Upload add_sub_sorl_v1_abs30_K3_50K

add_sub_sorl_v1_abs30_K3_50K/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": 151674
+}

add_sub_sorl_v1_abs30_K3_50K/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_abs30_K3_50K/metrics.json

@@ -0,0 +1,3477 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      832,
+      882,
+      932,
+      982,
+      1032,
+      1082,
+      1132,
+      1182,
+      1232,
+      1282,
+      1332,
+      1382,
+      1432,
+      1482,
+      1532,
+      1614,
+      1664,
+      1714,
+      1764,
+      1814,
+      1864,
+      1914,
+      1964,
+      2014,
+      2064,
+      2114,
+      2164,
+      2214,
+      2264,
+      2314,
+      2396,
+      2446,
+      2496,
+      2546,
+      2596,
+      2646,
+      2696,
+      2746,
+      2796,
+      2846,
+      2896,
+      2946,
+      2996,
+      3046,
+      3096,
+      3178,
+      3228,
+      3278,
+      3328,
+      3378,
+      3428,
+      3478,
+      3528,
+      3578,
+      3628,
+      3678,
+      3728,
+      3778,
+      3828,
+      3878,
+      3960,
+      4010,
+      4060,
+      4110,
+      4160,
+      4210,
+      4260,
+      4310,
+      4360,
+      4410,
+      4460,
+      4510,
+      4560,
+      4610,
+      4660,
+      4742,
+      4792,
+      4842,
+      4892,
+      4942,
+      4992,
+      5042,
+      5092,
+      5142,
+      5192,
+      5242,
+      5292,
+      5342,
+      5392,
+      5442,
+      5524,
+      5574,
+      5624,
+      5674,
+      5724,
+      5774,
+      5824,
+      5874,
+      5924,
+      5974,
+      6024,
+      6074,
+      6124,
+      6174,
+      6224,
+      6306,
+      6356,
+      6406,
+      6456,
+      6506,
+      6556,
+      6606,
+      6656,
+      6706,
+      6756,
+      6806,
+      6856,
+      6906,
+      6956,
+      7006,
+      7088,
+      7138,
+      7188,
+      7238,
+      7288,
+      7338,
+      7388,
+      7438,
+      7488,
+      7538,
+      7588,
+      7638,
+      7688,
+      7738,
+      7788,
+      7870,
+      7920,
+      7970,
+      8020,
+      8070,
+      8120,
+      8170,
+      8220,
+      8270,
+      8320,
+      8370,
+      8420,
+      8470,
+      8520,
+      8570,
+      8652,
+      8702,
+      8752,
+      8802,
+      8852,
+      8902,
+      8952,
+      9002,
+      9052,
+      9102,
+      9152,
+      9202,
+      9252,
+      9302,
+      9352,
+      9434,
+      9484,
+      9534,
+      9584,
+      9634,
+      9684,
+      9734,
+      9784,
+      9834,
+      9884,
+      9934,
+      9984,
+      10034,
+      10084,
+      10134,
+      10216,
+      10266,
+      10316,
+      10366,
+      10416,
+      10466,
+      10516,
+      10566,
+      10616,
+      10666,
+      10716,
+      10766,
+      10816,
+      10866,
+      10916,
+      10998,
+      11048,
+      11098,
+      11148,
+      11198,
+      11248,
+      11298,
+      11348,
+      11398,
+      11448,
+      11498,
+      11548,
+      11598,
+      11648,
+      11698,
+      11780,
+      11830,
+      11880,
+      11930,
+      11980,
+      12030,
+      12080,
+      12130,
+      12180,
+      12230,
+      12280,
+      12330,
+      12380,
+      12430,
+      12480,
+      12562,
+      12612,
+      12662,
+      12712,
+      12762,
+      12812,
+      12862,
+      12912,
+      12962,
+      13012,
+      13062,
+      13112,
+      13162,
+      13212,
+      13262,
+      13344,
+      13394,
+      13444,
+      13494,
+      13544,
+      13594,
+      13644,
+      13694,
+      13744,
+      13794,
+      13844,
+      13894,
+      13944,
+      13994,
+      14044,
+      14126,
+      14176,
+      14226,
+      14276,
+      14326,
+      14376,
+      14426,
+      14476,
+      14526,
+      14576,
+      14626,
+      14676,
+      14726,
+      14776,
+      14826,
+      14908,
+      14958,
+      15008,
+      15058,
+      15108,
+      15158,
+      15208,
+      15258,
+      15308,
+      15358,
+      15408,
+      15458,
+      15508,
+      15558,
+      15608
+    ],
+    "loss": [
+      13.509393692016602,
+      9.749732971191406,
+      5.9360880851745605,
+      3.8922414779663086,
+      2.9130501747131348,
+      2.5895895957946777,
+      2.3288536071777344,
+      1.7239751815795898,
+      1.3578572273254395,
+      0.6631852388381958,
+      0.7125436067581177,
+      0.22906304895877838,
+      0.11403137445449829,
+      -0.18299715220928192,
+      -0.1072545275092125,
+      -2.3244917392730713,
+      -9.210739135742188,
+      -10.258565902709961,
+      -11.842880249023438,
+      -12.723512649536133,
+      -13.822088241577148,
+      -13.786534309387207,
+      -14.823137283325195,
+      -13.667715072631836,
+      -15.485278129577637,
+      -14.717754364013672,
+      -14.689838409423828,
+      -14.371696472167969,
+      -14.705050468444824,
+      -14.5253324508667,
+      -14.996114730834961,
+      -14.544259071350098,
+      -14.397850036621094,
+      -15.307740211486816,
+      -14.620795249938965,
+      -14.343721389770508,
+      -14.886155128479004,
+      -14.080240249633789,
+      -14.099590301513672,
+      -14.724437713623047,
+      -14.093734741210938,
+      -13.550093650817871,
+      -11.328234672546387,
+      -10.802987098693848,
+      -8.087011337280273,
+      -7.161535263061523,
+      -6.76300048828125,
+      -5.766507148742676,
+      -5.0977959632873535,
+      -5.875992298126221,
+      -4.497377395629883,
+      -4.863527774810791,
+      -4.927530765533447,
+      -4.935651779174805,
+      -4.701904296875,
+      -4.331963539123535,
+      -3.767944574356079,
+      -3.393040418624878,
+      -3.2145793437957764,
+      -2.731534242630005,
+      -3.1932218074798584,
+      -4.5905680656433105,
+      -2.8137407302856445,
+      -3.041274309158325,
+      -2.6139461994171143,
+      -2.4471042156219482,
+      -2.4882166385650635,
+      -2.434157609939575,
+      -2.7672336101531982,
+      -2.69252610206604,
+      -2.0434446334838867,
+      -1.9935232400894165,
+      -3.210987091064453,
+      -2.203340530395508,
+      -1.61903977394104,
+      -1.8135666847229004,
+      -2.2731308937072754,
+      -1.7342028617858887,
+      -2.345428705215454,
+      -2.3986122608184814,
+      -2.6395304203033447,
+      -2.1111576557159424,
+      -1.9803094863891602,
+      -2.002279281616211,
+      -2.1121442317962646,
+      -2.3220841884613037,
+      -1.642890453338623,
+      -1.5122915506362915,
+      -1.352969765663147,
+      -1.790658950805664,
+      -2.0493268966674805,
+      -1.6816511154174805,
+      -1.5232160091400146,
+      -1.385091781616211,
+      -1.6739228963851929,
+      -3.1018519401550293,
+      -1.7351073026657104,
+      -2.2683258056640625,
+      -2.4009287357330322,
+      -1.8868049383163452,
+      -1.7282708883285522,
+      -1.1806395053863525,
+      -1.5342776775360107,
+      -1.2198463678359985,
+      -1.1997792720794678,
+      -1.6372817754745483,
+      -1.1737879514694214,
+      -1.8703954219818115,
+      -1.1837074756622314,
+      -2.0487072467803955,
+      -1.3988951444625854,
+      -1.3324480056762695,
+      -1.3330432176589966,
+      -1.4838277101516724,
+      -1.5514414310455322,
+      -1.4017307758331299,
+      -1.4976024627685547,
+      -1.3571642637252808,
+      -1.1720738410949707,
+      -1.2919104099273682,
+      -1.350249171257019,
+      -1.4729357957839966,
+      -1.1817578077316284,
+      -1.1564174890518188,
+      -1.5251946449279785,
+      -1.2372384071350098,
+      -0.932648777961731,
+      -1.0012640953063965,
+      -1.4444644451141357,
+      -1.2965625524520874,
+      -1.3345240354537964,
+      -1.3559203147888184,
+      -1.0545915365219116,
+      -1.1038869619369507,
+      -1.0856115818023682,
+      -1.0255887508392334,
