Upload add_sub_sorl_v1_abs2_100K

add_sub_sorl_v1_abs2_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": 151646
+}

add_sub_sorl_v1_abs2_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_abs2_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": [
+      21.55743980407715,
+      14.456762313842773,
+      13.332813262939453,
+      11.099372863769531,
+      9.76402759552002,
+      8.166985511779785,
+      6.677196502685547,
+      5.956841468811035,
+      5.798933029174805,
+      4.650259494781494,
+      4.677316665649414,
+      4.337405204772949,
+      4.556812286376953,
+      4.077780723571777,
+      3.8547043800354004,
+      2.160536050796509,
+      0.08490012586116791,
+      -0.15047906339168549,
+      0.34938758611679077,
+      -0.07935617864131927,
+      -0.062447674572467804,
+      -0.04229782521724701,
+      0.23106995224952698,
+      0.14142271876335144,
+      0.2878555953502655,
+      -0.8651682734489441,
+      -1.5847699642181396,
+      -2.3202571868896484,
+      -3.3132762908935547,
+      -3.8882954120635986,
+      -4.961565017700195,
+      -5.984193325042725,
+      -6.995052814483643,
+      -8.3079195022583,
+      -9.01822280883789,
+      -9.272224426269531,
+      -9.959793090820312,
+      -9.428740501403809,
+      -10.53104019165039,
+      -9.90003776550293,
+      -10.468411445617676,
+      -10.950371742248535,
+      -11.26554012298584,
+      -11.296932220458984,
+      -11.408262252807617,
+      -12.537223815917969,
+      -12.446996688842773,
+      -12.266963958740234,
+      -12.521815299987793,
+      -12.351020812988281,
+      -12.981197357177734,
+      -13.050939559936523,
+      -12.874207496643066,
+      -13.596314430236816,
+      -13.118048667907715,
+      -12.82033920288086,
+      -12.900100708007812,
+      -13.404964447021484,
+      -13.971785545349121,
+      -14.11911392211914,
+      -13.840298652648926,
+      -13.950313568115234,
+      -13.34890365600586,
+      -14.914931297302246,
+      -14.512497901916504,
+      -13.609143257141113,
+      -13.685526847839355,
+      -13.983019828796387,
+      -14.2177095413208,
+      -13.94830322265625,
+      -14.049311637878418,
+      -13.777447700500488,
+      -14.062230110168457,
+      -14.017431259155273,
+      -13.942742347717285,
+      -14.194819450378418,
+      -14.288540840148926,
+      -13.903726577758789,
+      -13.793651580810547,
+      -14.276308059692383,
+      -14.7862548828125,
+      -13.599206924438477,
+      -13.381665229797363,
+      -13.555388450622559,
+      -14.167590141296387,
+      -13.62869930267334,
+      -13.118552207946777,
+      -12.308067321777344,
+      -10.674836158752441,
+      -8.095012664794922,
+      -8.552412986755371,
+      -6.768813133239746,
+      -6.4329681396484375,
+      -5.3595757484436035,
+      -5.664283752441406,
+      -5.1527419090271,
+      -4.728729724884033,
+      -5.28671932220459,
+      -4.602601051330566,
+      -4.507763385772705,
+      -3.9181900024414062,
+      -4.268271446228027,
+      -4.134187698364258,
+      -4.568851947784424,
+      -4.330838680267334,
+      -4.221950054168701,
+      -3.783834934234619,
+      -4.0207133293151855,
+      -4.3817243576049805,
+      -3.8077564239501953,
+      -3.842586040496826,
+      -3.489835739135742,
+      -4.192014217376709,
+      -3.756579875946045,
+      -3.0009307861328125,
+      -2.9890785217285156,
+      -3.1000733375549316,
+      -3.1378750801086426,
+      -4.129668712615967,
+      -3.153613328933716,
+      -3.7414538860321045,
+      -3.187767267227173,
+      -3.159456729888916,
+      -2.40816330909729,
+      -3.079115629196167,
+      -2.8455893993377686,
+      -2.674211025238037,
+      -2.6699635982513428,
+      -3.312190055847168,
+      -2.572709798812866,
+      -2.7692463397979736,
+      -2.3675882816314697,
