Upload add_sub_sorl_v1_abs70_100K

add_sub_sorl_v1_abs70_100K/config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "SorlModelWrapper"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 510,
+  "initializer_range": 0.02,
+  "intermediate_size": 2040,
+  "layer_types": [
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 3,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151714
+}

add_sub_sorl_v1_abs70_100K/generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "transformers_version": "5.5.0",
+  "use_cache": true
+}

add_sub_sorl_v1_abs70_100K/metrics.json

@@ -0,0 +1,3537 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      800,
+      850,
+      900,
+      950,
+      1000,
+      1050,
+      1100,
+      1150,
+      1200,
+      1250,
+      1300,
+      1350,
+      1400,
+      1450,
+      1500,
+      1550,
+      1613,
+      1663,
+      1713,
+      1763,
+      1813,
+      1863,
+      1913,
+      1963,
+      2013,
+      2063,
+      2113,
+      2163,
+      2213,
+      2263,
+      2313,
+      2363,
+      2413,
+      2463,
+      2513,
+      2563,
+      2613,
+      2663,
+      2713,
+      2763,
+      2813,
+      2863,
+      2913,
+      2963,
+      3013,
+      3063,
+      3113,
+      3176,
+      3226,
+      3276,
+      3326,
+      3376,
+      3426,
+      3476,
+      3526,
+      3576,
+      3626,
+      3676,
+      3726,
+      3776,
+      3826,
+      3876,
+      3926,
+      3976,
+      4026,
+      4076,
+      4126,
+      4176,
+      4226,
+      4276,
+      4326,
+      4376,
+      4426,
+      4476,
+      4526,
+      4576,
+      4626,
+      4676,
+      4739,
+      4789,
+      4839,
+      4889,
+      4939,
+      4989,
+      5039,
+      5089,
+      5139,
+      5189,
+      5239,
+      5289,
+      5339,
+      5389,
+      5439,
+      5489,
+      5539,
+      5589,
+      5639,
+      5689,
+      5739,
+      5789,
+      5839,
+      5889,
+      5939,
+      5989,
+      6039,
+      6089,
+      6139,
+      6189,
+      6239,
+      6302,
+      6352,
+      6402,
+      6452,
+      6502,
+      6552,
+      6602,
+      6652,
+      6702,
+      6752,
+      6802,
+      6852,
+      6902,
+      6952,
+      7002,
+      7052,
+      7102,
+      7152,
+      7202,
+      7252,
+      7302,
+      7352,
+      7402,
+      7452,
+      7502,
+      7552,
+      7602,
+      7652,
+      7702,
+      7752,
+      7802,
+      7865,
+      7915,
+      7965,
+      8015,
+      8065,
+      8115,
+      8165,
+      8215,
+      8265,
+      8315,
+      8365,
+      8415,
+      8465,
+      8515,
+      8565,
+      8615,
+      8665,
+      8715,
+      8765,
+      8815,
+      8865,
+      8915,
+      8965,
+      9015,
+      9065,
+      9115,
+      9165,
+      9215,
+      9265,
+      9315,
+      9365,
+      9428,
+      9478,
+      9528,
+      9578,
+      9628,
+      9678,
+      9728,
+      9778,
+      9828,
+      9878,
+      9928,
+      9978,
+      10028,
+      10078,
+      10128,
+      10178,
+      10228,
+      10278,
+      10328,
+      10378,
+      10428,
+      10478,
+      10528,
+      10578,
+      10628,
+      10678,
+      10728,
+      10778,
+      10828,
+      10878,
+      10928,
+      10991,
+      11041,
+      11091,
+      11141,
+      11191,
+      11241,
+      11291,
+      11341,
+      11391,
+      11441,
+      11491,
+      11541,
+      11591,
+      11641,
+      11691,
+      11741,
+      11791,
+      11841,
+      11891,
+      11941,
+      11991,
+      12041,
+      12091,
+      12141,
+      12191,
+      12241,
+      12291,
+      12341,
+      12391,
+      12441,
+      12491,
+      12554,
+      12604,
+      12654,
+      12704,
+      12754,
+      12804,
+      12854,
+      12904,
+      12954,
+      13004,
+      13054,
+      13104,
+      13154,
+      13204,
+      13254,
+      13304,
+      13354,
+      13404,
+      13454,
+      13504,
+      13554,
+      13604,
+      13654,
+      13704,
+      13754,
+      13804,
+      13854,
+      13904,
+      13954,
+      14004,
+      14054,
+      14117,
+      14167,
+      14217,
+      14267,
+      14317,
+      14367,
+      14417,
+      14467,
+      14517,
+      14567,
+      14617,
+      14667,
+      14717,
+      14767,
+      14817,
+      14867,
+      14917,
+      14967,
+      15017,
+      15067,
+      15117,
+      15167,
+      15217,
+      15267,
+      15317,
+      15367,
+      15417,
+      15467,
+      15517,
+      15567,
+      15617
+    ],
+    "loss": [
+      15.745485305786133,
+      12.111101150512695,
+      8.688098907470703,
+      7.662339210510254,
+      5.153091907501221,
+      3.735525131225586,
+      2.607950448989868,
+      2.3635735511779785,
+      2.269740581512451,
+      1.8536663055419922,
+      2.110579252243042,
+      1.73744535446167,
+      1.4838066101074219,
+      1.1551334857940674,
+      0.8606105446815491,
+      -0.5420317053794861,
+      -1.9786415100097656,
+      -5.19218635559082,
+      -6.521421432495117,
+      -8.504491806030273,
+      -8.68906021118164,
+      -9.980416297912598,
+      -10.984786987304688,
+      -11.12669849395752,
+      -12.1531400680542,
+      -12.336996078491211,
+      -12.98429012298584,
+      -12.650969505310059,
+      -13.504485130310059,
+      -13.880614280700684,
+      -13.229496002197266,
+      -14.47417163848877,
+      -13.695523262023926,
+      -13.4900484085083,
+      -14.075793266296387,
+      -13.270401954650879,
+      -14.080594062805176,
+      -14.277887344360352,
+      -14.198729515075684,
+      -13.764762878417969,
+      -13.542289733886719,
+      -14.601493835449219,
+      -14.072729110717773,
+      -13.897228240966797,
+      -13.582523345947266,
+      -14.607087135314941,
+      -14.05154037475586,
+      -14.359332084655762,
+      -14.434036254882812,
+      -14.00304889678955,
+      -14.451115608215332,
+      -14.104546546936035,
+      -13.94878101348877,
+      -14.221521377563477,
+      -13.815980911254883,
+      -13.736690521240234,
+      -13.158988952636719,
+      -13.692000389099121,
+      -14.006065368652344,
+      -14.327851295471191,
+      -14.33360767364502,
+      -14.00327205657959,
+      -13.431060791015625,
+      -14.111815452575684,
+      -14.123462677001953,
+      -13.069084167480469,
+      -12.345903396606445,
+      -11.153604507446289,
+      -9.801177978515625,
+      -9.01672077178955,
+      -7.479533672332764,
+      -7.182117938995361,
+      -6.600742816925049,
+      -6.449777603149414,
+      -5.038028240203857,
+      -5.620625972747803,
+      -5.1884050369262695,
+      -5.449138164520264,
+      -4.972566604614258,
+      -5.019037246704102,
+      -5.8771257400512695,
+      -4.346200466156006,
+      -5.829677581787109,
+      -3.929460048675537,
+      -4.005841255187988,
+      -3.954021692276001,
+      -3.9247312545776367,
+      -3.7874293327331543,
+      -4.137342929840088,
+      -3.631946086883545,
+      -3.795776128768921,
+      -3.6702277660369873,
+      -3.6395363807678223,
+      -2.9685635566711426,
+      -3.527247667312622,
+      -3.400318145751953,
+      -3.0507147312164307,
+      -2.8359832763671875,
+      -2.8940179347991943,
+      -3.1444623470306396,
+      -2.8155367374420166,
+      -2.8892080783843994,
+      -3.103070020675659,
+      -2.5470175743103027,
+      -2.6847221851348877,
+      -2.423678159713745,
+      -2.263305425643921,
+      -2.0734896659851074,
+      -2.357597589492798,
+      -2.2026429176330566,
+      -2.0292112827301025,
+      -2.156585216522217,
+      -1.9658671617507935,
+      -2.144821882247925,
+      -2.7632017135620117,
+      -2.0000717639923096,
+      -2.1257712841033936,
+      -2.0397377014160156,
+      -2.131657838821411,
+      -2.2364962100982666,
+      -1.737606167793274,
+      -3.416055202484131,
+      -1.4052796363830566,
+      -1.6089859008789062,
+      -2.2010209560394287,
+      -1.9943928718566895,
+      -2.0204389095306396,
+      -1.4130545854568481,
+      -2.4625351428985596,
+      -1.4854553937911987,
+      -1.8527801036834717,
+      -1.5017026662826538,
+      -1.5360091924667358,
+      -1.3850282430648804,
+      -1.5596262216567993,
+      -1.8482146263122559,
+      -1.4947919845581055,
+      -1.800887107849121,
+      -1.3769874572753906,
+      -1.4963473081588745,
+      -1.5607776641845703,
+      -1.3631064891815186,
