039, "paired_rms_distance": 0.07627841830253601 } }, { "position": 17, "token": 2387, "piece": "aves", "fast": { "initial_mean_absolute": 0.025121040642261505, "trained_mean_absolute": 0.13322776556015015, "paired_rms_distance": 0.18789373338222504 }, "slow": { "initial_mean_absolute": 0.004954013042151928, "trained_mean_absolute": 0.03840704262256622, "paired_rms_distance": 0.07964204251766205 } }, { "position": 18, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.027274256572127342, "trained_mean_absolute": 0.1168430745601654, "paired_rms_distance": 0.16540592908859253 }, "slow": { "initial_mean_absolute": 0.005077056121081114, "trained_mean_absolute": 0.039328429847955704, "paired_rms_distance": 0.08135201781988144 } }, { "position": 19, "token": 258, "piece": " t", "fast": { "initial_mean_absolute": 0.02886609174311161, "trained_mean_absolute": 0.12510940432548523, "paired_rms_distance": 0.16544568538665771 }, "slow": { "initial_mean_absolute": 0.00522852735593915, "trained_mean_absolute": 0.039564382284879684, "paired_rms_distance": 0.0815202072262764 } }, { "position": 20, "token": 577, "piece": "ree", "fast": { "initial_mean_absolute": 0.027089867740869522, "trained_mean_absolute": 0.12454041093587875, "paired_rms_distance": 0.17540623247623444 }, "slow": { "initial_mean_absolute": 0.005402863956987858, "trained_mean_absolute": 0.040771059691905975, "paired_rms_distance": 0.08381255716085434 } }, { "position": 21, "token": 289, "piece": " and", "fast": { "initial_mean_absolute": 0.029798639938235283, "trained_mean_absolute": 0.12311931699514389, "paired_rms_distance": 0.17720337212085724 }, "slow": { "initial_mean_absolute": 0.005496256519109011, "trained_mean_absolute": 0.04271901026368141, "paired_rms_distance": 0.08730266988277435 } }, { "position": 22, "token": 1127, "piece": " return", "fast": { "initial_mean_absolute": 0.027503635734319687, "trained_mean_absolute": 0.13944201171398163, "paired_rms_distance": 0.18999338150024414 }, "slow": { "initial_mean_absolute": 0.005527321714907885, "trained_mean_absolute": 0.044639647006988525, "paired_rms_distance": 0.08979863673448563 } }, { "position": 23, "token": 84, "piece": "s", "fast": { "initial_mean_absolute": 0.029557982459664345, "trained_mean_absolute": 0.10869818925857544, "paired_rms_distance": 0.1510254442691803 }, "slow": { "initial_mean_absolute": 0.0056805857457220554, "trained_mean_absolute": 0.04548804461956024, "paired_rms_distance": 0.09051650762557983 } }, { "position": 24, "token": 291, "piece": " to", "fast": { "initial_mean_absolute": 0.029552942141890526, "trained_mean_absolute": 0.15557041764259338, "paired_rms_distance": 0.21760223805904388 }, "slow": { "initial_mean_absolute": 0.005905340425670147, "trained_mean_absolute": 0.047561489045619965, "paired_rms_distance": 0.09412117302417755 } }, { "position": 25, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.02823040448129177, "trained_mean_absolute": 0.15411585569381714, "paired_rms_distance": 0.21816392242908478 }, "slow": { "initial_mean_absolute": 0.00624646944925189, "trained_mean_absolute": 0.050036538392305374, "paired_rms_distance": 0.0983976200222969 } }, { "position": 26, "token": 1113, "piece": " same", "fast": { "initial_mean_absolute": 0.027242284268140793, "trained_mean_absolute": 0.16077390313148499, "paired_rms_distance": 0.22651508450508118 }, "slow": { "initial_mean_absolute": 0.006167738698422909, "trained_mean_absolute": 0.052422165870666504, "paired_rms_distance": 0.10297001153230667 } }, { "position": 27, "token": 283, "piece": " b", "fast": { "initial_mean_absolute": 0.024250993505120277, "trained_mean_absolute": 0.1387457251548767, "paired_rms_distance": 0.18996931612491608 }, "slow": { "initial_mean_absolute": 0.00618379469960928, "trained_mean_absolute": 0.05278445780277252, "paired_rms_distance": 