+      -2.5942609310150146,
+      -2.752772808074951,
+      -2.179344415664673,
+      -2.1692566871643066,
+      -3.2311666011810303,
+      -2.5556187629699707,
+      -2.5724711418151855,
+      -2.1499860286712646,
+      -3.1964876651763916,
+      -2.218101739883423,
+      -2.623993158340454,
+      -2.063426971435547,
+      -2.512239456176758,
+      -2.2239904403686523,
+      -2.098299741744995,
+      -2.2338473796844482,
+      -2.2596230506896973,
+      -2.153071880340576,
+      -2.321948766708374,
+      -1.9581347703933716,
+      -2.570591449737549,
+      -2.1422479152679443,
+      -1.8671281337738037,
+      -1.921321153640747,
+      -2.3773269653320312,
+      -2.2209441661834717,
+      -1.8629577159881592,
+      -1.8766900300979614,
+      -1.7334203720092773,
+      -1.6970460414886475,
+      -1.9741642475128174,
+      -1.9299850463867188,
+      -1.70906662940979,
+      -1.8495261669158936,
+      -1.7984683513641357,
+      -1.5267865657806396,
+      -1.6772974729537964,
+      -1.6340644359588623,
+      -3.483680486679077,
+      -1.6918689012527466,
+      -1.8547474145889282,
+      -1.629125952720642,
+      -1.6546547412872314,
+      -1.189652919769287,
+      -1.7392150163650513,
+      -1.6066049337387085,
+      -2.0829155445098877,
+      -1.5005329847335815,
+      -1.697817325592041,
+      -2.0126705169677734,
+      -1.7649344205856323,
+      -1.6129711866378784,
+      -2.1282825469970703,
+      -1.4418532848358154,
+      -1.6184873580932617,
+      -1.1675630807876587,
+      -1.8687047958374023,
+      -1.6724880933761597,
+      -2.17988920211792,
+      -1.3837471008300781,
+      -1.1479820013046265,
+      -1.249972939491272,
+      -1.8096187114715576,
+      -1.4028083086013794,
+      -1.5338959693908691,
+      -1.2512074708938599,
+      -1.3164658546447754,
+      -0.9766436219215393,
+      -1.290589690208435,
+      -1.2838385105133057,
+      -1.1990852355957031,
+      -1.4214012622833252,
+      -1.2117725610733032,
+      -1.4734677076339722,
+      -0.77535080909729,
+      -1.0201669931411743,
+      -1.6018017530441284,
+      -1.133147120475769,
+      -1.1521793603897095,
+      -0.9268485903739929,
+      -0.9853177070617676,
+      -1.1100549697875977,
+      -1.2448731660842896,
+      -1.3782854080200195,
+      -1.0712212324142456,
+      -0.8957306742668152,
+      -1.0588748455047607,
+      -0.7864165902137756,
+      -0.7465080618858337,
+      -0.9283392429351807,
+      -0.7342801094055176,
+      -1.0154541730880737,
+      -0.9934065341949463,
+      -1.2512985467910767,
+      -0.9935685396194458,
+      -1.2980875968933105,
+      -1.1094932556152344,
+      -1.0540815591812134,
+      -0.7584136724472046,
+      -0.8151233792304993,
+      -0.8131420016288757,
+      -0.9537174701690674,
+      -0.8363350033760071,
+      -1.402547836303711,
+      -0.8352566957473755,
+      -0.8664710521697998,
+      -1.0080583095550537,
+      -0.7523218393325806,
+      -1.4662609100341797,
+      -0.9632124304771423,
+      -1.0357917547225952,
+      -0.8007121682167053,
+      -0.8814998269081116,
+      -0.8099896311759949,
+      -0.7689655423164368,
+      -0.6990557312965393,
+      -0.6981830596923828,
+      -0.5704962611198425,
+      -0.5685226321220398,
+      -0.47645679116249084,
+      -0.7991287112236023,
+      -0.4307916760444641,
+      -0.2727436423301697,
+      -0.5455629229545593,
+      -0.3937355875968933,
+      -0.334403932094574,
+      -0.5208835601806641,
+      -0.42717838287353516,
+      -0.39090976119041443,
+      -0.4524388015270233,
+      -0.4628322422504425,
+      -0.4377143979072571,
+      -0.34418362379074097,
+      -0.421299546957016,
+      -0.43138304352760315,
+      -0.39728233218193054,
+      -0.3580074608325958,
+      -0.22887344658374786,
+      -0.25300633907318115,