+      -1.6475448608398438,
+      -1.3457427024841309,
+      -1.5554670095443726,
+      -1.341086983680725,
+      -1.3120797872543335,
+      -1.2237696647644043,
+      -1.2719624042510986,
+      -1.4146528244018555,
+      -1.7828205823898315,
+      -1.7062568664550781,
+      -1.50191330909729,
+      -1.1606098413467407,
+      -1.4483414888381958,
+      -1.3426216840744019,
+      -1.3150863647460938,
+      -1.595462441444397,
+      -2.0584230422973633,
+      -1.2525500059127808,
+      -1.0609592199325562,
+      -1.209620475769043,
+      -1.610905408859253,
+      -1.3973740339279175,
+      -1.1834378242492676,
+      -1.2822691202163696,
+      -1.4483617544174194,
+      -1.851280689239502,
+      -1.3067349195480347,
+      -0.9041375517845154,
+      -1.5270408391952515,
+      -1.0312458276748657,
+      -1.381829023361206,
+      -1.2376768589019775,
+      -1.1838841438293457,
+      -1.2037440538406372,
+      -1.167670488357544,
+      -1.975407361984253,
+      -1.1881239414215088,
+      -1.2769267559051514,
+      -1.51146399974823,
+      -1.393563985824585,
+      -1.1670476198196411,
+      -1.1838637590408325,
+      -1.2354512214660645,
+      -1.330101490020752,
+      -1.0780541896820068,
+      -0.9247320890426636,
+      -1.3360766172409058,
+      -0.9906800985336304,
+      -1.3341649770736694,
+      -0.9367088079452515,
+      -1.0170704126358032,
+      -1.2342417240142822,
+      -0.9409386515617371,
+      -1.1774961948394775,
+      -1.4847350120544434,
+      -0.8936091065406799,
+      -0.8424043655395508,
+      -0.9152222871780396,
+      -1.2678248882293701,
+      -1.1446444988250732,
+      -1.222661018371582,
+      -0.9633662700653076,
+      -1.0385525226593018,
+      -0.9137232899665833,
+      -1.0190773010253906,
+      -0.8557896018028259,
+      -0.8860712051391602,
+      -1.1912087202072144,
+      -1.3398643732070923,
+      -0.7046958208084106,
+      -0.8662760257720947,
+      -0.9078179597854614,
+      -0.9039081931114197,
+      -1.1256135702133179,
+      -1.0441449880599976,
+      -0.9961947202682495,
+      -0.7966858744621277,
+      -0.7069897055625916,
+      -0.9014972448348999,
+      -0.878194272518158,
+      -0.694083034992218,
+      -1.035062313079834,
+      -0.6719993948936462,
+      -0.8185563087463379,
+      -1.0316808223724365,
+      -1.0586570501327515,
+      -0.6989405751228333,
+      -0.9158240556716919,
+      -0.630041241645813,
+      -0.9732574224472046,
+      -0.7749252915382385,
+      -1.0481270551681519,
+      -0.7896981239318848,
+      -0.8524516820907593,
+      -0.8393667936325073,
+      -1.1963506937026978,
+      -0.8570467233657837,
+      -0.7653064131736755,
+      -0.8306721448898315,
+      -0.8403700590133667,
+      -0.8288451433181763,
+      -0.6389251351356506,
+      -0.9450255632400513,
+      -0.9943042397499084,
+      -0.8688981533050537,
+      -0.8509131073951721,
+      -0.7889793515205383,
+      -0.6726685762405396,
+      -0.6617797017097473,
+      -0.5594507455825806,
+      -0.9815265536308289,
+      -0.8038948178291321,
+      -0.6824551820755005,
+      -0.6856938004493713,
+      -0.5018048286437988,
+      -0.6337124109268188,
+      -0.5932860374450684,
+      -0.8555293083190918,
+      -0.9893091320991516,
+      -0.8064727187156677,
+      -0.5684929490089417,
+      -0.7127496004104614,
+      -0.7534486055374146,
+      -0.6665985584259033,
+      -0.56135094165802,
+      -0.6397708654403687,
+      -0.6071538925170898,
+      -0.7273363471031189,
+      -0.5128734707832336,
+      -0.6393722891807556,
+      -0.729474663734436,
+      -0.6338003873825073,
+      -0.6954477429389954,
+      -0.7880577445030212,
+      -0.7830257415771484,
+      -0.601180911064148,
+      -0.6578355431556702,
+      -0.6298449635505676,
+      -0.5262075662612915,
+      -0.593431830406189,
+      -0.5402166843414307,
+      -0.6533572673797607,
+      -0.3860713243484497,
+      -0.784864068031311,
+      -0.7605255842208862,
+      -0.6784576773643494,
+      -0.7199658751487732,
+      -0.8685442805290222,
+      -0.5253223180770874,
+      -0.4623960852622986,
+      -0.5535212755203247,
+      -0.5872893929481506,
+      -0.592187225818634,
+      -0.7782663106918335,
+      -0.6775127053260803,
+      -0.5392772555351257,
+      -0.5953889489173889,
+      -0.47411519289016724,
+      -0.5348408818244934,
+      -0.5214688777923584,
+      -0.381331205368042,
+      -0.4691905081272125,
+      -0.5091303586959839,
+      -0.606124758720398,
+      -0.6473522782325745,
+      -0.5565692186355591,
+      -0.6980366110801697,
+      -0.6461753249168396
+    ],
+    "base_loss": [
+      10.605161666870117,
+      8.040678024291992,
+      7.333432197570801,
+      6.27871561050415,
+      5.253532409667969,
+      3.5272746086120605,
+      2.136018991470337,
+      2.044652223587036,
+      1.9119266271591187,
+      1.8299065828323364,
+      1.836234450340271,
+      1.7979004383087158,
+      1.857666015625,
+      1.8216068744659424,
+      1.7819855213165283,
+      1.85440194606781,
+      1.8007665872573853,
+      2.045675277709961,
+      1.933180809020996,
+      1.9491740465164185,
+      1.8485418558120728,
+      1.8433297872543335,
+      1.8656617403030396,
+      1.7364046573638916,
+      1.763675570487976,
+      1.676373839378357,
+      1.7110265493392944,
+      1.615490198135376,
+      1.6555925607681274,
+      1.6771615743637085,
+      1.5803605318069458,
+      1.698895812034607,
+      1.6376783847808838,
+      1.5982328653335571,
+      1.639822244644165,
+      1.5648831129074097,
+      1.6399648189544678,
+      1.6553243398666382,
+      1.6260439157485962,
+      1.5946561098098755,
+      1.5571329593658447,
+      1.6678483486175537,
+      1.5978716611862183,
+      1.579860806465149,
+      1.547479510307312,
+      1.6490713357925415,
+      1.5853873491287231,
+      1.6186771392822266,
+      1.629547119140625,
+      1.5803824663162231,
+      1.6232846975326538,
+      1.5827962160110474,
+      1.5661370754241943,
+      1.595491647720337,
+      1.5508081912994385,
+      1.5436824560165405,
+      1.4762951135635376,
+      1.534109354019165,
+      1.5676032304763794,
+      1.6045726537704468,
+      1.6091138124465942,
+      1.5710519552230835,
+      1.5058962106704712,
+      1.5800859928131104,
+      1.5821422338485718,
+      1.4647464752197266,
+      1.3848673105239868,
+      1.2519539594650269,
+      1.1015363931655884,
+      1.0217984914779663,
+      0.845373272895813,
+      0.8137518167495728,
+      0.7536793947219849,
+      0.734664261341095,
+      0.6096369624137878,
+      0.6363228559494019,
+      0.5888699889183044,
+      0.6166903376579285,
+      0.5632797479629517,
+      0.5683057904243469,
+      0.6647682189941406,
+      0.49327102303504944,
+      0.6592168807983398,
+      0.4475482404232025,
+      0.45858272910118103,
+      0.4504123330116272,
+      0.4464643597602844,
+      0.434716135263443,
+      0.4710227847099304,
+      0.41418787837028503,
+      0.4333138167858124,
+      0.41829705238342285,
+      0.41405433416366577,
+      0.34256479144096375,
+      0.4015984535217285,
+      0.3865906894207001,
+      0.34953573346138,
+      0.3254949152469635,
+      0.33393600583076477,
+      0.36152055859565735,
+      0.3228608965873718,
+      0.33522674441337585,
+      0.3540439009666443,
+      0.29312536120414734,
+      0.31177449226379395,
+      0.2804255187511444,
+      0.2624456286430359,
+      0.24266862869262695,
+      0.27142927050590515,
+      0.256727010011673,
+      0.23595015704631805,
+      0.24898859858512878,
+      0.22764110565185547,
+      0.25850754976272583,
+      0.31883153319358826,
+      0.23297050595283508,
+      0.24542836844921112,
+      0.2364697903394699,
+      0.24855555593967438,
+      0.2585849165916443,
+      0.2022559493780136,
+      0.3883301913738251,
+      0.1663922220468521,
+      0.18828082084655762,
+      0.25600293278694153,
+      0.23163311183452606,
+      0.23427070677280426,
+      0.1647852659225464,
+      0.2818305194377899,
+      0.17450526356697083,
+      0.21574051678180695,
+      0.17741680145263672,
+      0.18111097812652588,
+      0.16546352207660675,
+      0.18169072270393372,
+      0.21692071855068207,
+      0.17771503329277039,
+      0.20947639644145966,
+      0.16193951666355133,
+      0.1745607554912567,
+      0.1820065975189209,