+      -1.0998259782791138,
+      -0.7609472274780273,
+      -1.4960383176803589,
+      -1.0091618299484253,
+      -1.098103404045105,
+      -0.9137800335884094,
+      -0.836209237575531,
+      -0.7871055603027344,
+      -0.9814184308052063,
+      -0.8756022453308105,
+      -0.9116973280906677,
+      -0.782789945602417,
+      -1.2794203758239746,
+      -0.8758397698402405,
+      -1.1934887170791626,
+      -0.9208111763000488,
+      -1.0771613121032715,
+      -1.5408060550689697,
+      -1.2266724109649658,
+      -0.9050666689872742,
+      -0.5775320529937744,
+      -1.0902734994888306,
+      -1.2348861694335938,
+      -1.3116742372512817,
+      -0.7466331720352173,
+      -0.9078770875930786,
+      -0.7932072877883911,
+      -0.8755908608436584,
+      -1.191214680671692,
+      -0.6539024114608765,
+      -0.8499408960342407,
+      -0.9262394309043884,
+      -0.5675032138824463,
+      -0.45079314708709717,
+      -0.6070061326026917,
+      -1.1951838731765747,
+      -0.7711151838302612,
+      -0.9214723706245422,
+      -1.322747826576233,
+      -1.0038766860961914,
+      -0.21057185530662537,
+      -0.7360962629318237,
+      -1.047795295715332,
+      -0.5658537149429321,
+      -0.5452471971511841,
+      -0.6649935841560364,
+      -0.9057523012161255,
+      -0.9455522894859314,
+      -0.6777364015579224,
+      -0.6301040053367615,
+      -0.6875035762786865,
+      -0.7744766473770142,
+      -0.36955589056015015,
+      -0.7586586475372314,
+      -0.8452391624450684,
+      -0.6979129314422607,
+      -0.4924622178077698,
+      -0.5034284591674805,
+      -0.07909169793128967,
+      -0.4357263743877411,
+      -0.4921090304851532,
+      -0.6217997670173645,
+      -0.3099788427352905,
+      -0.3058575987815857,
+      -0.491129070520401,
+      -0.4106674790382385,
+      -0.5844934582710266,
+      -0.26459887623786926,
+      -0.4641355872154236,
+      -0.4777032434940338,
+      -0.5105655193328857,
+      -0.7228485941886902,
+      -0.37706509232521057,
+      -0.40531256794929504,
+      -0.3295072913169861,
+      -0.4547842741012573,
+      -0.4582812190055847,
+      -0.4903515577316284,
+      -0.4897080361843109,
+      -0.36798498034477234,
+      -0.6037724018096924,
+      -0.2460249364376068,
+      -0.3986009955406189,
+      -0.5716195702552795,
+      -0.23653608560562134,
+      -0.31190627813339233,
+      -0.44520020484924316,
+      -0.2482786774635315,
+      -0.47306254506111145,
+      -0.33344653248786926,
+      -0.39349013566970825,
+      -0.33977243304252625,
+      -0.3621232807636261,
+      -0.48065319657325745,
+      -0.30519622564315796,
+      -0.3248542249202728,
+      -0.30721086263656616,
+      -0.4892882704734802,
+      -0.5734624266624451,
+      -0.3485679030418396,
+      -0.2581605017185211,
+      -0.3928232491016388,
+      -0.18135161697864532,
+      -0.2859618067741394,
+      -0.1581522524356842,
+      -0.24521522223949432,
+      -0.40374138951301575,
+      -0.47672492265701294,
+      -0.1581224650144577,
+      -0.29607728123664856,
+      -0.04652285575866699,
+      -0.20653583109378815,
+      -0.2537737488746643,
+      -0.4445362985134125,
+      -0.1114339828491211,
+      -0.489991694688797,
+      -0.40431979298591614,
+      -0.5255370736122131,
+      -0.2891177833080292,
+      -0.13589264452457428,
+      -0.24943886697292328,
+      -0.1608649492263794,
+      -0.35039061307907104,
+      -0.21615290641784668,
+      -0.43050703406333923,
+      -0.25556042790412903,
+      -0.2582490146160126,
+      -0.42218878865242004,
+      -0.3240232765674591,
+      -0.12729018926620483,
+      -0.14628957211971283,
+      -0.14741162955760956,
+      -0.274014413356781,
+      -0.11761685460805893,
+      -0.1516052484512329,
+      -0.16077259182929993,
+      -0.2511594593524933,
+      -0.010212086141109467,
+      -0.2537863254547119,
+      -0.29016000032424927,
+      -0.16137613356113434,
+      -0.21936403214931488,
+      -0.14548173546791077,
+      -0.14406554400920868,
+      -0.32225245237350464,
+      -0.21201518177986145,
+      -0.25709688663482666,
+      -0.09402480721473694,
+      -0.14027932286262512,
+      -0.08345134556293488,
+      -0.19392192363739014,
+      0.007833808660507202,
+      -0.20100173354148865,
+      -0.14701750874519348,
+      -0.33041656017303467,
+      -0.16883714497089386,
+      -0.08166240155696869,
+      -0.15378451347351074,
+      -0.21001671254634857,
+      -0.1452471911907196,
+      -0.09268718957901001,
+      7.594376802444458e-05,
+      -0.16848143935203552,
+      -0.052198126912117004
+    ],
+    "base_loss": [
+      9.620177268981934,
+      7.274699687957764,
+      6.204272747039795,
+      4.166467189788818,
+      2.2501654624938965,
+      1.9623774290084839,
+      1.9064360857009888,
+      1.9552589654922485,
+      1.8678728342056274,
+      1.7965959310531616,
+      1.8250160217285156,
+      1.788840651512146,
+      1.8576834201812744,
+      1.8137224912643433,
+      1.8437130451202393,
+      1.8456766605377197,
+      1.989670991897583,
+      1.849730134010315,
+      1.9440898895263672,
+      1.8215068578720093,
+      1.811987280845642,
+      1.7906662225723267,
+      1.8050600290298462,
+      1.6350735425949097,
+      1.8098596334457397,
+      1.7310606241226196,
+      1.73845636844635,
+      1.7080748081207275,
+      1.6872161626815796,
+      1.6848351955413818,
+      1.7108510732650757,
+      1.658853530883789,
+      1.6322201490402222,
+      1.728717565536499,
+      1.6502704620361328,
+      1.6198476552963257,
+      1.6768041849136353,
+      1.6007717847824097,
+      1.577912449836731,
+      1.6503422260284424,
+      1.5837620496749878,
+      1.51720130443573,
+      1.2733173370361328,
+      1.2153533697128296,
+      0.9087010622024536,
+      0.8046143651008606,
+      0.760909914970398,
+      0.653200089931488,
+      0.576078474521637,
+      0.6617172956466675,
+      0.5092791318893433,
+      0.5486401319503784,
+      0.5964444875717163,
+      0.5582998394966125,
+      0.5341429114341736,
+      0.489955335855484,
+      0.4256270229816437,
+      0.3848516643047333,
+      0.36419087648391724,
+      0.31207147240638733,
+      0.36289140582084656,
+      0.5198935270309448,
+      0.31770092248916626,
+      0.3470384478569031,
+      0.3001234233379364,
+      0.28539514541625977,
+      0.28449442982673645,
+      0.27902907133102417,
+      0.3160552978515625,
+      0.3043105900287628,
+      0.23538364470005035,
+      0.22816064953804016,
+      0.3635883033275604,
+      0.2505258023738861,
+      0.18705177307128906,
+      0.2077564001083374,
+      0.25986963510513306,
+      0.2002851516008377,
+      0.2696126401424408,
+      0.27336403727531433,
+      0.29874229431152344,
+      0.24122998118400574,
+      0.22744552791118622,
+      0.22859355807304382,
+      0.24051980674266815,
+      0.2640826106071472,
+      0.1906505525112152,
+      0.17343030869960785,
+      0.15632306039333344,
+      0.20517291128635406,
+      0.23487980663776398,
+      0.1935703456401825,
+      0.17757008969783783,
+      0.16017813980579376,
+      0.1929638385772705,
+      0.3533931374549866,
+      0.19844293594360352,
+      0.25687453150749207,
+      0.27493026852607727,
+      0.21510256826877594,
+      0.1985342800617218,
+      0.1377100944519043,
+      0.17895111441612244,
+      0.1440882384777069,
+      0.13970573246479034,
+      0.18795418739318848,
+      0.142202690243721,
+      0.21577799320220947,
+      0.13755567371845245,
+      0.23404911160469055,
+      0.16638506948947906,
+      0.15567395091056824,
+      0.1552688181400299,
+      0.1740662306547165,
+      0.18159958720207214,
+      0.1600363701581955,
+      0.1719866544008255,
+      0.15725398063659668,
+      0.13674183189868927,
+      0.15326379239559174,