+      -0.205618754029274,
+      -0.3025527000427246,
+      -0.2220858335494995,
+      -0.2500958740711212,
+      -0.3251419961452484,
+      -0.25898948311805725,
+      -0.2661969065666199,
+      -0.18099558353424072,
+      -0.3117075562477112,
+      -0.24770574271678925,
+      -0.26206326484680176,
+      -0.3404958248138428,
+      -0.22024191915988922,
+      -0.18646451830863953,
+      -0.2535249888896942,
+      -0.20417502522468567,
+      -0.2095787525177002,
+      -0.259319931268692,
+      -0.41926753520965576,
+      -0.23228217661380768,
+      -0.19904252886772156,
+      -0.14680063724517822,
+      -0.1614488959312439,
+      -0.14487634599208832,
+      -0.2808101773262024,
+      -0.18040938675403595,
+      -0.14410051703453064,
+      -0.049752090126276016,
+      -0.10657652467489243,
+      -0.05924094840884209,
+      -0.10887810587882996,
+      -0.2908860146999359,
+      -0.0975363627076149,
+      -0.17220471799373627,
+      -0.15736867487430573,
+      -0.1087040826678276,
+      -0.10741429030895233,
+      -0.11429610848426819,
+      -0.07702717185020447
+    ],
+    "base_loss": [
+      10.55781364440918,
+      8.231264114379883,
+      7.375144958496094,
+      6.467900276184082,
+      5.595902442932129,
+      4.134605407714844,
+      2.3774359226226807,
+      2.105088472366333,
+      1.9189599752426147,
+      1.8717594146728516,
+      1.8702201843261719,
+      1.8352535963058472,
+      1.8482831716537476,
+      1.869525671005249,
+      1.8779953718185425,
+      1.8948261737823486,
+      1.8043323755264282,
+      1.8525203466415405,
+      1.8378676176071167,
+      1.8408254384994507,
+      1.7881041765213013,
+      1.8258365392684937,
+      1.8472777605056763,
+      1.750261902809143,
+      1.731013298034668,
+      1.708060383796692,
+      1.7264175415039062,
+      1.6409509181976318,
+      1.7112095355987549,
+      1.7249946594238281,
+      1.6471461057662964,
+      1.7835009098052979,
+      1.7195311784744263,
+      1.7008330821990967,
+      1.7077165842056274,
+      1.6616802215576172,
+      1.7441548109054565,
+      1.7079802751541138,
+      1.7124301195144653,
+      1.6236799955368042,
+      1.6541255712509155,
+      1.6965360641479492,
+      1.6469358205795288,
+      1.6354162693023682,
+      1.6002006530761719,
+      1.710097074508667,
+      1.6456886529922485,
+      1.6939975023269653,
+      1.632445216178894,
+      1.6538705825805664,
+      1.710196614265442,
+      1.6665986776351929,
+      1.6424137353897095,
+      1.6636838912963867,
+      1.6194913387298584,
+      1.6220190525054932,
+      1.539172887802124,
+      1.6079224348068237,
+      1.6561269760131836,
+      1.6791114807128906,
+      1.6207760572433472,
+      1.6462295055389404,
+      1.5754941701889038,
+      1.6986427307128906,
+      1.660187005996704,
+      1.5879908800125122,
+      1.5971603393554688,
+      1.595866322517395,
+      1.620086908340454,
+      1.577508568763733,
+      1.5957127809524536,
+      1.5771719217300415,
+      1.5998361110687256,
+      1.585344672203064,
+      1.5755517482757568,
+      1.592895746231079,
+      1.606165885925293,
+      1.58901047706604,
+      1.5569877624511719,
+      1.6083017587661743,
+      1.6573187112808228,
+      1.529327630996704,
+      1.5545157194137573,
+      1.5298712253570557,
+      1.5886039733886719,
+      1.5447196960449219,
+      1.472448706626892,
+      1.4083069562911987,
+      1.1975148916244507,
+      0.9487571716308594,
+      0.9713836908340454,
+      0.7721220254898071,
+      0.7260175347328186,
+      0.6046575903892517,
+      0.6525946259498596,
+      0.5858612060546875,
+      0.5463818311691284,
+      0.596659243106842,
+      0.5324410200119019,
+      0.5113546848297119,
+      0.4467690885066986,
+      0.49367403984069824,
+      0.4715045392513275,