+      0.15958893299102783,
+      0.1951303482055664,
+      0.15986397862434387,
+      0.1828688383102417,
+      0.15678441524505615,
+      0.15453293919563293,
+      0.14549635350704193,
+      0.15015602111816406,
+      0.16589994728565216,
+      0.2083023488521576,
+      0.19756896793842316,
+      0.1769312620162964,
+      0.13748231530189514,
+      0.16848985850811005,
+      0.15637251734733582,
+      0.1542062908411026,
+      0.18510974943637848,
+      0.23901475965976715,
+      0.14729569852352142,
+      0.13061518967151642,
+      0.1632332056760788,
+      0.18807141482830048,
+      0.16241005063056946,
+      0.13961409032344818,
+      0.15025213360786438,
+      0.16993758082389832,
+      0.2136787623167038,
+      0.15274199843406677,
+      0.11502443999052048,
+      0.17928506433963776,
+      0.12414716929197311,
+      0.16716186702251434,
+      0.14670038223266602,
+      0.1401141881942749,
+      0.1426459550857544,
+      0.1390915811061859,
+      0.22745878994464874,
+      0.14163002371788025,
+      0.15047256648540497,
+      0.1763717383146286,
+      0.16612128913402557,
+      0.13839386403560638,
+      0.14029578864574432,
+      0.14828573167324066,
+      0.15654674172401428,
+      0.12895287573337555,
+      0.10966147482395172,
+      0.1566745936870575,
+      0.11825021356344223,
+      0.15899570286273956,
+      0.11417065560817719,
+      0.12207361310720444,
+      0.1464504897594452,
+      0.11314932256937027,
+      0.14121291041374207,
+      0.1731998473405838,
+      0.10785476118326187,
+      0.10474668443202972,
+      0.11097441613674164,
+      0.14920303225517273,
+      0.1357085406780243,
+      0.14436888694763184,
+      0.11719918251037598,
+      0.12284661829471588,
+      0.11041203886270523,
+      0.12166093289852142,
+      0.10262764990329742,
+      0.10752516239881516,
+      0.14113996922969818,
+      0.15606527030467987,
+      0.08586384356021881,
+      0.10451855510473251,
+      0.10989884287118912,
+      0.10847078263759613,
+      0.1371469348669052,
+      0.12350263446569443,
+      0.11938208341598511,
+      0.09504014253616333,
+      0.0879065990447998,
+      0.10795427858829498,
+      0.10538385808467865,
+      0.08429454267024994,
+      0.12348323315382004,
+      0.08275620639324188,
+      0.09877373278141022,
+      0.12323183566331863,
+      0.12549912929534912,
+      0.08515826612710953,
+      0.1083882674574852,
+      0.07765626907348633,
+      0.1146833747625351,
+      0.09415219724178314,
+      0.12638306617736816,
+      0.09567337483167648,
+      0.10269604623317719,
+      0.10050521790981293,
+      0.13955256342887878,
+      0.10265278071165085,
+      0.09245339781045914,
+      0.1012834683060646,
+      0.10195635259151459,
+      0.0989336296916008,
+      0.07850481569766998,
+      0.11410516500473022,
+      0.11699012666940689,
+      0.10463123768568039,
+      0.10138637572526932,
+      0.09315840154886246,
+      0.08152943849563599,
+      0.08043662458658218,
+      0.0696566104888916,
+      0.11788948625326157,
+      0.09615107625722885,
+      0.084773488342762,
+      0.08405060321092606,
+      0.06331674009561539,
+      0.0781053826212883,
+      0.07224418967962265,
+      0.10328739136457443,
+      0.1186985969543457,
+      0.09797065705060959,
+      0.07149884849786758,
+      0.08715981990098953,
+      0.09124607592821121,
+      0.08693216741085052,
+      0.07098497450351715,
+      0.07794861495494843,
+      0.07411247491836548,
+      0.08904915302991867,
+      0.06491655856370926,
+      0.0776715949177742,
+      0.08977745473384857,
+      0.07693883031606674,
+      0.08406933397054672,
+      0.09536425024271011,
+      0.09354664385318756,
+      0.07439922541379929,
+      0.08047791570425034,
+      0.07951271533966064,
+      0.0660519227385521,
+      0.07324367016553879,
+      0.06856359541416168,
+      0.08066720515489578,
+      0.0501202717423439,
+      0.09389287978410721,
+      0.09163025766611099,
+      0.08235735446214676,
+      0.08668408542871475,
+      0.104389987885952,
+      0.06577304750680923,
+      0.05824172496795654,
+      0.06805268675088882,
+      0.07189692556858063,
+      0.07321818172931671,
+      0.09301216155290604,
+      0.08259880542755127,
+      0.06701328605413437,
+      0.07296095043420792,
+      0.05979323387145996,
+      0.06692303717136383,
+      0.06562996655702591,
+      0.04991487041115761,
+      0.05980746075510979,
+      0.06248386576771736,
+      0.0744619369506836,
+      0.08007364720106125,
+      0.06882565468549728,
+      0.08849557489156723,
+      0.07901816070079803
+    ],
+    "info_loss": [
+      -0.11871623992919922,
+      -0.14137744903564453,
+      -0.27831506729125977,
+      -0.1530928611755371,
+      -0.20770788192749023,
+      -0.09761548042297363,
+      -0.049141645431518555,
+      -0.0583651065826416,
+      -0.05298948287963867,
+      -0.08501315116882324,
+      -0.0592193603515625,
+      -0.09322202205657959,
+      -0.12354552745819092,
+      -0.15344345569610596,
+      -0.17847394943237305,
+      -0.3257317543029785,
+      -0.46242451667785645,
+      -0.8056128025054932,
+      -0.9225881099700928,
+      -1.1132266521453857,
+      -1.1149460077285767,
+      -1.2374181747436523,
+      -1.3308024406433105,
+      -1.3271249532699585,
+      -1.4295438528060913,
+      -1.436357855796814,
+      -1.501106858253479,
+      -1.4593278169631958,
+      -1.5458439588546753,
+      -1.5861430168151855,
+      -1.5073487758636475,
+      -1.6434880495071411,
+      -1.5584927797317505,
+      -1.5328888893127441,
+      -1.5932952165603638,
+      -1.504364013671875,
+      -1.5911858081817627,
+      -1.6138381958007812,
+      -1.602057933807373,
+      -1.5550190210342407,
+      -1.5279000997543335,
+      -1.6439439058303833,
+      -1.5867046117782593,
+      -1.5650498867034912,
+      -1.5295556783676147,
+      -1.6402710676193237,
+      -1.5791656970977783,
+      -1.6110172271728516,
+      -1.6211673021316528,
+      -1.5715925693511963,
+      -1.620406150817871,
+      -1.5801424980163574,
+      -1.5641402006149292,
+      -1.5935267210006714,
+      -1.5490223169326782,
+      -1.5424306392669678,
+      -1.4748547077178955,
+      -1.5328766107559204,
+      -1.5664801597595215,
+      -1.603726863861084,
+      -1.6080092191696167,
+      -1.5664794445037842,
+      -1.503515601158142,
+      -1.5792337656021118,
+      -1.580399751663208,
+      -1.4628851413726807,
+      -1.3819822072982788,
+      -1.2503565549850464,
+      -1.0993303060531616,
+      -1.0141922235488892,
+      -0.8425378203392029,
+      -0.8103099465370178,
+      -0.7460905909538269,
+      -0.7272024750709534,
+      -0.574376106262207,
+      -0.6344446539878845,
+      -0.5875888466835022,
+      -0.6155530214309692,
+      -0.5619609355926514,
+      -0.5675796270370483,
+      -0.6624935865402222,
+      -0.4922202527523041,
+      -0.6576066017150879,
+      -0.4468822181224823,
+      -0.45621412992477417,
+      -0.44929325580596924,
+      -0.44582489132881165,
+      -0.4315442442893982,
+      -0.47013962268829346,
+      -0.41347256302833557,
+      -0.4320799708366394,
+      -0.41779038310050964,
+      -0.41327568888664246,
+      -0.3419603109359741,
+      -0.40124937891960144,
+      -0.3858768939971924,
+      -0.34896141290664673,
+      -0.32399991154670715,
+      -0.3315870463848114,
+      -0.3593605160713196,
+      -0.3226620852947235,
+      -0.33104127645492554,
+      -0.35347723960876465,
+      -0.2920803427696228,
+      -0.3079918622970581,
+      -0.2797461450099945,
+      -0.26059624552726746,
+      -0.24012351036071777,
+      -0.27054256200790405,
+      -0.2565530240535736,
+      -0.23481540381908417,
+      -0.24855494499206543,
+      -0.22735653817653656,
+      -0.24865524470806122,
+      -0.3175896108150482,
+      -0.2319718897342682,
+      -0.2450624704360962,
+      -0.23596806824207306,
+      -0.24763233959674835,
+      -0.25789570808410645,
+      -0.20208360254764557,
+      -0.3878355324268341,
+      -0.1660861074924469,
+      -0.1880086362361908,
+      -0.2552400529384613,
+      -0.2314588874578476,
+      -0.2339198887348175,
+      -0.16447849571704865,
+      -0.28169310092926025,
+      -0.17403823137283325,
+      -0.21548306941986084,
+      -0.17581123113632202,
+      -0.1803392618894577,