+      0.15582379698753357,
+      0.16946016252040863,
+      0.135700523853302,
+      0.13945254683494568,
+      0.17806246876716614,
+      0.14694233238697052,
+      0.12567099928855896,
+      0.11750459671020508,
+      0.16599784791469574,
+      0.15093906223773956,
+      0.15538986027240753,
+      0.15526553988456726,
+      0.12194647639989853,
+      0.1285392940044403,
+      0.1254655122756958,
+      0.12065047770738602,
+      0.12589503824710846,
+      0.09028182923793793,
+      0.17123977839946747,
+      0.1230500265955925,
+      0.12619636952877045,
+      0.10803552716970444,
+      0.09788729250431061,
+      0.09319708496332169,
+      0.11327429860830307,
+      0.10097881406545639,
+      0.10638836771249771,
+      0.09176915138959885,
+      0.14741040766239166,
+      0.10740412771701813,
+      0.1361636370420456,
+      0.12075847387313843,
+      0.12333475798368454,
+      0.17989730834960938,
+      0.14102159440517426,
+      0.11652173101902008,
+      0.06868256628513336,
+      0.12516002357006073,
+      0.141461580991745,
+      0.15087445080280304,
+      0.08888103067874908,
+      0.1065240427851677,
+      0.09194660931825638,
+      0.10156156122684479,
+      0.14206565916538239,
+      0.0768805667757988,
+      0.10412381589412689,
+      0.11042814701795578,
+      0.10985857993364334,
+      0.0573594756424427,
+      0.07315342873334885,
+      0.13767044246196747,
+      0.09264820069074631,
+      0.10909143835306168,
+      0.15286903083324432,
+      0.12395979464054108,
+      0.0777847170829773,
+      0.09892841428518295,
+      0.12121900171041489,
+      0.06989651918411255,
+      0.06681704521179199,
+      0.07782609015703201,
+      0.11105065792798996,
+      0.10854284465312958,
+      0.07878432422876358,
+      0.07466436922550201,
+      0.08287449926137924,
+      0.09146697074174881,
+      0.04641620069742203,
+      0.08733348548412323,
+      0.09853975474834442,
+      0.0975751206278801,
+      0.06128227710723877,
+      0.061950184404850006,
+      0.06638076901435852,
+      0.052723176777362823,
+      0.057864073663949966,
+      0.07467615604400635,
+      0.08593636751174927,
+      0.041386980563402176,
+      0.05814815312623978,
+      0.050658438354730606,
+      0.06955800205469131,
+      0.04857265204191208,
+      0.0566335991024971,
+      0.05846080556511879,
+      0.0625414326786995,
+      0.08379863202571869,
+      0.04710562154650688,
+      0.048435620963573456,
+      0.04122187942266464,
+      0.053831856697797775,
+      0.05445948988199234,
+      0.05889381840825081,
+      0.058188822120428085,
+      0.04711601510643959,
+      0.07027188688516617,
+      0.030178751796483994,
+      0.04915710538625717,
+      0.06689925491809845,
+      0.03054959885776043,
+      0.04069678112864494,
+      0.05244181677699089,
+      0.03186619281768799,
+      0.05536025017499924,
+      0.043754756450653076,
+      0.04784201830625534,
+      0.042507562786340714,
+      0.06403299421072006,
+      0.05792354419827461,
+      0.03965076431632042,
+      0.040603041648864746,
+      0.03727508336305618,
+      0.05885818973183632,
+      0.06674869358539581,
+      0.0428953543305397,
+      0.03251257166266441,
+      0.0478118360042572,
+      0.023396877571940422,
+      0.03521552309393883,
+      0.02183755859732628,
+      0.030320053920149803,
+      0.04852340742945671,
+      0.05709812045097351,
+      0.020526060834527016,
+      0.03813812509179115,
+      0.02402515895664692,
+      0.02654096856713295,
+      0.032051630318164825,
+      0.054777830839157104,
+      0.018536703661084175,
+      0.05893069505691528,
+      0.0486418791115284,
+      0.06543079763650894,
+      0.035314351320266724,
+      0.018703743815422058,
+      0.03151312842965126,
+      0.023690471425652504,
+      0.04184756055474281,
+      0.028494222089648247,
+      0.04992935433983803,
+      0.030887356027960777,
+      0.03226137161254883,
+      0.050099123269319534,
+      0.0390518493950367,
+      0.017795631662011147,
+      0.019216744229197502,
+      0.022917110472917557,
+      0.03587643429636955,
+      0.016961442306637764,
+      0.021900543943047523,
+      0.021325627341866493,
+      0.030909908935427666,
+      0.007513721007853746,
+      0.03314381465315819,
+      0.03860088065266609,
+      0.021070923656225204,
+      0.02842792496085167,
+      0.020011277869343758,
+      0.02126014418900013,
+      0.039515621960163116,
+      0.03040228597819805,
+      0.032819073647260666,
+      0.01422138325870037,
+      0.017915237694978714,
+      0.01341172493994236,
+      0.02505233883857727,
+      0.00482439249753952,
+      0.026765646412968636,
+      0.02130223624408245,
+      0.04042539373040199,
+      0.02248985692858696,
+      0.01175006479024887,
+      0.020824184641242027,
+      0.02868003584444523,
+      0.018705619499087334,
+      0.014115615747869015,
+      0.004133924841880798,
+      0.022322585806250572,
+      0.008793945424258709
+    ],
+    "info_loss": [
+      -0.3039093017578125,
+      -0.25568628311157227,
+      -0.31884098052978516,
+      -0.19432926177978516,
+      -0.054767608642578125,
+      -0.049605607986450195,
+      -0.04014873504638672,
+      -0.05167961120605469,
+      -0.07442402839660645,
+      -0.13443315029144287,
+      -0.12893474102020264,
+      -0.17337191104888916,
+      -0.19108903408050537,
+      -0.21625876426696777,
+      -0.20988011360168457,
+      -0.4322943687438965,
+      -1.1377909183502197,
+      -1.229161024093628,
+      -1.3951693773269653,
+      -1.4703682661056519,
+      -1.580564260482788,
+      -1.5719064474105835,
+      -1.6765954494476318,
+      -1.5430896282196045,
+      -1.7418283224105835,
+      -1.6556910276412964,
+      -1.6543322801589966,
+      -1.6190367937088013,
+      -1.650298833847046,
+      -1.633048415184021,
+      -1.6830683946609497,
+      -1.6315804719924927,
+      -1.6118990182876587,
+      -1.7115004062652588,
+      -1.6398777961730957,
+      -1.6054381132125854,
+      -1.6660842895507812,
+      -1.5763437747955322,
+      -1.575171947479248,
+      -1.6467103958129883,
+      -1.5790048837661743,
+      -1.514373540878296,
+      -1.2676775455474854,
+      -1.2094448804855347,
+      -0.9074189066886902,
+      -0.8032690286636353,
+      -0.7583819031715393,
+      -0.6490398645401001,
+      -0.5744845867156982,
+      -0.6607133150100708,
+      -0.5075458884239197,
+      -0.5478048920631409,
+      -0.5586112141609192,
+      -0.5572612881660461,
+      -0.5309542417526245,
+      -0.48941725492477417,
+      -0.4249153137207031,
+      -0.3829994797706604,
+      -0.3637515902519226,
+      -0.31148210167884827,
+      -0.3624016046524048,
+      -0.5192978978157043,
+      -0.3173016309738159,
+      -0.34558427333831787,
+      -0.2973107695579529,
+      -0.28006038069725037,
+      -0.28304430842399597,
+      -0.27716898918151855,
+      -0.31370648741722107,
+      -0.3037734031677246,
+      -0.23494072258472443,
+      -0.22656802833080292,
+      -0.3627624213695526,
+      -0.25011780858039856,
+      -0.18683567643165588,
+      -0.20739343762397766,
+      -0.2596602737903595,
+      -0.19889728724956512,
+      -0.26800841093063354,
+      -0.27274608612060547,
+      -0.2986396551132202,
+      -0.24110020697116852,
+      -0.22729624807834625,
+      -0.2283105105161667,
+      -0.2394106090068817,
+      -0.2636500298976898,
+      -0.1884169727563858,
+      -0.1727910190820694,
+      -0.15604889392852783,
+      -0.20469872653484344,
+      -0.2344672530889511,
+      -0.19138728082180023,
+      -0.1769782155752182,
+      -0.160061314702034,
+      -0.19289621710777283,
+      -0.35286980867385864,
+      -0.1973426192998886,
+      -0.2565523087978363,
+      -0.27472347021102905,