+      0.5395259261131287,
+      0.48863282799720764,
+      0.4811270833015442,
+      0.4409574866294861,
+      0.46015387773513794,
+      0.5071967244148254,
+      0.4407365024089813,
+      0.43596014380455017,
+      0.3976394236087799,
+      0.4789476990699768,
+      0.4261191785335541,
+      0.3488524854183197,
+      0.3421429991722107,
+      0.3513649106025696,
+      0.3556698262691498,
+      0.4667738378047943,
+      0.357208251953125,
+      0.4270372688770294,
+      0.3612685203552246,
+      0.36596614122390747,
+      0.28447532653808594,
+      0.3500048518180847,
+      0.34377458691596985,
+      0.3037779629230499,
+      0.3040713369846344,
+      0.374905526638031,
+      0.2936929166316986,
+      0.31861549615859985,
+      0.2704242169857025,
+      0.2960323691368103,
+      0.31331783533096313,
+      0.2655356824398041,
+      0.2478264421224594,
+      0.3676256537437439,
+      0.2925736606121063,
+      0.2932095229625702,
+      0.27233609557151794,
+      0.36574485898017883,
+      0.25314006209373474,
+      0.29876837134361267,
+      0.23810824751853943,
+      0.2855685353279114,
+      0.2536938488483429,
+      0.24312074482440948,
+      0.25708886981010437,
+      0.2590205669403076,
+      0.24564266204833984,
+      0.2646576762199402,
+      0.22410044074058533,
+      0.2995555102825165,
+      0.24821220338344574,
+      0.23046858608722687,
+      0.22727033495903015,
+      0.27086523175239563,
+      0.2533358037471771,
+      0.2169254869222641,
+      0.22717027366161346,
+      0.2051292210817337,
+      0.1950482279062271,
+      0.23457269370555878,
+      0.22084100544452667,
+      0.1970510482788086,
+      0.21207870543003082,
+      0.2063349336385727,
+      0.17636503279209137,
+      0.19287574291229248,
+      0.18833866715431213,
+      0.3938497006893158,
+      0.19834022223949432,
+      0.21254512667655945,
+      0.1910787969827652,
+      0.19024720788002014,
+      0.14706312119960785,
+      0.19971981644630432,
+      0.18616066873073578,
+      0.2377772331237793,
+      0.1731453239917755,
+      0.19504819810390472,
+      0.2301284223794937,
+      0.2027578353881836,
+      0.18562926352024078,
+      0.2430158108472824,
+      0.1668236255645752,
+      0.18670785427093506,
+      0.1360914409160614,
+      0.21598540246486664,
+      0.19217689335346222,
+      0.24856576323509216,
+      0.16024446487426758,
+      0.1372070610523224,
+      0.15015220642089844,
+      0.2074291706085205,
+      0.16735093295574188,
+      0.1768689900636673,
+      0.14595003426074982,
+      0.15265120565891266,
+      0.11484047025442123,
+      0.1501741111278534,
+      0.14941106736660004,
+      0.1398736834526062,
+      0.1644693911075592,
+      0.14105084538459778,
+      0.17046555876731873,
+      0.09274041652679443,
+      0.12018904834985733,
+      0.18430931866168976,
+      0.13885965943336487,
+      0.13433656096458435,
+      0.10935930907726288,
+      0.11638494580984116,
+      0.1296434849500656,
+      0.14469733834266663,
+      0.1595931202173233,
+      0.12532873451709747,
+      0.10585658997297287,
+      0.12406021356582642,
+      0.09367185086011887,
+      0.0893135592341423,
+      0.10987697541713715,
+      0.08813031017780304,
+      0.11914919316768646,
+      0.11809062957763672,
+      0.14532360434532166,
+      0.12125253677368164,
+      0.1505296230316162,
+      0.12963902950286865,
+      0.12339673936367035,
+      0.0905638113617897,
+      0.09690284729003906,
+      0.09675943851470947,
+      0.11234831809997559,
+      0.09925606101751328,
+      0.16212299466133118,
+      0.09914389997720718,
+      0.10257159173488617,
+      0.11831194162368774,
+      0.08988779038190842,
+      0.16919860243797302,
+      0.11332643032073975,
+      0.1219133734703064,