+      -0.1650605946779251,
+      -0.1814371794462204,
+      -0.21547359228134155,
+      -0.17492720484733582,
+      -0.20845045149326324,
+      -0.16139690577983856,
+      -0.17442575097084045,
+      -0.18162329494953156,
+      -0.15935660898685455,
+      -0.1947454810142517,
+      -0.1596723347902298,
+      -0.18266050517559052,
+      -0.15669506788253784,
+      -0.15445366501808167,
+      -0.14542245864868164,
+      -0.1500117927789688,
+      -0.1657772809267044,
+      -0.20730118453502655,
+      -0.1968885213136673,
+      -0.1759709268808365,
+      -0.1373818963766098,
+      -0.16828355193138123,
+      -0.15632285177707672,
+      -0.1541530042886734,
+      -0.18477436900138855,
+      -0.23842284083366394,
+      -0.14703422784805298,
+      -0.12765443325042725,
+      -0.14535963535308838,
+      -0.18799753487110138,
+      -0.1622273027896881,
+      -0.13954323530197144,
+      -0.14984118938446045,
+      -0.16965319216251373,
+      -0.21306651830673218,
+      -0.1525978446006775,
+      -0.1124684140086174,
+      -0.17801117897033691,
+      -0.1240897998213768,
+      -0.1620187759399414,
+      -0.1460704207420349,
+      -0.1397659331560135,
+      -0.14243143796920776,
+      -0.1384785920381546,
+      -0.2272264063358307,
+      -0.14157363772392273,
+      -0.15038394927978516,
+      -0.17587241530418396,
+      -0.16556702554225922,
+      -0.13828729093074799,
+      -0.13996198773384094,
+      -0.14555414021015167,
+      -0.15643732249736786,
+      -0.1286497712135315,
+      -0.10931475460529327,
+      -0.15654276311397552,
+      -0.11817201972007751,
+      -0.15730591118335724,
+      -0.11406108736991882,
+      -0.12194474041461945,
+      -0.14630363881587982,
+      -0.11232269555330276,
+      -0.1398424506187439,
+      -0.1729625016450882,
+      -0.10761153697967529,
+      -0.10124911367893219,
+      -0.11021573841571808,
+      -0.14916908740997314,
+      -0.13568000495433807,
+      -0.1437206268310547,
+      -0.1160874143242836,
+      -0.12278911471366882,
+      -0.11030962318181992,
+      -0.12157618999481201,
+      -0.10260012000799179,
+      -0.10745620727539062,
+      -0.13962514698505402,
+      -0.15587244927883148,
+      -0.08550996333360672,
+      -0.10447712242603302,
+      -0.10986152291297913,
+      -0.10843180865049362,
+      -0.132799431681633,
+      -0.12327650934457779,
+      -0.1192261204123497,
+      -0.09495997428894043,
+      -0.08783882856369019,
+      -0.10792768746614456,
+      -0.10432849824428558,
+      -0.08426912128925323,
+      -0.12343040853738785,
+      -0.08272845298051834,
+      -0.09874361008405685,
+      -0.12303923070430756,
+      -0.12547725439071655,
+      -0.08496551960706711,
+      -0.1083354726433754,
+      -0.07763051986694336,
+      -0.11461926251649857,
+      -0.09409280121326447,
+      -0.12636896967887878,
+      -0.09553775936365128,
+      -0.10268252342939377,
+      -0.10048936307430267,
+      -0.13953720033168793,
+      -0.10260877013206482,
+      -0.09243543446063995,
+      -0.10126668214797974,
+      -0.10184351354837418,
+      -0.09890984743833542,
+      -0.07839582115411758,
+      -0.11393488198518753,
+      -0.1169787347316742,
+      -0.10461471229791641,
+      -0.10137353837490082,
+      -0.09313812106847763,
+      -0.08150708675384521,
+      -0.08026055991649628,
+      -0.0696309357881546,
+      -0.11784257739782333,
+      -0.09613538533449173,
+      -0.08474813401699066,
+      -0.0840376764535904,
+      -0.06305735558271408,
+      -0.07809284329414368,
+      -0.07223228365182877,
+      -0.10327018052339554,
+      -0.11867464333772659,
+      -0.09795786440372467,
+      -0.0714866891503334,
+      -0.08673679083585739,
+      -0.09118502587080002,
+      -0.08327877521514893,
+      -0.07097714394330978,
+      -0.07780350744724274,
+      -0.07409871369600296,
+      -0.08903937041759491,
+      -0.06490036100149155,
+      -0.07765685021877289,
+      -0.08974980562925339,
+      -0.07693041861057281,
+      -0.08406312018632889,
+      -0.09535522013902664,
+      -0.09353701770305634,
+      -0.07436812669038773,
+      -0.08039438724517822,
+      -0.07950684428215027,
+      -0.06602951139211655,
+      -0.07323478162288666,
+      -0.06851720064878464,
+      -0.08064503967761993,
+      -0.050105925649404526,
+      -0.09388339519500732,
+      -0.09162209182977676,
+      -0.0823427215218544,
+      -0.08666205406188965,
+      -0.10436543822288513,
+      -0.06575608253479004,
+      -0.05822037532925606,
+      -0.06804681569337845,
+      -0.07189025729894638,
+      -0.07319743931293488,
+      -0.09294571727514267,
+      -0.08258873224258423,
+      -0.06700537353754044,
+      -0.0729491114616394,
+      -0.059782713651657104,
+      -0.06691405922174454,
+      -0.06561791151762009,
+      -0.04990833252668381,
+      -0.05980100482702255,
+      -0.062476858496665955,
+      -0.07444346696138382,
+      -0.08006814867258072,
+      -0.06881572306156158,
+      -0.08848292380571365,
+      -0.07901027798652649
+    ],
+    "abs_loss": [
+      4.248014450073242,
+      4.234400749206543,
+      4.188644886016846,
+      4.105509281158447,
+      3.925417423248291,
+      3.289015769958496,
+      3.1008191108703613,
+      2.9618568420410156,
+      3.020190954208374,
+      2.9729349613189697,
+      2.8805203437805176,
+      3.0525832176208496,
+      2.9079132080078125,
+      2.9795520305633545,
+      2.9216318130493164,
+      3.0001397132873535,
+      2.919717311859131,
+      2.9154083728790283,
+      2.957773208618164,
+      2.91117787361145,
+      2.740760326385498,
+      2.5675408840179443,
+      2.148101329803467,
+      2.039168357849121,
+      1.8643121719360352,
+      1.7577546834945679,
+      1.5585510730743408,
+      1.7541223764419556,
+      1.645475149154663,
+      1.5547354221343994,
+      1.5044190883636475,
+      1.5476247072219849,
+      1.5362157821655273,
+      1.439476728439331,
+      1.419737696647644,
+      1.427870273590088,
+      1.2846004962921143,
+      1.4998186826705933,
+      1.4611632823944092,
+      1.4195325374603271,
+      1.2427997589111328,
+      1.3025373220443726,
+      1.525195598602295,
+      1.1897931098937988,
+      1.1708741188049316,
+      1.1220957040786743,
+      1.1160176992416382,
+      0.9629797339439392,
+      1.0605902671813965,
+      0.876197874546051,
+      0.9425597190856934,
+      0.6959391832351685,
+      0.6551886200904846,
+      0.6491069793701172,
+      0.8603583574295044,
+      0.7898715734481812,
+      0.796775221824646,
+      0.6029624342918396,
+      0.5876208543777466,
+      0.6276936531066895,
+      0.6227383613586426,
+      0.5754898190498352,
+      0.6257544159889221,
+      0.6736498475074768,
+      0.5511360168457031,
+      0.4754224419593811,
+      0.5097036361694336,
+      0.6583508253097534,
+      0.6496922969818115,
+      0.6905272603034973,
+      0.5980067253112793,
+      0.6397318243980408,
+      0.42971271276474,
+      0.5821137428283691,
+      0.4668859839439392,
+      0.582308292388916,
+      0.5925179719924927,
+      0.48843228816986084,
+      0.5182346105575562,
+      0.5913019180297852,
+      0.5356505513191223,
+      0.5443198680877686,
+      0.4625856876373291,
+      0.5666908025741577,
+      0.5563081502914429,
+      0.6170629262924194,
+      0.6245989203453064,
+      0.41324830055236816,
+      0.6230974793434143,
+      0.49977999925613403,
+      0.5869432091712952,
+      0.4126822054386139,
+      0.4490078389644623,
+      0.5880581140518188,
+      0.5484663248062134,
+      0.4388089179992676,
+      0.5843925476074219,
+      0.5470926761627197,
+      0.4574468433856964,
+      0.5226950645446777,
+      0.5557544231414795,
+      0.4983421266078949,
+      0.5043578147888184,
+      0.48045799136161804,
+      0.5108127593994141,
+      0.5677489638328552,
+      0.44493192434310913,
+      0.4161803126335144,
+      0.5129793882369995,
+      0.5522379875183105,
+      0.5025249123573303,
+      0.48488345742225647,
+      0.4656607508659363,
+      0.3821223974227905,
+      0.45388126373291016,
+      0.5386711359024048,
+      0.5228223204612732,
+      0.4839162230491638,
+      0.4646385610103607,
+      0.44527536630630493,
+      0.48525339365005493,
+      0.523290753364563,
+      0.505596399307251,
+      0.49558740854263306,
+      0.5670284032821655,
+      0.4900159537792206,