+      -0.2148154079914093,
+      -0.1984354555606842,
+      -0.1368931382894516,
+      -0.17813047766685486,
+      -0.14283229410648346,
+      -0.1391829252243042,
+      -0.1878795474767685,
+      -0.13922472298145294,
+      -0.21557334065437317,
+      -0.137115016579628,
+      -0.23376700282096863,
+      -0.16109754145145416,
+      -0.15478913486003876,
+      -0.15512211620807648,
+      -0.17280195653438568,
+      -0.18132664263248444,
+      -0.1594957411289215,
+      -0.17192837595939636,
+      -0.15713313221931458,
+      -0.1366598904132843,
+      -0.1502690315246582,
+      -0.15552976727485657,
+      -0.16888968646526337,
+      -0.1356629729270935,
+      -0.13934867084026337,
+      -0.17790979146957397,
+      -0.14454492926597595,
+      -0.11023528128862381,
+      -0.11658834666013718,
+      -0.16530659794807434,
+      -0.1505998969078064,
+      -0.15333877503871918,
+      -0.15518233180046082,
+      -0.12159797549247742,
+      -0.12795329093933105,
+      -0.1253867745399475,
+      -0.12043915688991547,
+      -0.125812828540802,
+      -0.09025537967681885,
+      -0.17112232744693756,
+      -0.11628134548664093,
+      -0.12609297037124634,
+      -0.10718151926994324,
+      -0.09783569723367691,
+      -0.09083214402198792,
+      -0.11275403201580048,
+      -0.10062570124864578,
+      -0.10635444521903992,
+      -0.09162037819623947,
+      -0.1473710536956787,
+      -0.10226716101169586,
+      -0.13601011037826538,
+      -0.11056432127952576,
+      -0.12312618643045425,
+      -0.1776789426803589,
+      -0.14076223969459534,
+      -0.10711970180273056,
+      -0.06863991916179657,
+      -0.1251138150691986,
+      -0.14144030213356018,
+      -0.15062983334064484,
+      -0.08860492706298828,
+      -0.10636203736066818,
+      -0.09127325564622879,
+      -0.10154711455106735,
+      -0.13964059948921204,
+      -0.07681436836719513,
+      -0.09958827495574951,
+      -0.10726893693208694,
+      -0.0754103809595108,
+      -0.05706685781478882,
+      -0.07299036532640457,
+      -0.13755835592746735,
+      -0.09162119776010513,
+      -0.10907827317714691,
+      -0.15164072811603546,
+      -0.11811167746782303,
+      -0.03249345347285271,
+      -0.0903124213218689,
+      -0.12119682878255844,
+      -0.06970714777708054,
+      -0.06679243594408035,
+      -0.07746848464012146,
+      -0.1049945279955864,
+      -0.10831177979707718,
+      -0.07852256298065186,
+      -0.07461845874786377,
+      -0.0801382064819336,
+      -0.09072548896074295,
+      -0.0461912527680397,
+      -0.08726274222135544,
+      -0.0985114574432373,
+      -0.08456294983625412,
+      -0.06125200167298317,
+      -0.06186487525701523,
+      -0.020790398120880127,
+      -0.052596382796764374,
+      -0.057847101241350174,
+      -0.07461639493703842,
+      -0.04391394183039665,
+      -0.04002353176474571,
+      -0.05813704803586006,
+      -0.050615258514881134,
+      -0.06931646913290024,
+      -0.03439071774482727,
+      -0.056624509394168854,
+      -0.058373335748910904,
+      -0.06252697855234146,
+      -0.08378857374191284,
+      -0.04694179818034172,
+      -0.048417072743177414,
+      -0.041149038821458817,
+      -0.05321168899536133,
+      -0.05434707552194595,
+      -0.05887436121702194,
+      -0.05817076563835144,
+      -0.04709482938051224,
+      -0.07026635110378265,
+      -0.03015829622745514,
+      -0.04905325546860695,
+      -0.06654560565948486,
+      -0.030544336885213852,
+      -0.04069141298532486,
+      -0.05243115499615669,
+      -0.03185577690601349,
+      -0.05531829595565796,
+      -0.043741751462221146,
+      -0.047834210097789764,
+      -0.04247414320707321,
+      -0.04620106518268585,
+      -0.05786474794149399,
+      -0.039284951984882355,
+      -0.04059477895498276,
+      -0.037263140082359314,
+      -0.05885086953639984,
+      -0.06672535091638565,
+      -0.0428934171795845,
+      -0.03250708058476448,
+      -0.047808755189180374,
+      -0.023388231173157692,
+      -0.03517588973045349,
+      -0.021835356950759888,
+      -0.030310988426208496,
+      -0.048488788306713104,
+      -0.057092029601335526,
+      -0.020513668656349182,
+      -0.03813441842794418,
+      -0.009540590457618237,
+      -0.026532385498285294,
+      -0.032044872641563416,
+      -0.054148320108652115,
+      -0.01852763630449772,
+      -0.05834845080971718,
+      -0.04863329231739044,
+      -0.061629801988601685,
+      -0.035191670060157776,
+      -0.018583303317427635,
+      -0.031511276960372925,
+      -0.023687515407800674,
+      -0.04183052107691765,
+      -0.02848932333290577,
+      -0.04992211237549782,
+      -0.03088497184216976,
+      -0.03225800395011902,
+      -0.050096165388822556,
+      -0.03905072063207626,
+      -0.017788220196962357,
+      -0.01920984499156475,
+      -0.022910727187991142,
+      -0.035870496183633804,
+      -0.016952384263277054,
+      -0.02188054285943508,
+      -0.02132139354944229,
+      -0.030908163636922836,
+      -0.006695533636957407,
+      -0.03312639519572258,
+      -0.03859920799732208,
+      -0.021069664508104324,
+      -0.028425302356481552,
+      -0.01996333710849285,
+      -0.02125830017030239,
+      -0.03950563073158264,
+      -0.030400089919567108,
+      -0.032812416553497314,
+      -0.014219382777810097,
+      -0.017884526401758194,
+      -0.013407579623162746,
+      -0.025049446150660515,
+      -0.00478351628407836,
+      -0.026763346046209335,
+      -0.021270256489515305,
+      -0.04042299464344978,
+      -0.022487716749310493,
+      -0.011748102493584156,
+      -0.020802181214094162,
+      -0.028673797845840454,
+      -0.01862839236855507,
+      -0.014111454598605633,
+      -0.004131872672587633,
+      -0.022314749658107758,
+      -0.008790398947894573
+    ],
+    "abs_loss": [
+      3.3947887420654297,
+      3.355994701385498,
+      3.1868395805358887,
+      2.9957025051116943,
+      2.636652708053589,
+      2.5180301666259766,
+      1.9289097785949707,
+      1.1535060405731201,
+      1.0622076988220215,
+      1.2105638980865479,
+      1.2612888813018799,
+      1.1270835399627686,
+      1.164677381515503,
+      1.1221411228179932,
+      1.0970821380615234,
+      0.787369966506958,
+      1.1733012199401855,
+      1.088056206703186,
+      1.086652398109436,
+      1.0431888103485107,
+      0.8837997913360596,
+      0.9671907424926758,
+      0.8009629249572754,
+      0.7664799690246582,
+      0.7298077344894409,
+      0.5918371677398682,
+      0.5854489803314209,
+      0.6336506009101868,
+      0.7560532093048096,
+      0.4996872842311859,
+      0.4896530508995056,
+      0.5172319412231445,
+      0.5553824305534363,
+      0.5992914438247681,
+      0.4079827666282654,
+      0.6127545833587646,
+      0.5886710286140442,
+      0.4055051803588867,
+      0.5210465788841248,
+      0.4203464388847351,
+      0.4285019338130951,
+      0.47940152883529663,
+      0.4155282974243164,
+      0.439352422952652,
+      0.4332502782344818,
+      0.5129095315933228,
+      0.4495353698730469,
+      0.3472867012023926,
+      0.40901464223861694,
+      0.3672488331794739,
+      0.3482757806777954,
+      0.37994515895843506,
+      0.43616002798080444,
+      0.3600550591945648,
+      0.3320258855819702,
+      0.355334997177124,
+      0.289895236492157,
+      0.2699585556983948,
+      0.26594990491867065,
+      0.2789926528930664,
+      0.28545188903808594,
+      0.24128937721252441,
+      0.21557548642158508,
+      0.29902154207229614,
+      0.29513996839523315,
+      0.2261316329240799,
+      0.23958441615104675,
+      0.31923651695251465,
+      0.27118515968322754,
+      0.23833268880844116,
+      0.17458440363407135,
+      0.2082197368144989,
+      0.25783029198646545,
+      0.23325616121292114,
+      0.28747764229774475,