+      0.09550021588802338,
+      0.1042398065328598,
+      0.09629204124212265,
+      0.09185947477817535,
+      0.08394961804151535,
+      0.0838543176651001,
+      0.06975314766168594,
+      0.06948263198137283,
+      0.05922183766961098,
+      0.09506983309984207,
+      0.054161328822374344,
+      0.03661850467324257,
+      0.06693246215581894,
+      0.050016459077596664,
+      0.044011440128088,
+      0.06418493390083313,
+      0.053776971995830536,
+      0.04977472871541977,
+      0.05943726375699043,
+      0.05769496411085129,
+      0.0548989363014698,
+      0.044522084295749664,
+      0.05372154712677002,
+      0.05429927259683609,
+      0.05040301755070686,
+      0.04686831310391426,
+      0.03171539679169655,
+      0.034375984221696854,
+      0.02913111075758934,
+      0.03988438844680786,
+      0.030962970107793808,
+      0.03404800966382027,
+      0.04251670092344284,
+      0.03505539894104004,
+      0.03583652153611183,
+      0.026452062651515007,
+      0.04097387567162514,
+      0.033787403255701065,
+      0.035386499017477036,
+      0.04416670277714729,
+      0.030734002590179443,
+      0.026986896991729736,
+      0.034428056329488754,
+      0.028954336419701576,
+      0.02955194003880024,
+      0.03509492427110672,
+      0.05284568667411804,
+      0.03208259865641594,
+      0.02837185375392437,
+      0.022582465782761574,
+      0.024208368733525276,
+      0.0223563052713871,
+      0.03746243566274643,
+      0.026329679414629936,
+      0.022274931892752647,
+      0.011795973405241966,
+      0.018098199740052223,
+      0.012842332012951374,
+      0.018380993977189064,
+      0.038584090769290924,
+      0.017115477472543716,
+      0.025396117940545082,
+      0.023746242746710777,
+      0.01833145134150982,
+      0.018189962953329086,
+      0.018952786922454834,
+      0.014812135137617588
+    ],
+    "info_loss": [
+      -0.23081111907958984,
+      -0.5864462852478027,
+      -0.36399364471435547,
+      -0.28661489486694336,
+      -0.17821931838989258,
+      -0.10320901870727539,
+      -0.028664588928222656,
+      -0.05982160568237305,
+      -0.052922725677490234,
+      -0.16179072856903076,
+      -0.15811514854431152,
+      -0.18838560581207275,
+      -0.16759395599365234,
+      -0.21485137939453125,
+      -0.21427619457244873,
+      -0.21580040454864502,
+      -0.20179104804992676,
+      -0.213822603225708,
+      -0.16826021671295166,
+      -0.2020508050918579,
+      -0.1943281888961792,
+      -0.20006966590881348,
+      -0.16979622840881348,
+      -0.1685103178024292,
+      -0.15185558795928955,
+      -0.265255331993103,
+      -0.33891570568084717,
+      -0.40550482273101807,
+      -0.5117264986038208,
+      -0.5712318420410156,
+      -0.6699534058570862,
+      -0.7878265976905823,
+      -0.8829475045204163,
+      -1.0115578174591064,
+      -1.083652138710022,
+      -1.1040446758270264,
+      -1.180250644683838,
+      -1.1229536533355713,
+      -1.2351194620132446,
+      -1.1612054109573364,
+      -1.2211158275604248,
+      -1.2747255563735962,
+      -1.3001933097839355,
+      -1.3030011653900146,
+      -1.310207724571228,
+      -1.4331682920455933,
+      -1.4174022674560547,
+      -1.4047023057937622,
+      -1.4245320558547974,
+      -1.4083890914916992,
+      -1.4774441719055176,
+      -1.4791162014007568,
+      -1.458984136581421,
+      -1.5335596799850464,
+      -1.4807966947555542,
+      -1.451379418373108,
+      -1.4512931108474731,
+      -1.5090150833129883,
+      -1.5702227354049683,
+      -1.5864689350128174,
+      -1.5529896020889282,
+      -1.5662482976913452,
+      -1.4988164901733398,
+      -1.667850136756897,
+      -1.6237952709197998,
+      -1.5258361101150513,
+      -1.5347386598587036,
+      -1.5641248226165771,
+      -1.590108871459961,
+      -1.5585733652114868,