+      0.5530015826225281,
+      0.4023090600967407,
+      0.3795967102050781,
+      0.45430195331573486,
+      0.4961245656013489,
+      0.37589558959007263,
+      0.5071943998336792,
+      0.39159664511680603,
+      0.3987557888031006,
+      0.4503222703933716,
+      0.40217238664627075,
+      0.3273792862892151,
+      0.38206928968429565,
+      0.45704320073127747,
+      0.3849458396434784,
+      0.4018760323524475,
+      0.4915027618408203,
+      0.4851424992084503,
+      0.3422090709209442,
+      0.3465098440647125,
+      0.4339431822299957,
+      0.35385948419570923,
+      0.500917911529541,
+      0.3932620882987976,
+      0.31006529927253723,
+      0.34924912452697754,
+      0.33225369453430176,
+      0.42161425948143005,
+      0.3447437882423401,
+      0.28126439452171326,
+      0.3864516019821167,
+      0.4131661057472229,
+      0.34799614548683167,
+      0.2833597660064697,
+      0.4262927174568176,
+      0.38478782773017883,
+      0.3123430609703064,
+      0.39662599563598633,
+      0.3168265223503113,
+      0.3374044597148895,
+      0.4509154260158539,
+      0.4092380702495575,
+      0.3720371723175049,
+      0.32245540618896484,
+      0.3193129897117615,
+      0.4018149971961975,
+      0.32657575607299805,
+      0.3378874659538269,
+      0.28595322370529175,
+      0.36344659328460693,
+      0.35009145736694336,
+      0.33734726905822754,
+      0.35329532623291016,
+      0.35059797763824463,
+      0.2901272773742676,
+      0.3283654451370239,
+      0.3822149932384491,
+      0.27559030055999756,
+      0.2803605794906616,
+      0.2743658423423767,
+      0.38650450110435486,
+      0.29735124111175537,
+      0.2636435627937317,
+      0.3230719566345215,
+      0.2562853991985321,
+      0.2919286787509918,
+      0.2846037447452545,
+      0.34604984521865845,
+      0.333662748336792,
+      0.3608507215976715,
+      0.29711613059043884,
+      0.2485959827899933,
+      0.27522337436676025,
+      0.23814435303211212,
+      0.28669074177742004,
+      0.3444576859474182,
+      0.2782653570175171,
+      0.2843083143234253,
+      0.2942905128002167,
+      0.31752514839172363,
+      0.2586301863193512,
+      0.2982809245586395,
+      0.41684481501579285,
+      0.33103907108306885,
+      0.2990052103996277,
+      0.27157121896743774,
+      0.27298983931541443,
+      0.32912784814834595,
+      0.2982273995876312,
+      0.2846553921699524,
+      0.24887534976005554,
+      0.27428680658340454,
+      0.30408716201782227,
+      0.3042374551296234,
+      0.2922416925430298,
+      0.253709614276886,
+      0.28282374143600464,
+      0.23307877779006958,
+      0.2911456227302551,
+      0.3480558395385742,
+      0.30271637439727783,
+      0.305656373500824,
+      0.31656521558761597,
+      0.230843186378479,
+      0.3473528027534485,
+      0.2783201038837433,
+      0.20797383785247803,
+      0.26997676491737366,
+      0.24719452857971191,
+      0.2668984532356262,
+      0.2427939772605896,
+      0.2640213966369629,
+      0.3293270468711853,
+      0.3124449551105499,
+      0.2541097402572632,
+      0.2846413850784302,
+      0.28788357973098755,
+      0.2490839958190918,
+      0.31947681307792664,
+      0.23902346193790436,
+      0.19850093126296997,
+      0.2748686373233795,
+      0.2085222750902176,
+      0.21743398904800415,
+      0.2422069013118744,
+      0.24782437086105347,
+      0.28087347745895386,
+      0.22643589973449707,
+      0.2488379180431366,
+      0.36067014932632446,
+      0.22233793139457703,
+      0.2832460105419159,
+      0.24721194803714752,
+      0.23525303602218628,
+      0.29794538021087646,
+      0.30321621894836426,
+      0.20213299989700317,
+      0.27157583832740784,
+      0.259025514125824,
+      0.21834829449653625,
+      0.21391159296035767,
+      0.29406437277793884,
+      0.22560474276542664,
+      0.3151233196258545,
+      0.3564601242542267,
+      0.21590474247932434,
+      0.2805882394313812,
+      0.2424798309803009,
+      0.24720430374145508,
+      0.24170127511024475,
+      0.26862311363220215,
+      0.2529475688934326,
+      0.2638781666755676,
+      0.30753597617149353,
+      0.3029334843158722,
+      0.24195852875709534,
+      0.29423952102661133,
+      0.26509591937065125,
+      0.2073935717344284,
+      0.2789330780506134,
+      0.27929404377937317,
+      0.26568663120269775,
+      0.21595841646194458,
+      0.24741975963115692,
+      0.29185062646865845,
+      0.27799612283706665,
+      0.23569640517234802,
+      0.2136664092540741,
+      0.19363853335380554,
+      0.2502439022064209,
+      0.2523377537727356,
+      0.2553808093070984,
+      0.2667678892612457,
+      0.30567339062690735,
+      0.18897244334220886,
+      0.2077244222164154,
+      0.22972536087036133,
+      0.2555011510848999,
+      0.1943330019712448,
+      0.21900857985019684,
+      0.2112642526626587,
+      0.21387708187103271,
+      0.33129119873046875,
+      0.22131985425949097
+    ],
+    "zipf_loss": [
+      5.902684688568115,
+      5.060757637023926,
+      3.7189528942108154,
+      2.5040011405944824,
+      1.5840965509414673,
+      0.8555037975311279,
+      0.6532659530639648,
+      0.6063866019248962,
+      0.5856895446777344,
+      0.5765976905822754,
+      0.5784863233566284,
+      0.5665068626403809,
+      0.5708044767379761,
+      0.5700060129165649,
+      0.5712013244628906,
+      0.5608698725700378,
+      0.5528655052185059,
+      0.5267254114151001,
+      0.4755018353462219,
+      0.3874831199645996,
+      0.3377814292907715,
+      0.29368188977241516,
+      0.24276527762413025,
+      0.20423030853271484,
+      0.19219207763671875,
+      0.17443324625492096,
+      0.15989655256271362,
+      0.15140657126903534,
+      0.1338139772415161,
+      0.14818087220191956,
+      0.1131894439458847,
+      0.10705186426639557,
+      0.0981040894985199,
+      0.09666009992361069,
+      0.07536322623491287,
+      0.06556831300258636,
+      0.06283900141716003,
+      0.05518872290849686,
+      0.04969022050499916,
+      0.048817746341228485,
+      0.05529868230223656,
+      0.03984316438436508,
+      0.04392576217651367,
+      0.05443081259727478,
+      0.04846706613898277,
+      0.034341730177402496,
+      0.04312724992632866,
+      0.035865068435668945,
+      0.04203113168478012,
+      0.04487438499927521,
+      0.03540477901697159,
+      0.04448772966861725,
+      0.060964684933423996,
+      0.053344178944826126,
+      0.03739790618419647,
+      0.0649472177028656,
+      0.033584870398044586,
+      0.04236011952161789,
+      0.032370828092098236,
+      0.042076606303453445,
+      0.07509628683328629,
+      0.032922953367233276,
+      0.03562316298484802,
+      0.033071745187044144,
+      0.04327850043773651,
+      0.0474778413772583,
+      0.038080718368291855,
+      0.03217237442731857,
+      0.025619802996516228,
+      0.03434999659657478,
+      0.04067017138004303,
+      0.0432572066783905,
+      0.06351292878389359,
+      0.029371127486228943,
+      0.049407556653022766,
+      0.02926669269800186,
+      0.03936169296503067,
+      0.040858425199985504,
+      0.031939730048179626,
+      0.029323050752282143,
+      0.029477233067154884,
+      0.028299378231167793,
+      0.04091307148337364,
+      0.03514455258846283,
+      0.04208677262067795,
+      0.026792136952280998,
+      0.024593273177742958,
+      0.05197222903370857,
+      0.030720636248588562,
+      0.03861355036497116,
+      0.03301554173231125,
+      0.04811073839664459,
+      0.03426514193415642,
+      0.04966907203197479,
+      0.028800886124372482,
+      0.027979116886854172,
+      0.030924420803785324,
+      0.023811589926481247,
+      0.0421719029545784,
+      0.03535270318388939,
+      0.03264780342578888,
+      0.03614368289709091,
+      0.027222566306591034,
+      0.032614752650260925,
+      0.032340727746486664,
+      0.03658295422792435,
+      0.035718291997909546,
+      0.043458759784698486,
+      0.025101013481616974,
+      0.050936803221702576,
+      0.03274014592170715,
+      0.03148721158504486,
+      0.03349090367555618,
+      0.0450107641518116,
+      0.04847481846809387,
+      0.03280961140990257,
+      0.02714274264872074,
+      0.03508167341351509,
+      0.049645956605672836,
+      0.03934843838214874,
+      0.03244860842823982,
+      0.02164081111550331,
+      0.03862951323390007,