0.1030735969543457 } }, { "position": 28, "token": 3106, "piece": "ranch", "fast": { "initial_mean_absolute": 0.02722608670592308, "trained_mean_absolute": 0.1561286449432373, "paired_rms_distance": 0.21533454954624176 }, "slow": { "initial_mean_absolute": 0.0062729595229029655, "trained_mean_absolute": 0.053279586136341095, "paired_rms_distance": 0.10343468189239502 } }, { "position": 29, "token": 15, "piece": ".", "fast": { "initial_mean_absolute": 0.027204250916838646, "trained_mean_absolute": 0.1451653689146042, "paired_rms_distance": 0.19409975409507751 }, "slow": { "initial_mean_absolute": 0.006366088520735502, "trained_mean_absolute": 0.05452138930559158, "paired_rms_distance": 0.1057380959391594 } }, { "position": 30, "token": 325, "piece": " The", "fast": { "initial_mean_absolute": 0.028056863695383072, "trained_mean_absolute": 0.14834435284137726, "paired_rms_distance": 0.20292389392852783 }, "slow": { "initial_mean_absolute": 0.006537768058478832, "trained_mean_absolute": 0.05629131942987442, "paired_rms_distance": 0.10890255123376846 } }, { "position": 31, "token": 331, "piece": " g", "fast": { "initial_mean_absolute": 0.02673790603876114, "trained_mean_absolute": 0.13212458789348602, "paired_rms_distance": 0.17825007438659668 }, "slow": { "initial_mean_absolute": 0.006677511613816023, "trained_mean_absolute": 0.056813955307006836, "paired_rms_distance": 0.10946960002183914 } }, { "position": 32, "token": 428, "piece": "ard", "fast": { "initial_mean_absolute": 0.03064233437180519, "trained_mean_absolute": 0.12694622576236725, "paired_rms_distance": 0.18031705915927887 }, "slow": { "initial_mean_absolute": 0.006992768496274948, "trained_mean_absolute": 0.058065399527549744, "paired_rms_distance": 0.11123945564031601 } }, { "position": 33, "token": 821, "piece": "ener", "fast": { "initial_mean_absolute": 0.03312137350440025, "trained_mean_absolute": 0.15233616530895233, "paired_rms_distance": 0.21351048350334167 }, "slow": { "initial_mean_absolute": 0.007271833252161741, "trained_mean_absolute": 0.059117257595062256, "paired_rms_distance": 0.11368755251169205 } }, { "position": 34, "token": 2001, "piece": " wat", "fast": { "initial_mean_absolute": 0.030544588342308998, "trained_mean_absolute": 0.13562050461769104, "paired_rms_distance": 0.1779838502407074 }, "slow": { "initial_mean_absolute": 0.007504153531044722, "trained_mean_absolute": 0.05888989195227623, "paired_rms_distance": 0.11196986585855484 } }, { "position": 35, "token": 1662, "piece": "ches", "fast": { "initial_mean_absolute": 0.030274786055088043, "trained_mean_absolute": 0.16951271891593933, "paired_rms_distance": 0.24107517302036285 }, "slow": { "initial_mean_absolute": 0.007578900083899498, "trained_mean_absolute": 0.0608212985098362, "paired_rms_distance": 0.11596748977899551 } }, { "position": 36, "token": 404, "piece": " from", "fast": { "initial_mean_absolute": 0.03183096647262573, "trained_mean_absolute": 0.14618992805480957, "paired_rms_distance": 0.22571925818920135 }, "slow": { "initial_mean_absolute": 0.0076020811684429646, "trained_mean_absolute": 0.06229180842638016, "paired_rms_distance": 0.11866322159767151 } }, { "position": 37, "token": 260, "piece": " a", "fast": { "initial_mean_absolute": 0.0308818556368351, "trained_mean_absolute": 0.15759137272834778, "paired_rms_distance": 0.22696435451507568 }, "slow": { "initial_mean_absolute": 0.007771452888846397, "trained_mean_absolute": 0.06372097879648209, "paired_rms_distance": 0.12167297303676605 } }, { "position": 38, "token": 1642, "piece": " wind", "fast": { "initial_mean_absolute": 0.029783004894852638, "trained_mean_absolute": 0.16880807280540466, "paired_rms_distance": 0.231013223528862 }, "slow": { "initial_mean_absolute": 0.00784686952829361, "trained_mean_absolute": 