+      0.2220531404018402,
+      0.19746264815330505,
+      0.2527940273284912,
+      0.20642422139644623,
+      0.18559518456459045,
+      0.20667535066604614,
+      0.23093782365322113,
+      0.24432438611984253,
+      0.20482295751571655,
+      0.17530974745750427,
+      0.25710201263427734,
+      0.24117588996887207,
+      0.21339458227157593,
+      0.18219798803329468,
+      0.29019445180892944,
+      0.2061009705066681,
+      0.2884983420372009,
+      0.15619494020938873,
+      0.19746707379817963,
+      0.15183740854263306,
+      0.24942252039909363,
+      0.16572964191436768,
+      0.21800562739372253,
+      0.18709537386894226,
+      0.18629208207130432,
+      0.22767946124076843,
+      0.1663786768913269,
+      0.2522010803222656,
+      0.1544596552848816,
+      0.1796630322933197,
+      0.16058242321014404,
+      0.13038626313209534,
+      0.15064063668251038,
+      0.1634083390235901,
+      0.14795365929603577,
+      0.11809875071048737,
+      0.16260576248168945,
+      0.1509130597114563,
+      0.14089354872703552,
+      0.1798515021800995,
+      0.1337878406047821,
+      0.16343703866004944,
+      0.14893639087677002,
+      0.13968947529792786,
+      0.15690213441848755,
+      0.15960702300071716,
+      0.10107538104057312,
+      0.10875316709280014,
+      0.17253854870796204,
+      0.13098755478858948,
+      0.10898811370134354,
+      0.11676128953695297,
+      0.15887998044490814,
+      0.06520901620388031,
+      0.08105364441871643,
+      0.18425777554512024,
+      0.08747240155935287,
+      0.11760591715574265,
+      0.05891865864396095,
+      0.09311813116073608,
+      0.10404924303293228,
+      0.10930569469928741,
+      0.07826673984527588,
+      0.08597943186759949,
+      0.09667380154132843,
+      0.08121703565120697,
+      0.14723467826843262,
+      0.05955659598112106,
+      0.0923231914639473,
+      0.07016702741384506,
+      0.09162533283233643,
+      0.07628462463617325,
+      0.0766228511929512,
+      0.09439100325107574,
+      0.08760590851306915,
+      0.0737878605723381,
+      0.06837128102779388,
+      0.08421748876571655,
+      0.0957871600985527,
+      0.07666170597076416,
+      0.1123364269733429,
+      0.09760688245296478,
+      0.07807846367359161,
+      0.06708869338035583,
+      0.08171576261520386,
+      0.06592215597629547,
+      0.08457791060209274,
+      0.06967782974243164,
+      0.0786462351679802,
+      0.06119159981608391,
+      0.06770946830511093,
+      0.07555916905403137,
+      0.09257392585277557,
+      0.07516339421272278,
+      0.07899697124958038,
+      0.07565949112176895,
+      0.10470855236053467,
+      0.04123685508966446,
+      0.04719582572579384,
+      0.057023901492357254,
+      0.06000470742583275,
+      0.050972916185855865,
+      0.026791758835315704,
+      0.05733746662735939,
+      0.057162001729011536,
+      0.06315764784812927,
+      0.06523626297712326,
+      0.06238469481468201,
+      0.056802794337272644,
+      0.03760264068841934,
+      0.057449936866760254,
+      0.056212738156318665,
+      0.06407733261585236,
+      0.05097174644470215,
+      0.05976347252726555,
+      0.09436068683862686,
+      0.04945018142461777,
+      0.11757603287696838,
+      0.07025796175003052,
+      0.056657664477825165,
+      0.050152845680713654,
+      0.053100623190402985,
+      0.06337989121675491,
+      0.052439115941524506,
+      0.029815109446644783,
+      0.05111522600054741,
+      0.04386456310749054,
+      0.04625169187784195,
+      0.05024842917919159,
+      0.07870125025510788,
+      0.057452261447906494,
+      0.04711850360035896,
+      0.03932467848062515,
+      0.040417902171611786,
+      0.044037483632564545,
+      0.038460854440927505,
+      0.0472777783870697,
+      0.030307834967970848,
+      0.03239193931221962,
+      0.041062939912080765,
+      0.028412584215402603,
+      0.04923238605260849,
+      0.02583213895559311,
+      0.03357554227113724,
+      0.045960620045661926,
+      0.032667823135852814,
+      0.0440668985247612,
+      0.04598768800497055,
+      0.020485155284404755,
+      0.03069138713181019,
+      0.09542465955018997,
+      0.04972872883081436,
+      0.04988779500126839,
+      0.06253587454557419,
+      0.041065312922000885,
+      0.056552380323410034,
+      0.03858733922243118,
+      0.0346544086933136,
+      0.04244701936841011,
+      0.04882288724184036,
+      0.02505997195839882,
+      0.08355213701725006,
+      0.04368632659316063,
+      0.02518036775290966,
+      0.03222693130373955,
+      0.02847285196185112,
+      0.028555555269122124,
+      0.04500667750835419,
+      0.039119310677051544,
+      0.042069531977176666,
+      0.04982626438140869,
+      0.06467273831367493,
+      0.05387020483613014,
+      0.044043101370334625,
+      0.04368911683559418,
+      0.04396820440888405,
+      0.059396810829639435,
+      0.04429316520690918,
+      0.03188247233629227,
+      0.0384872704744339,
+      0.04625459387898445,
+      0.03219570964574814,
+      0.0690075159072876,
+      0.038136277347803116,
+      0.046186305582523346,
+      0.0670032873749733,
+      0.03440630063414574,
+      0.04002414643764496,
+      0.0336119569838047,
+      0.027710601687431335,
+      0.032281503081321716,
+      0.024941034615039825,
+      0.03399371728301048,
+      0.043286774307489395,
+      0.031762830913066864,
+      0.06252194941043854,
+      0.03299562633037567,
+      0.03929908946156502,
+      0.018995758146047592,
+      0.06292541325092316,
+      0.035221878439188004,
+      0.03341846168041229,
+      0.04938807711005211,
+      0.027772381901741028,
+      0.023751210421323776,
+      0.040190670639276505,
+      0.03436459228396416,
+      0.03515885770320892,
+      0.0373615026473999,
+      0.021354738622903824,
+      0.04102501645684242,
+      0.036710917949676514,
+      0.036844585090875626,
+      0.051790397614240646,
+      0.026980645954608917,
+      0.04401412978768349,
+      0.038550522178411484,
+      0.02942569926381111,
+      0.04987340420484543,
+      0.03775779530405998,
+      0.02882983535528183,
+      0.05363527312874794,
+      0.027315087616443634,
+      0.020971231162548065,
+      0.035575009882450104
+    ],
+    "zipf_loss": [
+      6.58882999420166,
+      4.696296691894531,
+      2.60154128074646,
+      1.3694965839385986,
+      0.9468953609466553,
+      0.8714652061462402,
+      0.6310137510299683,
+      0.1701616644859314,
+      0.12800396978855133,
+      0.08986442536115646,
+      0.050746090710163116,
+      0.061233147978782654,
+      0.05077056214213371,
+      0.053653884679079056,
+      0.03812534362077713,
+      0.07403817772865295,
+      0.06016801297664642,
+      0.0745084285736084,
+      0.056058380752801895,
+      0.05434504151344299,
+      0.08318846672773361,
+      0.045144904404878616,
+      0.05766122043132782,
+      0.05146005004644394,
+      0.050165221095085144,
+      0.048910077661275864,
+      0.056483615189790726,
+      0.047232866287231445,
+      0.03511703759431839,
+      0.070347361266613,
+      0.07475403696298599,
+      0.0609690397977829,
+      0.0333818756043911,
+      0.018616575747728348,
+      0.08691272884607315,
+      0.029536176472902298,
+      0.03901568427681923,
+      0.04187442362308502,
+      0.02211182191967964,
+      0.05028867721557617,
+      0.06970130652189255,
+      0.028500426560640335,
+      0.033670824021101,
+      0.03217380493879318,
+      0.03515150398015976,
+      0.01524924673140049,
+      0.014954814687371254,
+      0.03596249967813492,
+      0.030069705098867416,
+      0.03269797936081886,
+      0.03397458419203758,
+      0.027886394411325455,
+      0.01852090284228325,
+      0.04265565052628517,
+      0.04029293358325958,
+      0.03672034293413162,