+      0.03326089680194855,
+      0.03867378830909729,
+      0.03956107050180435,
+      0.02918875403702259,
+      0.026714175939559937,
+      0.034605640918016434,
+      0.03499145060777664,
+      0.036697641015052795,
+      0.04140329360961914,
+      0.035553041845560074,
+      0.060954466462135315,
+      0.033179301768541336,
+      0.04456836357712746,
+      0.03654774650931358,
+      0.041403159499168396,
+      0.036835212260484695,
+      0.027645012363791466,
+      0.03495408594608307,
+      0.030682988464832306,
+      0.055629342794418335,
+      0.042602382600307465,
+      0.05404828116297722,
+      0.034428223967552185,
+      0.03452951833605766,
+      0.04957253113389015,
+      0.02790767326951027,
+      0.03789380565285683,
+      0.05088239163160324,
+      0.03013448975980282,
+      0.04763920605182648,
+      0.03356539458036423,
+      0.03152993321418762,
+      0.03610782325267792,
+      0.03359227254986763,
+      0.025854911655187607,
+      0.05199098587036133,
+      0.04216059297323227,
+      0.04234059154987335,
+      0.04226396232843399,
+      0.04976421967148781,
+      0.02282628044486046,
+      0.04069782793521881,
+      0.03215024620294571,
+      0.03314099460840225,
+      0.024782083928585052,
+      0.029297877103090286,
+      0.073276586830616,
+      0.0418546199798584,
+      0.045323487371206284,
+      0.03853926062583923,
+      0.04253818094730377,
+      0.04506561905145645,
+      0.04157973825931549,
+      0.04301469773054123,
+      0.03566312417387962,
+      0.0506528839468956,
+      0.04138036444783211,
+      0.04187582805752754,
+      0.06314846873283386,
+      0.039209991693496704,
+      0.04790130630135536,
+      0.04376837611198425,
+      0.05028841271996498,
+      0.040840305387973785,
+      0.029018813744187355,
+      0.04631214216351509,
+      0.04048272967338562,
+      0.054269835352897644,
+      0.06053861230611801,
+      0.05184292793273926,
+      0.04773920774459839,
+      0.03577277809381485,
+      0.04363023117184639,
+      0.041978538036346436,
+      0.04979187995195389,
+      0.03781772404909134,
+      0.052146218717098236,
+      0.0459938645362854,
+      0.042001236230134964,
+      0.04234977439045906,
+      0.05187782645225525,
+      0.03706303983926773,
+      0.04720836877822876,
+      0.04916055500507355,
+      0.03775591403245926,
+      0.03928117826581001,
+      0.030798811465501785,
+      0.032894425094127655,
+      0.03738286718726158,
+      0.04667762666940689,
+      0.04798555001616478,
+      0.04211629182100296,
+      0.03676827251911163,
+      0.04022987186908722,
+      0.04925571754574776,
+      0.027465006336569786,
+      0.05306820571422577,
+      0.040601249784231186,
+      0.03433593362569809,
+      0.03603116795420647,
+      0.05245066061615944,
+      0.043414343148469925,
+      0.035300515592098236,
+      0.04520798474550247,
+      0.04005079343914986,
+      0.03389983996748924,
+      0.036058053374290466,
+      0.03387245535850525,
+      0.030419837683439255,
+      0.051053088158369064,
+      0.06218193098902702,
+      0.04528661072254181,
+      0.04498759284615517,
+      0.040742211043834686,
+      0.03306664898991585,
+      0.03345557674765587,
+      0.035349950194358826,
+      0.05530022829771042,
+      0.04764458164572716,
+      0.03253139555454254,
+      0.04161981865763664,
+      0.04827052354812622,
+      0.03459079563617706,
+      0.05276760458946228,
+      0.033949095755815506,
+      0.028391260653734207,
+      0.03912941366434097,
+      0.03616856038570404,
+      0.04241953045129776,
+      0.050922323018312454,
+      0.038664426654577255,
+      0.05536889657378197,
+      0.034565310925245285,
+      0.04321820288896561,
+      0.040786027908325195,
+      0.032071419060230255,
+      0.050359830260276794,
+      0.048944178968667984,
+      0.044813625514507294,
+      0.05466174706816673,
+      0.040300920605659485,
+      0.04125305265188217,
+      0.057422131299972534,
+      0.056044429540634155,
+      0.030909128487110138,
+      0.03716030344367027,
+      0.04249592870473862,
+      0.03556758910417557,
+      0.037934184074401855,
+      0.05018708482384682,
+      0.034316956996917725,
+      0.03639369457960129,
+      0.045960068702697754,
+      0.03193551301956177,
+      0.042806319892406464,
+      0.039242617785930634,
+      0.05495717376470566,
+      0.037742339074611664,
+      0.04147646576166153,
+      0.046967800706624985,
+      0.04591631889343262,
+      0.04412830248475075,
+      0.0321836844086647,
+      0.03613559156656265,
+      0.03604351356625557,
+      0.03837473690509796,
+      0.04597822576761246,
+      0.037280354648828506,
+      0.03376632183790207,
+      0.035324495285749435,
+      0.03834962472319603,
+      0.0472051165997982,
+      0.03315430134534836,
+      0.04054207354784012,
+      0.03822512552142143,
+      0.034464433789253235,
+      0.03335137665271759,
+      0.048479434102773666,
+      0.04830781742930412,
+      0.04486474022269249,
+      0.04346194863319397,
+      0.033721063286066055,
+      0.04194709658622742,
+      0.05212908983230591,
+      0.04137463867664337,
+      0.06516798585653305,
+      0.04277731478214264
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8454185724258423,
+      0.6524177193641663,
+      0.3698425590991974,
+      0.2007085680961609,
+      0.1236860603094101,
+      0.08822250366210938,
+      0.0750516727566719,
+      0.05863531678915024,
+      0.04750237613916397,
+      0.04451904818415642,
+      0.04229466989636421,
+      0.040987420827150345,
+      0.04738488793373108,
+      0.05249207466840744,
+      0.05897305905818939,
+      0.060233548283576965,
+      0.062270693480968475,
+      0.07207406312227249,
+      0.08198414742946625,
+      0.09681784361600876,
+      0.1064775362610817,
+      0.11615313589572906,
+      0.12525823712348938,
+      0.13203047215938568,
+      0.13891912996768951,
+      0.14662884175777435,
+      0.15332895517349243,
+      0.15664303302764893,
+      0.15949848294258118,
+      0.1633111983537674,
+      0.16485799849033356,
+      0.16638898849487305,
+      0.16665266454219818,
+      0.16638712584972382,
+      0.1657113879919052,
+      0.1661071628332138,
+      0.1650349497795105,
+      0.16582758724689484,
+      0.16719463467597961,
+      0.16592173278331757,
+      0.16679780185222626,
+      0.17262960970401764,
+      0.17457710206508636,
+      0.1743481308221817,
+      0.1783905178308487,
+      0.18017259240150452,
+      0.18341462314128876,
+      0.18411006033420563,
+      0.18720033764839172,
+      0.1891525238752365,
+      0.18835467100143433,
+      0.19066214561462402,
+      0.19213086366653442,
+      0.19740790128707886,
+      0.19639301300048828,
+      0.19970081746578217,
+      0.20187750458717346,
+      0.20390638709068298,
+      0.2044963538646698,
+      0.20567278563976288,
+      0.2079830914735794,
+      0.21193991601467133,
+      0.21391405165195465,
+      0.21922127902507782,
+      0.22368207573890686,
+      0.22522194683551788,
+      0.22709041833877563,
+      0.22769533097743988,
+      0.23018060624599457,
+      0.23341462016105652,
+      0.23343972861766815,
+      0.23775576055049896,
+      0.24117596447467804,
+      0.24501429498195648,
+      0.24653051793575287,
+      0.25028756260871887,
+      0.25320684909820557,
+      0.2552432417869568,
+      0.2562824487686157,
+      0.25940391421318054,
+      0.26296067237854004,
+      0.26571977138519287,
+      0.2707919776439667,
+      0.27494823932647705,
+      0.2768678665161133,
+      0.27713948488235474,
+      0.279514342546463,
+      0.2820461094379425,
+      0.28318119049072266,
+      0.2849445044994354,
+      0.2857610285282135,
+      0.28739556670188904,
+      0.29032039642333984,
+      0.2934110164642334,
+      0.2961346507072449,
+      0.2951403260231018,
+      0.2951270043849945,
+      0.2940308749675751,
+      0.29594114422798157,
+      0.2994272708892822,
+      0.3011847734451294,
+      0.30261480808258057,
+      0.30125999450683594,
+      0.30381637811660767,
+      0.3056132197380066,
+      0.30930984020233154,
+      0.3089517652988434,
+      0.30901747941970825,
+      0.316816121339798,
+      0.3187602460384369,
+      0.3199279010295868,
+      0.31832966208457947,
+      0.32230615615844727,
+      0.32506951689720154,