0.06558546423912048, "paired_rms_distance": 0.12464433908462524 } }, { "position": 39, "token": 348, "piece": "ow", "fast": { "initial_mean_absolute": 0.02791045419871807, "trained_mean_absolute": 0.1332325041294098, "paired_rms_distance": 0.19176779687404633 }, "slow": { "initial_mean_absolute": 0.007913431152701378, "trained_mean_absolute": 0.06638959795236588, "paired_rms_distance": 0.1258428394794464 } }, { "position": 40, "token": 15, "piece": ".", "fast": { "initial_mean_absolute": 0.030796803534030914, "trained_mean_absolute": 0.14819614589214325, "paired_rms_distance": 0.20383211970329285 }, "slow": { "initial_mean_absolute": 0.008087879046797752, "trained_mean_absolute": 0.06754420697689056, "paired_rms_distance": 0.1274624913930893 } }, { "position": 41, "token": 2849, "piece": " Later", "fast": { "initial_mean_absolute": 0.029594531282782555, "trained_mean_absolute": 0.15101368725299835, "paired_rms_distance": 0.21380238234996796 }, "slow": { "initial_mean_absolute": 0.008166651241481304, "trained_mean_absolute": 0.06847678869962692, "paired_rms_distance": 0.12926085293293 } }, { "position": 42, "token": 13, "piece": ",", "fast": { "initial_mean_absolute": 0.028162745758891106, "trained_mean_absolute": 0.15002107620239258, "paired_rms_distance": 0.2069675624370575 }, "slow": { "initial_mean_absolute": 0.008166944608092308, "trained_mean_absolute": 0.06798283755779266, "paired_rms_distance": 0.12715710699558258 } }, { "position": 43, "token": 3027, "piece": " rain", "fast": { "initial_mean_absolute": 0.028395934030413628, "trained_mean_absolute": 0.16000355780124664, "paired_rms_distance": 0.21191717684268951 }, "slow": { "initial_mean_absolute": 0.008215800859034061, "trained_mean_absolute": 0.06742538511753082, "paired_rms_distance": 0.12475398927927017 } }, { "position": 44, "token": 2430, "piece": " fall", "fast": { "initial_mean_absolute": 0.02654573693871498, "trained_mean_absolute": 0.1298060417175293, "paired_rms_distance": 0.16520152986049652 }, "slow": { "initial_mean_absolute": 0.008186656050384045, "trained_mean_absolute": 0.06698735058307648, "paired_rms_distance": 0.12215186655521393 } }, { "position": 45, "token": 84, "piece": "s", "fast": { "initial_mean_absolute": 0.029752377420663834, "trained_mean_absolute": 0.10644800961017609, "paired_rms_distance": 0.1443432867527008 }, "slow": { "initial_mean_absolute": 0.008104382082819939, "trained_mean_absolute": 0.06647462397813797, "paired_rms_distance": 0.12048110365867615 } }, { "position": 46, "token": 327, "piece": " on", "fast": { "initial_mean_absolute": 0.027463313192129135, "trained_mean_absolute": 0.1510087251663208, "paired_rms_distance": 0.2131786346435547 }, "slow": { "initial_mean_absolute": 0.008074909448623657, "trained_mean_absolute": 0.06702855229377747, "paired_rms_distance": 0.1213284432888031 } }, { "position": 47, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.026203792542219162, "trained_mean_absolute": 0.1500207781791687, "paired_rms_distance": 0.2159687727689743 }, "slow": { "initial_mean_absolute": 0.008219931274652481, "trained_mean_absolute": 0.06854475289583206, "paired_rms_distance": 0.12412944436073303 } }, { "position": 48, "token": 473, "piece": " le", "fast": { "initial_mean_absolute": 0.03043615072965622, "trained_mean_absolute": 0.13656407594680786, "paired_rms_distance": 0.18984653055667877 }, "slow": { "initial_mean_absolute": 0.0084832226857543, "trained_mean_absolute": 0.06863681972026825, "paired_rms_distance": 0.1245230883359909 } }, { "position": 49, "token": 2387, "piece": "aves", "fast": { "initial_mean_absolute": 0.026669081300497055, "trained_mean_absolute": 0.143196702003479, "paired_rms_distance": 0.20741119980812073 }, "slow": { "initial_mean_absolute": 0.008566980250179768, "trained_mean_absolute": 0.06964829564094543, "paired_rms_distance": 