+      0.02659197896718979,
+      0.02510666288435459,
+      0.03215096890926361,
+      0.04331589862704277,
+      0.03935756906867027,
+      0.05838822200894356,
+      0.020017120987176895,
+      0.03762765973806381,
+      0.029524000361561775,
+      0.04549149423837662,
+      0.03377367556095123,
+      0.02657960169017315,
+      0.026657648384571075,
+      0.017064208164811134,
+      0.05312030762434006,
+      0.02317437157034874,
+      0.027265695855021477,
+      0.023985985666513443,
+      0.03351731225848198,
+      0.030405916273593903,
+      0.04385584592819214,
+      0.029205456376075745,
+      0.04440043494105339,
+      0.036925189197063446,
+      0.02745618112385273,
+      0.03552067652344704,
+      0.04077502712607384,
+      0.03174997866153717,
+      0.02391131781041622,
+      0.02462334930896759,
+      0.026511013507843018,
+      0.020848870277404785,
+      0.032976262271404266,
+      0.02213592827320099,
+      0.03985583782196045,
+      0.00980145763605833,
+      0.053376540541648865,
+      0.03559654951095581,
+      0.04689183831214905,
+      0.04851072281599045,
+      0.023303033784031868,
+      0.018522314727306366,
+      0.05266643315553665,
+      0.02761736884713173,
+      0.03478141129016876,
+      0.03394386172294617,
+      0.04285586625337601,
+      0.04894232004880905,
+      0.034377939999103546,
+      0.03750133141875267,
+      0.06321785598993301,
+      0.054495833814144135,
+      0.03354610502719879,
+      0.040118470788002014,
+      0.033885374665260315,
+      0.043508730828762054,
+      0.04781779646873474,
+      0.05603636056184769,
+      0.06224025413393974,
+      0.019811511039733887,
+      0.033350951969623566,
+      0.042019397020339966,
+      0.04381423816084862,
+      0.041825853288173676,
+      0.03326395899057388,
+      0.03639336675405502,
+      0.02829602360725403,
+      0.08036279678344727,
+      0.06274212151765823,
+      0.050369687378406525,
+      0.03235689923167229,
+      0.03122669830918312,
+      0.03608278930187225,
+      0.05039194971323013,
+      0.0250480305403471,
+      0.03189028054475784,
+      0.02768118493258953,
+      0.04121473804116249,
+      0.03347885236144066,
+      0.04774739593267441,
+      0.02147674560546875,
+      0.04349803924560547,
+      0.03534719720482826,
+      0.020934192463755608,
+      0.028508242219686508,
+      0.035276252776384354,
+      0.03830478712916374,
+      0.018786411732435226,
+      0.025830931961536407,
+      0.020513445138931274,
+      0.03783036023378372,
+      0.03398241102695465,
+      0.03744061663746834,
+      0.030667103826999664,
+      0.02306997776031494,
+      0.05723650008440018,
+      0.022343957796692848,
+      0.0465073361992836,
+      0.03226206824183464,
+      0.038375016301870346,
+      0.030423959717154503,
+      0.027896907180547714,
+      0.03134632855653763,
+      0.03557799756526947,
+      0.0439428985118866,
+      0.04076138138771057,
+      0.020610880106687546,
+      0.03045407496392727,
+      0.05700647830963135,
+      0.030589675530791283,
+      0.03426211699843407,
+      0.026764456182718277,
+      0.06922566890716553,
+      0.054616279900074005,
+      0.04217809811234474,
+      0.032258354127407074,
+      0.04832490161061287,
+      0.0554993562400341,
+      0.035088032484054565,
+      0.04727983847260475,
+      0.03148068115115166,
+      0.06542038917541504,
+      0.03722023218870163,
+      0.05560505762696266,
+      0.0495443157851696,
+      0.025341596454381943,
+      0.026903830468654633,
+      0.023342445492744446,
+      0.024944588541984558,
+      0.035671256482601166,
+      0.025382786989212036,
+      0.034903496503829956,
+      0.04084325209259987,
+      0.020658954977989197,
+      0.0318995900452137,
+      0.045196421444416046,
+      0.04701791703701019,
+      0.04624432325363159,
+      0.05676574632525444,
+      0.03249902278184891,
+      0.023187857121229172,
+      0.04335003346204758,
+      0.03798030689358711,
+      0.050009217113256454,
+      0.026981737464666367,
+      0.040440212935209274,
+      0.034488074481487274,
+      0.025710783898830414,
+      0.03760576993227005,
+      0.04182407259941101,
+      0.04745098203420639,
+      0.02730601467192173,
+      0.04120548814535141,
+      0.026018794625997543,
+      0.036915123462677,
+      0.018772978335618973,
+      0.027699263766407967,
+      0.036259058862924576,
+      0.02970455214381218,
+      0.05300603061914444,
+      0.02369600161910057,
+      0.02279607206583023,
+      0.039416879415512085,
+      0.022341210395097733,
+      0.035090915858745575,
+      0.04990437254309654,
+      0.022070778533816338,
+      0.03636438027024269,
+      0.021691016852855682,
+      0.05067375302314758,
+      0.03203708678483963,
+      0.037472669035196304,
+      0.029600756242871284,
+      0.035964228212833405,
+      0.04234728217124939,
+      0.036631789058446884,
+      0.024680007249116898,
+      0.036117538809776306,
+      0.022160109132528305,
+      0.034964922815561295,
+      0.026042522862553596,
+      0.03308381885290146,
+      0.026615764945745468,
+      0.02735886164009571,
+      0.035516466945409775,
+      0.02471904456615448,
+      0.028122467920184135,
+      0.03318537026643753,
+      0.02228119783103466,
+      0.04214617982506752,
+      0.01839062012732029,
+      0.026860028505325317,
+      0.03021903522312641,
+      0.03780020400881767,
+      0.05090884864330292,
+      0.02862248383462429,
+      0.028941920027136803,
+      0.022141888737678528,
+      0.02363584190607071,
+      0.026611194014549255,
+      0.030941203236579895,
+      0.04541895538568497,
+      0.02225341834127903,
+      0.03562748432159424,
+      0.012084392830729485,
+      0.018961315974593163,
+      0.02806723490357399,
+      0.025312531739473343,
+      0.024661026895046234,
+      0.02956821769475937,
+      0.02409803308546543,
+      0.055379170924425125,
+      0.04448545724153519,
+      0.03176925331354141,
+      0.03904742747545242,
+      0.02781613916158676,
+      0.023082362487912178,
+      0.04732995480298996,
+      0.03804126754403114,
+      0.0537090003490448,
+      0.024907726794481277,
+      0.03152225539088249,
+      0.03136312961578369,
+      0.044882193207740784,
+      0.029269149526953697,
+      0.05814700573682785,
+      0.03469230234622955,
+      0.03021147847175598,
+      0.01851522922515869,
+      0.03311022371053696,
+      0.027849100530147552,
+      0.04716012254357338,
+      0.03468704968690872,
+      0.04168476164340973,
+      0.028986569494009018,
+      0.029695110395550728,
+      0.02112598717212677,
+      0.028425775468349457,
+      0.04426546394824982,
+      0.019448116421699524,
+      0.028948213905096054,
+      0.03452923521399498,
+      0.030246354639530182,
+      0.02335442043840885
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.7164890170097351,
+      0.44001519680023193,
+      0.23450502753257751,
+      0.1798117458820343,
+      0.11821601539850235,
+      0.08064808696508408,
+      0.04913584515452385,
+      0.04442092031240463,
+      0.03780680522322655,
+      0.03147492557764053,
+      0.03255445510149002,
+      0.03306354209780693,
+      0.031127508729696274,
+      0.03149908781051636,
+      0.03147974982857704,
+      0.02890501357614994,
+      0.04472016915678978,
+      0.05679545924067497,
+      0.0668342337012291,
+      0.07476525753736496,
+      0.08515781164169312,
+      0.09092093259096146,
+      0.09147346019744873,
+      0.09923742711544037,
+      0.10627821832895279,
+      0.11134133487939835,
+      0.11911923438310623,
+      0.11903499811887741,
+      0.1265963613986969,
+      0.1247737854719162,
+      0.1380072683095932,
+      0.139429971575737,
+      0.14270682632923126,
+      0.1469593346118927,