+      0.3242188096046448,
+      0.32370448112487793,
+      0.325268417596817,
+      0.32495614886283875,
+      0.3243638277053833,
+      0.3254850506782532,
+      0.32709378004074097,
+      0.32725614309310913,
+      0.32756146788597107,
+      0.32778775691986084,
+      0.33066636323928833,
+      0.32944557070732117,
+      0.3283247649669647,
+      0.3301420211791992,
+      0.3320577144622803,
+      0.336099773645401,
+      0.3349405825138092,
+      0.33438968658447266,
+      0.3332967758178711,
+      0.3337518274784088,
+      0.3349280059337616,
+      0.33751121163368225,
+      0.33745864033699036,
+      0.3390352129936218,
+      0.3396795392036438,
+      0.3428172767162323,
+      0.34306180477142334,
+      0.3420213460922241,
+      0.34256380796432495,
+      0.34335482120513916,
+      0.3430495858192444,
+      0.3445117771625519,
+      0.34731537103652954,
+      0.3471394181251526,
+      0.34982484579086304,
+      0.35192805528640747,
+      0.3546449840068817,
+      0.35223621129989624,
+      0.3524305820465088,
+      0.3521151542663574,
+      0.35328346490859985,
+      0.3530151844024658,
+      0.3538852632045746,
+      0.35223114490509033,
+      0.3535531759262085,
+      0.35533207654953003,
+      0.35749757289886475,
+      0.359323650598526,
+      0.360779732465744,
+      0.36273473501205444,
+      0.364918977022171,
+      0.3663212060928345,
+      0.3653111457824707,
+      0.36530864238739014,
+      0.36515799164772034,
+      0.3680950105190277,
+      0.36992835998535156,
+      0.37001073360443115,
+      0.3704085648059845,
+      0.3728475868701935,
+      0.3723159730434418,
+      0.3730833828449249,
+      0.37724408507347107,
+      0.378434956073761,
+      0.3795548677444458,
+      0.3785182535648346,
+      0.37985289096832275,
+      0.38019806146621704,
+      0.3810216784477234,
+      0.3825012743473053,
+      0.38243168592453003,
+      0.38221603631973267,
+      0.3820144832134247,
+      0.3820664882659912,
+      0.3802727162837982,
+      0.38027769327163696,
+      0.38286295533180237,
+      0.3812105357646942,
+      0.38233181834220886,
+      0.38291850686073303,
+      0.3830583095550537,
+      0.386683851480484,
+      0.3840213119983673,
+      0.38416072726249695,
+      0.38992780447006226,
+      0.3902909755706787,
+      0.3925168514251709,
+      0.3939124643802643,
+      0.3944869637489319,
+      0.3921501040458679,
+      0.3928128778934479,
+      0.3936733305454254,
+      0.39786452054977417,
+      0.39970171451568604,
+      0.4004788398742676,
+      0.4015267789363861,
+      0.40280917286872864,
+      0.40395042300224304,
+      0.4055359363555908,
+      0.40717577934265137,
+      0.40721243619918823,
+      0.409464567899704,
+      0.40915629267692566,
+      0.4096061885356903,
+      0.40949881076812744,
+      0.41049879789352417,
+      0.41203904151916504,
+      0.41238167881965637,
+      0.41321825981140137,
+      0.4143255949020386,
+      0.41407695412635803,
+      0.4160165786743164,
+      0.4173142910003662,
+      0.41775164008140564,
+      0.41922488808631897,
+      0.42065107822418213,
+      0.4218386709690094,
+      0.42317867279052734,
+      0.4253239035606384,
+      0.42656201124191284,
+      0.42637738585472107,
+      0.4268428385257721,
+      0.42705947160720825,
+      0.4287395477294922,
+      0.42988792061805725,
+      0.43054214119911194,
+      0.431358277797699,
+      0.434470534324646,
+      0.43372422456741333,
+      0.4343772828578949,
+      0.436478853225708,
+      0.43682318925857544,
+      0.4364190101623535,
+      0.4362836480140686,
+      0.43758946657180786,
+      0.43739020824432373,
+      0.4382403492927551,
+      0.43866875767707825,
+      0.43970364332199097,
+      0.4394722282886505,
+      0.44059357047080994,
+      0.44121912121772766,
+      0.4426378607749939,
+      0.4445840120315552,
+      0.4441210627555847,
+      0.4451994299888611,
+      0.4466753304004669,
+      0.4467293322086334,
+      0.4470214247703552,
+      0.4473244547843933,
+      0.4482899010181427,
+      0.449049711227417,
+      0.4494488537311554,
+      0.44887575507164,
+      0.4482535719871521,
+      0.44895297288894653,
+      0.4490957260131836,
+      0.449077308177948,
+      0.4497356414794922,
+      0.44975101947784424,
+      0.4500557482242584,
+      0.4509201943874359,
+      0.4517847001552582,
+      0.45305636525154114,
+      0.45346972346305847,
+      0.4530041813850403,
+      0.45415493845939636,
+      0.45489701628685,
+      0.4551505148410797,
+      0.4561237394809723,
+      0.4570634365081787,
+      0.4576168954372406,
+      0.45753759145736694,
+      0.4584166407585144,
+      0.45879897475242615,
+      0.45931005477905273,
+      0.4600575864315033,
+      0.46008801460266113,
+      0.46043217182159424,
+      0.4608006179332733,
+      0.4609127342700958,
+      0.46162301301956177,
+      0.4619598984718323,
+      0.46156224608421326,
+      0.4617094397544861,
+      0.4619976580142975,
+      0.4621671140193939,
+      0.46231216192245483,
+      0.46249955892562866,
+      0.4629148244857788,
+      0.4631115794181824,
+      0.46322980523109436,
+      0.4634542167186737,
+      0.46354347467422485,
+      0.4636399447917938,
+      0.46376949548721313
+    ],
+    "lr": [
+      4.188034188034189e-06,
+      8.461538461538462e-06,
+      1.2735042735042738e-05,
+      1.700854700854701e-05,
+      2.1282051282051282e-05,
+      2.5555555555555557e-05,
+      2.9829059829059833e-05,
+      3.4102564102564105e-05,
+      3.837606837606838e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      4e-05,
+      3.992995647012267e-05,
+      3.963316185199842e-05,
+      3.933636723387416e-05,
+      3.90395726157499e-05,
+      3.874277799762565e-05,
+      3.844598337950139e-05,
+      3.814918876137713e-05,
+      3.785239414325287e-05,
+      3.7555599525128613e-05,
+      3.725880490700435e-05,
+      3.696201028888009e-05,
+      3.666521567075585e-05,
+      3.6368421052631585e-05,
+      3.607162643450732e-05,
+      3.577483181638307e-05,
+      3.547803719825881e-05,
+      3.518124258013455e-05,
+      3.4884447962010287e-05,
+      3.458765334388604e-05,
+      3.4290858725761776e-05,
+      3.399406410763751e-05,
+      3.3697269489513265e-05,
+      3.3400474871389e-05,
+      3.310368025326474e-05,
+      3.280688563514048e-05,
+      3.251009101701623e-05,
+      3.2213296398891967e-05,
+      3.191650178076771e-05,
+      3.1542540561931145e-05,
+      3.124574594380688e-05,
+      3.094895132568263e-05,
+      3.065215670755838e-05,
+      3.0355362089434113e-05,
+      3.0058567471309854e-05,
+      2.97617728531856e-05,
+      2.9464978235061336e-05,
+      2.9168183616937077e-05,
+      2.8871388998812818e-05,
+      2.857459438068857e-05,
+      2.8277799762564307e-05,
+      2.798100514444005e-05,
+      2.7684210526315793e-05,
+      2.738741590819153e-05,
+      2.709062129006727e-05,
+      2.6793826671943013e-05,
+      2.6497032053818764e-05,
+      2.62002374356945e-05,
+      2.5903442817570243e-05,
+      2.5606648199445987e-05,
+      2.5309853581321725e-05,
+      2.5013058963197466e-05,
+      2.4716264345073214e-05,
+      2.441946972694896e-05,
+      2.4122675108824693e-05,
+      2.3825880490700434e-05,
+      2.352908587257618e-05,
+      2.3232291254451916e-05,
+      2.293549663632766e-05,
+      2.2638702018203408e-05,
+      2.226474079936684e-05,
+      2.196794618124258e-05,
+      2.1671151563118328e-05,
+      2.1374356944994066e-05,
+      2.1077562326869803e-05,
+      2.0780767708745554e-05,
+      2.0483973090621296e-05,
+      2.018717847249703e-05,
+      1.989038385437277e-05,
+      1.959358923624852e-05,
+      1.929679461812426e-05,
+      1.8999999999999998e-05,
+      1.8703205381875746e-05,
+      1.8406410763751487e-05,
+      1.8109616145627224e-05,
+      1.7812821527502965e-05,
+      1.7516026909378713e-05,
+      1.7219232291254454e-05,
+      1.6922437673130192e-05,
+      1.662564305500594e-05,
+      1.632884843688168e-05,
+      1.603205381875742e-05,
+      1.573525920063316e-05,
+      1.5438464582508908e-05,
+      1.5141669964384647e-05,
+      1.4844875346260385e-05,
+      1.4548080728136134e-05,
+      1.4251286110011875e-05,
+      1.3954491491887611e-05,
+      1.3657696873763363e-05,
+      1.33609022556391e-05,
+      1.298694103680253e-05,
+      1.2690146418678277e-05,