0.1264296919107437 } }, { "position": 50, "token": 289, "piece": " and", "fast": { "initial_mean_absolute": 0.03215816617012024, "trained_mean_absolute": 0.12867392599582672, "paired_rms_distance": 0.18555974960327148 }, "slow": { "initial_mean_absolute": 0.008778284303843975, "trained_mean_absolute": 0.0703163668513298, "paired_rms_distance": 0.12771163880825043 } }, { "position": 51, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.03179185092449188, "trained_mean_absolute": 0.14317777752876282, "paired_rms_distance": 0.20570501685142517 }, "slow": { "initial_mean_absolute": 0.00916118361055851, "trained_mean_absolute": 0.07148713618516922, "paired_rms_distance": 0.12956561148166656 } }, { "position": 52, "token": 2154, "piece": " bird", "fast": { "initial_mean_absolute": 0.029559368267655373, "trained_mean_absolute": 0.13550648093223572, "paired_rms_distance": 0.19805817306041718 }, "slow": { "initial_mean_absolute": 0.009400616399943829, "trained_mean_absolute": 0.0725371390581131, "paired_rms_distance": 0.13175088167190552 } }, { "position": 53, "token": 662, "piece": " she", "fast": { "initial_mean_absolute": 0.032290440052747726, "trained_mean_absolute": 0.14776763319969177, "paired_rms_distance": 0.20502077043056488 }, "slow": { "initial_mean_absolute": 0.009581880643963814, "trained_mean_absolute": 0.07317475974559784, "paired_rms_distance": 0.1328267753124237 } }, { "position": 54, "token": 77, "piece": "l", "fast": { "initial_mean_absolute": 0.029751259833574295, "trained_mean_absolute": 0.12007185816764832, "paired_rms_distance": 0.16183431446552277 }, "slow": { "initial_mean_absolute": 0.00958943460136652, "trained_mean_absolute": 0.07283541560173035, "paired_rms_distance": 0.13179956376552582 } }, { "position": 55, "token": 754, "piece": "ters", "fast": { "initial_mean_absolute": 0.031431764364242554, "trained_mean_absolute": 0.1464737504720688, "paired_rms_distance": 0.20577691495418549 }, "slow": { "initial_mean_absolute": 0.009603864513337612, "trained_mean_absolute": 0.07306463271379471, "paired_rms_distance": 0.13302038609981537 } }, { "position": 56, "token": 283, "piece": " b", "fast": { "initial_mean_absolute": 0.024017641320824623, "trained_mean_absolute": 0.1280021369457245, "paired_rms_distance": 0.1730339080095291 }, "slow": { "initial_mean_absolute": 0.009624085389077663, "trained_mean_absolute": 0.07201749086380005, "paired_rms_distance": 0.13072480261325836 } }, { "position": 57, "token": 1437, "piece": "ene", "fast": { "initial_mean_absolute": 0.027066435664892197, "trained_mean_absolute": 0.1252766251564026, "paired_rms_distance": 0.1775621920824051 }, "slow": { "initial_mean_absolute": 0.00965234637260437, "trained_mean_absolute": 0.07183799147605896, "paired_rms_distance": 0.1307988315820694 } }, { "position": 58, "token": 697, "piece": "ath", "fast": { "initial_mean_absolute": 0.0278007872402668, "trained_mean_absolute": 0.1262674331665039, "paired_rms_distance": 0.18489304184913635 }, "slow": { "initial_mean_absolute": 0.009657789021730423, "trained_mean_absolute": 0.07212211191654205, "paired_rms_distance": 0.13157828152179718 } }, { "position": 59, "token": 706, "piece": " them", "fast": { "initial_mean_absolute": 0.029673617333173752, "trained_mean_absolute": 0.12635643780231476, "paired_rms_distance": 0.18348610401153564 }, "slow": { "initial_mean_absolute": 0.009618701413273811, "trained_mean_absolute": 0.07258550822734833, "paired_rms_distance": 0.1323103904724121 } }, { "position": 60, "token": 15, "piece": ".", "fast": { "initial_mean_absolute": 0.031119320541620255, "trained_mean_absolute": 0.14282184839248657, "paired_rms_distance": 0.19179528951644897 }, "slow": { "initial_mean_absolute": 0.009714766405522823, "trained_mean_absolute": 0.07315576076507568, "paired_rms_distance": 