+      0.1489012986421585,
+      0.1440083086490631,
+      0.14821738004684448,
+      0.14707738161087036,
+      0.15268000960350037,
+      0.15831196308135986,
+      0.16472434997558594,
+      0.1661268174648285,
+      0.17111200094223022,
+      0.17259204387664795,
+      0.17933180928230286,
+      0.1868675798177719,
+      0.19625088572502136,
+      0.20457005500793457,
+      0.21157294511795044,
+      0.2194136679172516,
+      0.22429223358631134,
+      0.23341265320777893,
+      0.23742900788784027,
+      0.23962612450122833,
+      0.23993082344532013,
+      0.25011855363845825,
+      0.2589331269264221,
+      0.26473522186279297,
+      0.2694035768508911,
+      0.272711843252182,
+      0.2754519283771515,
+      0.2818911373615265,
+      0.29113882780075073,
+      0.29514259099960327,
+      0.29763442277908325,
+      0.3000759184360504,
+      0.30084431171417236,
+      0.3064722716808319,
+      0.31723761558532715,
+      0.3191567659378052,
+      0.3253643810749054,
+      0.33208969235420227,
+      0.334012895822525,
+      0.3363591432571411,
+      0.33951324224472046,
+      0.34777316451072693,
+      0.34541478753089905,
+      0.33628666400909424,
+      0.3373204469680786,
+      0.34333908557891846,
+      0.34256234765052795,
+      0.34841620922088623,
+      0.3535518944263458,
+      0.35466042160987854,
+      0.3657434284687042,
+      0.36002078652381897,
+      0.3547402620315552,
+      0.3438624441623688,
+      0.34744876623153687,
+      0.35149863362312317,
+      0.35756582021713257,
+      0.3576912581920624,
+      0.3594180643558502,
+      0.3629657030105591,
+      0.3711114227771759,
+      0.3733091950416565,
+      0.3697828948497772,
+      0.3735300302505493,
+      0.37614548206329346,
+      0.3754543364048004,
+      0.37430867552757263,
+      0.3725973665714264,
+      0.3632843494415283,
+      0.36546799540519714,
+      0.3608647584915161,
+      0.3605231046676636,
+      0.36258602142333984,
+      0.3645479083061218,
+      0.3697208762168884,
+      0.3715536892414093,
+      0.37075474858283997,
+      0.3678865134716034,
+      0.36681437492370605,
+      0.36476776003837585,
+      0.37008997797966003,
+      0.37324196100234985,
+      0.3813824951648712,
+      0.38447922468185425,
+      0.3872334063053131,
+      0.3909793496131897,
+      0.39336153864860535,
+      0.390764981508255,
+      0.39270278811454773,
+      0.3944382965564728,
+      0.4021323025226593,
+      0.4101596176624298,
+      0.41367971897125244,
+      0.41102364659309387,
+      0.42576465010643005,
+      0.42502540349960327,
+      0.42283493280410767,
+      0.4223448634147644,
+      0.4292895197868347,
+      0.4299042224884033,
+      0.43099603056907654,
+      0.4349670708179474,
+      0.430941641330719,
+      0.4276216924190521,
+      0.4218938946723938,
+      0.43078887462615967,
+      0.43886658549308777,
+      0.4456350803375244,
+      0.4460766613483429,
+      0.4431578814983368,
+      0.44184592366218567,
+      0.44215628504753113,
+      0.4425334334373474,
+      0.44719398021698,
+      0.45072248578071594,
+      0.45337986946105957,
+      0.4500051438808441,
+      0.4455207288265228,
+      0.4497999846935272,
+      0.4499672055244446,
+      0.4508799910545349,
+      0.4514467418193817,
+      0.4518853425979614,
+      0.45511433482170105,
+      0.45494544506073,
+      0.45767542719841003,
+      0.4608568549156189,
+      0.4599456191062927,
+      0.45884469151496887,
+      0.4611905515193939,
+      0.45628681778907776,
+      0.45790180563926697,
+      0.456195205450058,
+      0.4569515585899353,
+      0.45791909098625183,
+      0.46010375022888184,
+      0.45950251817703247,
+      0.4618186950683594,
+      0.4617295265197754,
+      0.4598270058631897,
+      0.45474281907081604,
+      0.4536970853805542,
+      0.45817336440086365,
+      0.45483019948005676,
+      0.4549138844013214,
+      0.4594607651233673,
+      0.46500587463378906,
+      0.4646036624908447,
+      0.46177607774734497,
+      0.4655199646949768,
+      0.4650300145149231,
+      0.4626225531101227,
+      0.46814680099487305,
+      0.46539047360420227,
+      0.46605440974235535,
+      0.4705409109592438,
+      0.4698832929134369,
+      0.47260361909866333,
+      0.4737638235092163,
+      0.47194957733154297,
+      0.47447729110717773,
+      0.47706830501556396,
+      0.4774821698665619,
+      0.4845934510231018,
+      0.48296087980270386,
+      0.4830245077610016,
+      0.47902020812034607,
+      0.4795837998390198,
+      0.4817299246788025,
+      0.4865899384021759,
+      0.4874388575553894,
+      0.4878205955028534,
+      0.4869936406612396,
+      0.4862651824951172,
+      0.48645102977752686,
+      0.48484084010124207,
+      0.488354355096817,
+      0.4892062544822693,
+      0.48806798458099365,
+      0.48925110697746277,
+      0.49163854122161865,
+      0.49381622672080994,
+      0.49342846870422363,
+      0.4912415146827698,
+      0.4916841983795166,
+      0.4906071126461029,
+      0.4897904098033905,
+      0.49452245235443115,
+      0.4923728406429291,
+      0.4922875463962555,
+      0.49448657035827637,
+      0.4951864182949066,
+      0.49540138244628906,
+      0.4948039650917053,
+      0.4952405095100403,
+      0.49712300300598145,
+      0.4969750642776489,
+      0.49810075759887695,
+      0.4981243312358856,
+      0.4974529445171356,
+      0.499134361743927,
+      0.49833613634109497,
+      0.5001863837242126,
+      0.4997479021549225,
+      0.5002753138542175,
+      0.501311719417572,
+      0.5036635398864746,
+      0.5029648542404175,
+      0.5034725666046143,
+      0.5049551129341125,
+      0.5059987306594849,
+      0.5073911547660828,
+      0.5067667365074158,
+      0.5083858370780945,
+      0.5091679096221924,
+      0.5096098780632019,
+      0.5089676976203918,
+      0.5098400712013245,
+      0.5101442933082581,
+      0.5129626989364624,
+      0.5143404006958008,
+      0.5133976936340332,
+      0.5152299404144287,
+      0.5157557725906372,
+      0.5187842845916748,
+      0.5187036395072937,
+      0.5189111232757568,
+      0.519492506980896,
+      0.520779550075531,
+      0.5204878449440002,
+      0.5201342105865479,
+      0.5208142399787903,
+      0.5219997763633728,
+      0.5224781632423401,
+      0.5233966112136841,
+      0.5231385231018066,
+      0.5231932401657104,
+      0.5238833427429199,
+      0.5240063667297363,
+      0.5245592594146729,
+      0.5234469175338745,
+      0.524029552936554,
+      0.5236588716506958,
+      0.52469402551651,
+      0.5251315832138062,
+      0.5246515274047852,
+      0.5242875218391418,
+      0.5250592231750488,
+      0.5248522758483887,
+      0.5247886776924133,
+      0.5251317620277405,
+      0.5267933011054993,
+      0.5268675088882446,
+      0.5270206332206726,
+      0.5275089740753174,
+      0.5276087522506714,
+      0.5281514525413513,
+      0.5281208157539368,
+      0.5284842252731323,
+      0.528671383857727,
+      0.5284337401390076,
+      0.5283054113388062,
+      0.5284829139709473,
+      0.5284526944160461,
+      0.5282247066497803,
+      0.5278456211090088
+    ],
+    "lr": [
+      8.376068376068378e-06,
+      1.6923076923076924e-05,
+      2.5470085470085475e-05,
+      3.401709401709402e-05,
+      4.2564102564102564e-05,
+      5.1111111111111115e-05,
+      5.9658119658119666e-05,
+      6.820512820512821e-05,
+      7.675213675213675e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      8e-05,
+      7.98600052728711e-05,
+      7.92668072765621e-05,
+      7.86736092802531e-05,
+      7.808041128394411e-05,
+      7.748721328763512e-05,
+      7.689401529132612e-05,
+      7.630081729501713e-05,
+      7.570761929870815e-05,
+      7.511442130239916e-05,
+      7.452122330609016e-05,