+      1.2393351800554018e-05,
+      1.2096557182429756e-05,
+      1.1799762564305506e-05,
+      1.1502967946181245e-05,
+      1.1206173328056984e-05,
+      1.0909378709932724e-05,
+      1.0612584091808472e-05,
+      1.0315789473684213e-05,
+      1.0018994855559949e-05,
+      9.7222002374357e-06,
+      9.425405619311438e-06,
+      9.128611001187177e-06,
+      8.831816383062916e-06,
+      8.535021764938666e-06,
+      8.238227146814405e-06,
+      7.941432528690143e-06,
+      7.644637910565893e-06,
+      7.347843292441632e-06,
+      7.0510486743173715e-06,
+      6.75425405619311e-06,
+      6.45745943806886e-06,
+      6.160664819944598e-06,
+      5.8638702018203375e-06,
+      5.567075583696086e-06,
+      5.270280965571826e-06,
+      4.973486347447565e-06,
+      4.676691729323314e-06,
+      4.379897111199053e-06,
+      4.083102493074792e-06
+    ],
+    "emb_lr": [],
+    "eval_step": [
+      1550,
+      3113,
+      4676,
+      6239,
+      7802,
+      9365,
+      10928,
+      12491,
+      14054,
+      15617
+    ],
+    "eval_accuracy": [
+      0.85,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0
+    ]
+  },
+  "final_accuracy": 1.0,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.996694214876033,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.93,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9881656804733728,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9831081081081081,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.82,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9692307692307692,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.9261083743842364,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 0.9914529914529915,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.58,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.8172043010752689,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.9013452914798207,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.39,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 0.64375,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.6644951140065146,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.22,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.32,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.4425,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.55,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.6,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.676,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.935,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.990625,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 0.9821428571428571,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.9810964083175804,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 0.9583333333333334,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.65,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 0.97,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8393782383419689,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.794392523364486,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.56,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.828125,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.7986111111111112,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.48,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8300653594771242,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.7422680412371134,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.72,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9207650273224044,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.9316239316239316,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.87,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9800332778702163,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 0.9494949494949495,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.989247311827957,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 0.993103448275862,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 0.9583333333333334,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.8,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 0.9882352941176471,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.8950276243093923,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.5,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.9902912621359223,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.7114093959731543,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 0.892018779342723,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.39,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.55,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.7333333333333333,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.54,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.54,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.695,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.915,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9933333333333333,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9887640449438202,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 0.9811320754716981,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9772209567198178,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.77,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8934010152284264,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.912621359223301,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.73,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.902834008097166,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.8888888888888888,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.6,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.889261744966443,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.8613861386138614,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.6966666666666667,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 4,
+      "mode": "sorl",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 1.0,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 1.0,
+  "sft_overall_accuracy": 0.6966666666666667
+}

add_sub_sorl_v1_abs70_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a22585d5e73e8ab5432b201cbfb9c017aa84e8d3422df85ec5407c5771999d1b
+size 650548580

add_sub_sorl_v1_abs70_100K/train_config.json

@@ -0,0 +1,84 @@
+{
+  "num_rollouts": 4,
+  "K": 4,
+  "max_iterations": 2,
+  "memory_span_abs": 1792,
+  "memory_span_traj": 1792,
+  "temperature": 1.0,
+  "ar_search": false,
+  "response_only_abs": false,
+  "alpha_info_gain": 10.0,
+  "alpha_abs": 0.1,
+  "alpha_soft_zipf": 1.0,
+  "alpha_ortho": 0.0,
+  "alpha_anchor": 0.0,
+  "alpha_jacobi": 0.0,
+  "decay": 0.8,
+  "target_vocab_util": 0.8,
+  "min_abs_ppl": 0.0,
+  "zipf_alpha": 1.0,
+  "lr": 4e-05,
+  "emb_lr_mult": 1.0,
+  "weight_decay": 0.01,
+  "warmup_steps": 468,
+  "cooldown_frac": 0.4,
+  "max_grad_norm": 1.0,
+  "vq_abs_pretrain_steps": 0,
+  "vq_abs_pretrain_lr": 0.001,
+  "vq_abs_pretrain_layer": -1,
+  "vq_abs_pretrain_batch_size": 256,
+  "vq_abs_pretrain_target_vectors": 20000,
+  "batch_size": 64,
+  "gradient_accumulation_steps": 1,
+  "num_epochs": 10,
+  "emb_warmup_steps": 0,
+  "log_every": 50,
+  "eval_every": 1562,
+  "save_every": 999999,
+  "eval_samples": 100,
+  "output_dir": "ckpt/sweep/as_sorl_abs70_K4_100K",
+  "eval_K": 4,
+  "alpha_traj": 0.0,
+  "corrupt_method": "shuffle",
+  "corrupt_ratio": 0.3,
+  "alpha_contrastive": 1.0,
+  "gamma_contrastive": 0.5,
+  "alpha_masked_traj": 0.0,
+  "mask_nl_ratio": 0.3,
+  "mask_nl_mode": "fixed",
+  "mask_nl_fixed_id": 0,
+  "use_ste": true,
+  "n_inner": 1,
+  "random_K": null,
+  "strip_suffix": null,
+  "compress_prefix": null,
+  "random_mem_span": null,
+  "warmup_ratio": 0.03,
+  "beta2": 0.999,
+  "seed": 42,
+  "n_digits": 6,
+  "n_layer": 2,
+  "n_head": 3,
+  "n_embd": 510,
+  "ops": "add_sub",
+  "abs_vocab": 70,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162560462,
+  "run_name": "add_sub_sorl_v1_abs70_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T18:34:15.781726+00:00",
+  "tokenizer": "Qwen/Qwen3-0.6B",
+  "dataset_repo": "thoughtworks/arithmetic-sorl-data",
+  "dataset_config": "add_sub_6digit",
+  "model_repo": "thoughtworks/arithmetic-sorl",
+  "trainer_version": "v1",
+  "wandb_run_id": "0azo3lkz",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/0azo3lkz",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.6966666666666667,
+  "eval_method": "ArithmeticEvaluator"
+}