0.13302530348300934 } }, { "position": 61, "token": 1502, "piece": " When", "fast": { "initial_mean_absolute": 0.02876034751534462, "trained_mean_absolute": 0.13813428580760956, "paired_rms_distance": 0.19722123444080353 }, "slow": { "initial_mean_absolute": 0.009730516001582146, "trained_mean_absolute": 0.07376134395599365, "paired_rms_distance": 0.13407468795776367 } }, { "position": 62, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.027878623455762863, "trained_mean_absolute": 0.13501963019371033, "paired_rms_distance": 0.19697578251361847 }, "slow": { "initial_mean_absolute": 0.009917857125401497, "trained_mean_absolute": 0.07474692165851593, "paired_rms_distance": 0.13576403260231018 } }, { "position": 63, "token": 1840, "piece": " sk", "fast": { "initial_mean_absolute": 0.027520447969436646, "trained_mean_absolute": 0.13848668336868286, "paired_rms_distance": 0.1888769119977951 }, "slow": { "initial_mean_absolute": 0.01000820379704237, "trained_mean_absolute": 0.07416357100009918, "paired_rms_distance": 0.13462653756141663 } }, { "position": 64, "token": 90, "piece": "y", "fast": { "initial_mean_absolute": 0.027063563466072083, "trained_mean_absolute": 0.11908626556396484, "paired_rms_distance": 0.1659853756427765 }, "slow": { "initial_mean_absolute": 0.010034856386482716, "trained_mean_absolute": 0.07394822686910629, "paired_rms_distance": 0.13418905436992645 } }, { "position": 65, "token": 3402, "piece": " cle", "fast": { "initial_mean_absolute": 0.02959086000919342, "trained_mean_absolute": 0.13687482476234436, "paired_rms_distance": 0.17415569722652435 }, "slow": { "initial_mean_absolute": 0.010100890882313251, "trained_mean_absolute": 0.07323292642831802, "paired_rms_distance": 0.13100746273994446 } }, { "position": 66, "token": 883, "piece": "ars", "fast": { "initial_mean_absolute": 0.029972054064273834, "trained_mean_absolute": 0.13368599116802216, "paired_rms_distance": 0.1783643215894699 }, "slow": { "initial_mean_absolute": 0.010212041437625885, "trained_mean_absolute": 0.07320421934127808, "paired_rms_distance": 0.13044047355651855 } }, { "position": 67, "token": 13, "piece": ",", "fast": { "initial_mean_absolute": 0.030445292592048645, "trained_mean_absolute": 0.13469363749027252, "paired_rms_distance": 0.1775035262107849 }, "slow": { "initial_mean_absolute": 0.010405568405985832, "trained_mean_absolute": 0.07182541489601135, "paired_rms_distance": 0.1266971230506897 } }, { "position": 68, "token": 386, "piece": " it", "fast": { "initial_mean_absolute": 0.033329714089632034, "trained_mean_absolute": 0.11810867488384247, "paired_rms_distance": 0.1578831970691681 }, "slow": { "initial_mean_absolute": 0.010643625631928444, "trained_mean_absolute": 0.07168950140476227, "paired_rms_distance": 0.12584340572357178 } }, { "position": 69, "token": 2201, "piece": " cross", "fast": { "initial_mean_absolute": 0.0327671617269516, "trained_mean_absolute": 0.11699528992176056, "paired_rms_distance": 0.1566292941570282 }, "slow": { "initial_mean_absolute": 0.01090650912374258, "trained_mean_absolute": 0.0710221379995346, "paired_rms_distance": 0.12367373704910278 } }, { "position": 70, "token": 280, "piece": "es", "fast": { "initial_mean_absolute": 0.030891625210642815, "trained_mean_absolute": 0.11489203572273254, "paired_rms_distance": 0.1594458967447281 }, "slow": { "initial_mean_absolute": 0.011096288450062275, "trained_mean_absolute": 0.07115552574396133, "paired_rms_distance": 0.12375681102275848 } }, { "position": 71, "token": 262, "piece": " the", "fast": { "initial_mean_absolute": 0.03171834349632263, "trained_mean_absolute": 0.13222067058086395, "paired_rms_distance": 0.1860329955816269 }, "slow": { "initial_mean_absolute": 0.011441567912697792, "trained_mean_absolute": 0.07207001000642776, "paired_rms_distance": 0.12480498105287552 } }, { "position": 72, "token": 331, "piece": " g", "fast": { "initial_mean_absolute": 0.027393635362386703, "trained_mean_absolute": 0.12624973058700562, "paired_rms_distance": 0.17261992394924164 }, "slow": { "initial_mean_absolute": 0.011437773704528809, "trained_mean_absolute": 0.07173813879489899, "paired_rms_distance": 0.12421131134033203 } }, { "position": 73, "token": 428, "piece": "ard", "fast": { "initial_mean_absolute": 0.03402942419052124, "trained_mean_absolute": 0.1260901838541031, "paired_rms_distance": 0.1751817911863327 }, "slow": { "initial_mean_absolute": 0.01160092931240797, "trained_mean_absolute": 0.0722440630197525, "paired_rms_distance": 0.12497936934232712 } }, { "position": 74, "token": 277, "piece": "en", "fast": { "initial_mean_absolute": 0.03227340430021286, "trained_mean_absolute": 0.11733895540237427, "paired_rms_distance": 0.16304965317249298 }, "slow": { "initial_mean_absolute": 0.011754150502383709, "trained_mean_absolute": 0.07230909168720245, "paired_rms_distance": 0.12500280141830444 } }, { "position": 75, "token": 759, "piece": " again", "fast": { "initial_mean_absolute": 0.032604020088911057, "trained_mean_absolute": 0.12062519043684006, "paired_rms_distance": 0.17651718854904175 }, "slow": { "initial_mean_absolute": 0.011893972754478455, "trained_mean_absolute": 0.07291946560144424, "paired_rms_distance": 0.12616148591041565 } }, { "position": 76, "token": 15, "piece": ".", "fast": { "initial_mean_absolute": 0.03278205171227455, "trained_mean_absolute": 0.1389051377773285, "paired_rms_distance": 0.18638983368873596 }, "slow": { "initial_mean_absolute": 0.012158679775893688, "trained_mean_absolute": 0.07353381812572479, "paired_rms_distance": 0.12708376348018646 } } ], "interpretation": "Identical teacher-forced input from zero state. State distance is not a measure of semantic knowledge or animal behavior." }, "parameter_changes": [ { "name": "edge_log_gain", "parameters": 76130, "initial_l2": 0.0, "trained_l2": 68.94664764404297, "change_l2": 68.94664764404297 }, { "name": "alpha_logit", "parameters": 1024, "initial_l2": 27.73988914489746, "trained_l2": 41.384220123291016, "change_l2": 19.07520866394043 }, { "name": "beta_logit", "parameters": 1024, "initial_l2": 42.34975051879883, "trained_l2": 38.42070770263672, "change_l2": 10.632875442504883 }, { "name": "recurrent_logit", "parameters": 1, "initial_l2": 0.0, "trained_l2": 0.21551662683486938, "change_l2": 0.21551662683486938 }, { "name": "output_bias", "parameters": 4096, "initial_l2": 0.0, "trained_l2": 6.07493257522583, "change_l2": 6.07493257522583 }, { "name": "embedding.weight", "parameters": 393216, "initial_l2": 125.24240112304688, "trained_l2": 137.31344604492188, "change_l2": 125.97926330566406 }, { "name": "input.weight", "parameters": 98304, "initial_l2": 9.26546859741211, "trained_l2": 30.794281005859375, "change_l2": 29.105073928833008 }, { "name": "input.bias", "parameters": 1024, "initial_l2": 0.0, "trained_l2": 1.5154465436935425, "change_l2": 1.5154465436935425 }, { "name": "norm.weight", "parameters": 256, "initial_l2": 16.0, "trained_l2": 12.40145492553711, "change_l2": 4.7852888107299805 }, { "name": "norm.bias", "parameters": 256, "initial_l2": 0.0, "trained_l2": 3.151198148727417, "change_l2": 3.151198148727417 }, { "name": "readout.weight", "parameters": 24576, "initial_l2": 3.927276611328125, "trained_l2": 12.471360206604004, "change_l2": 11.4475679397583 }, { "name": "readout.bias", "parameters": 96, "initial_l2": 0.0, "trained_l2": 0.9496904611587524, "change_l2": 0.9496904611587524 } ], "protocol": "docs/MECHANISM-PROTOCOL.md", "caution": "Exploratory validation diagnostics after training. Acute removal without retraining is not a matched architectural ablation or a topology advantage test. No held-out test data is used." }