+      7.392802530978118e-05,
+      7.333482731347219e-05,
+      7.236198259952545e-05,
+      7.176878460321646e-05,
+      7.117558660690747e-05,
+      7.058238861059848e-05,
+      6.998919061428949e-05,
+      6.939599261798048e-05,
+      6.88027946216715e-05,
+      6.820959662536251e-05,
+      6.761639862905352e-05,
+      6.702320063274453e-05,
+      6.643000263643554e-05,
+      6.583680464012655e-05,
+      6.524360664381755e-05,
+      6.465040864750859e-05,
+      6.405721065119958e-05,
+      6.308436593725284e-05,
+      6.249116794094385e-05,
+      6.189796994463487e-05,
+      6.130477194832587e-05,
+      6.0711573952016877e-05,
+      6.011837595570788e-05,
+      5.9525177959398896e-05,
+      5.8931979963089905e-05,
+      5.833878196678091e-05,
+      5.774558397047194e-05,
+      5.715238597416294e-05,
+      5.6559187977853943e-05,
+      5.5965989981544967e-05,
+      5.537279198523596e-05,
+      5.477959398892697e-05,
+      5.380674927498023e-05,
+      5.321355127867123e-05,
+      5.2620353282362255e-05,
+      5.202715528605326e-05,
+      5.143395728974426e-05,
+      5.084075929343529e-05,
+      5.024756129712629e-05,
+      4.9654363300817296e-05,
+      4.9061165304508305e-05,
+      4.846796730819932e-05,
+      4.787476931189032e-05,
+      4.7281571315581334e-05,
+      4.668837331927235e-05,
+      4.609517532296336e-05,
+      4.550197732665436e-05,
+      4.452913261270761e-05,
+      4.393593461639862e-05,
+      4.334273662008965e-05,
+      4.2749538623780655e-05,
+      4.215634062747166e-05,
+      4.156314263116268e-05,
+      4.0969944634853683e-05,
+      4.0376746638544686e-05,
+      3.978354864223571e-05,
+      3.919035064592671e-05,
+      3.8597152649617715e-05,
+      3.800395465330872e-05,
+      3.741075665699975e-05,
+      3.681755866069075e-05,
+      3.622436066438175e-05,
+      3.525151595043501e-05,
+      3.465831795412603e-05,
+      3.4065119957817036e-05,
+      3.347192196150804e-05,
+      3.287872396519904e-05,
+      3.2285525968890064e-05,
+      3.169232797258107e-05,
+      3.109912997627207e-05,
+      3.05059319799631e-05,
+      2.9912733983654103e-05,
+      2.9319535987345105e-05,
+      2.8726337991036132e-05,
+      2.8133139994727138e-05,
+      2.753994199841814e-05,
+      2.6946744002109144e-05,
+      2.5973899288162397e-05,
+      2.538070129185342e-05,
+      2.4787503295544423e-05,
+      2.419430529923543e-05,
+      2.3601107302926452e-05,
+      2.3007909306617458e-05,
+      2.241471131030846e-05,
+      2.1821513313999464e-05,
+      2.122831531769049e-05,
+      2.0635117321381496e-05,
+      2.00419193250725e-05,
+      1.944872132876352e-05,
+      1.8855523332454525e-05,
+      1.8262325336145528e-05,
+      1.766912733983655e-05,
+      1.6696282625889808e-05,
+      1.610308462958081e-05,
+      1.5509886633271817e-05,
+      1.491668863696282e-05,
+      1.4323490640653843e-05,
+      1.3730292644344847e-05,
+      1.3137094648035852e-05,
+      1.2543896651726875e-05,
+      1.1950698655417879e-05,
+      1.1357500659108882e-05,
+      1.0764302662799886e-05,
+      1.017110466649091e-05,
+      9.577906670181914e-06,
+      8.984708673872917e-06,
+      8.391510677563942e-06
+    ],
+    "emb_lr": [],
+    "eval_step": [
+      750,
+      1532,
+      2314,
+      3096,
+      3878,
+      4660,
+      5442,
+      6224,
+      7006,
+      7788,
+      8570,
+      9352,
+      10134,
+      10916,
+      11698,
+      12480,
+      13262,
+      14044,
+      14826,
+      15608
+    ],
+    "eval_accuracy": [
+      0.0,
+      0.01,
+      0.0,
+      0.0,
+      0.0,
+      0.01,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0
+    ]
+  },
+  "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": 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": 0.88,
+        "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": 0.9354838709677419,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.81,
+        "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": 0.90625,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.9446254071661238,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.64,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.75,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.8275,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.87,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.88,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.946,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.99,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.996875,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.998109640831758,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 0.9791666666666666,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.88,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9378238341968912,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.89,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.95703125,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.9513888888888888,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.954248366013072,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.9329896907216495,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.94,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9836065573770492,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.9957264957264957,
+            "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": 0.6,
+        "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": 0.7315436241610739,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.2,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.36,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.7633333333333333,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.09,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.31,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.5875,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.995,
+        "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": 0.9977220956719818,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.81,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9035532994923858,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.79,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9190283400809717,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.9477124183006536,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.73,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9161073825503355,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.9306930693069307,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.8316666666666667,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 3,
+      "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": 0.8316666666666667
+}

add_sub_sorl_v1_abs30_K3_50K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71138eb1b211df34431f196a198b3c938bef41da28e9d5b41eae4d9a62a276d3
+size 650385300

add_sub_sorl_v1_abs30_K3_50K/train_config.json

@@ -0,0 +1,84 @@
+{
+  "num_rollouts": 4,
+  "K": 3,
+  "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": 8e-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": 20,
+  "emb_warmup_steps": 0,
+  "log_every": 50,
+  "eval_every": 781,
+  "save_every": 999999,
+  "eval_samples": 100,
+  "output_dir": "ckpt/sweep/as_sorl_abs30_K3_50K",
+  "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": 30,
+  "dataset_size": 50000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162519662,
+  "run_name": "add_sub_sorl_v1_abs30_K3_50K",
+  "git_commit": "dc8dd776fb0c30a4c9073052dcc5e943e0fd80c6",
+  "timestamp": "2026-04-13T04:59:45.995142+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": "gx9albkf",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/gx9albkf",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.8316666666666667,
+  "eval_method": "ArithmeticEvaluator"
+}