+      -1.5704230070114136,
+      -1.5418198108673096,
+      -1.5720810890197754,
+      -1.5662081241607666,
+      -1.5576754808425903,
+      -1.5846681594848633,
+      -1.5954533815383911,
+      -1.5551222562789917,
+      -1.5410363674163818,
+      -1.5942590236663818,
+      -1.6501209735870361,
+      -1.5187057256698608,
+      -1.4995180368423462,
+      -1.514379620552063,
+      -1.5813732147216797,
+      -1.5231415033340454,
+      -1.4648243188858032,
+      -1.3773880004882812,
+      -1.1932123899459839,
+      -0.9102264642715454,
+      -0.9582034945487976,
+      -0.7599639892578125,
+      -0.7217062711715698,
+      -0.6021677851676941,
+      -0.6374126672744751,
+      -0.5795924663543701,
+      -0.5332291126251221,
+      -0.5940366983413696,
+      -0.5192331671714783,
+      -0.5076391696929932,
+      -0.4421822428703308,
+      -0.48185449838638306,
+      -0.4663180410861969,
+      -0.5165236592292786,
+      -0.4875980317592621,
+      -0.47597524523735046,
+      -0.4281480610370636,
+      -0.453742116689682,
+      -0.49453476071357727,
+      -0.4305174648761749,
+      -0.43351098895072937,
+      -0.3944139778614044,
+      -0.47276315093040466,
+      -0.4239259362220764,
+      -0.3406214714050293,
+      -0.33879050612449646,
+      -0.35079655051231384,
+      -0.35499829053878784,
+      -0.4652974009513855,
+      -0.35673290491104126,
+      -0.4225170314311981,
+      -0.3605479300022125,
+      -0.3581975996494293,
+      -0.27489882707595825,
+      -0.34855759143829346,
+      -0.3245709240436554,
+      -0.3034340739250183,
+      -0.3030414879322052,
+      -0.3743423819541931,
+      -0.2922758162021637,
+      -0.3144170939922333,
+      -0.26943954825401306,
+      -0.29466721415519714,
+      -0.312252938747406,
+      -0.2501186728477478,
+      -0.24738256633281708,
+      -0.36551225185394287,
+      -0.29044491052627563,
+      -0.2921958565711975,
+      -0.24785564839839935,
+      -0.36184948682785034,
+      -0.252761572599411,
+      -0.29790693521499634,
+      -0.23578904569149017,
+      -0.28542351722717285,
+      -0.25339555740356445,
+      -0.23976655304431915,
+      -0.2547180652618408,
+      -0.25748777389526367,
+      -0.24550583958625793,
+      -0.26428404450416565,
+      -0.22384725511074066,
+      -0.29263776540756226,
+      -0.2446698695421219,
+      -0.2153833955526352,
+      -0.22048817574977875,
+      -0.2704443633556366,
+      -0.25305241346359253,
+      -0.21361403167247772,
+      -0.21601071953773499,
+      -0.19948036968708038,
+      -0.1948372721672058,
+      -0.22649627923965454,
+      -0.22070898115634918,
+      -0.19623607397079468,
+      -0.2117931991815567,
+      -0.2061016857624054,
+      -0.17594103515148163,
+      -0.19263973832130432,
+      -0.18786264955997467,
+      -0.393390417098999,
+      -0.19464483857154846,
+      -0.21235060691833496,
+      -0.18764138221740723,
+      -0.1901116818189621,
+      -0.13929884135723114,
+      -0.1995152086019516,
+      -0.18489839136600494,
+      -0.23769024014472961,
+      -0.172989621758461,
+      -0.1949077546596527,
+      -0.22990308701992035,
+      -0.20239092409610748,
+      -0.18548186123371124,
+      -0.24277549982070923,
+      -0.16649411618709564,
+      -0.18614347279071808,
+      -0.1359878033399582,
+      -0.2140895277261734,
+      -0.19208790361881256,
+      -0.24846652150154114,
+      -0.16002069413661957,
+      -0.13413934409618378,
+      -0.14563290774822235,
+      -0.20732542872428894,
+      -0.1626378893852234,
+      -0.17669813334941864,
+      -0.14533713459968567,
+      -0.15253326296806335,
+      -0.11477691680192947,
+      -0.1496976613998413,
+      -0.14894747734069824,
+      -0.13951662182807922,
+      -0.1642075926065445,
+      -0.1409120261669159,
+      -0.1700175404548645,
+      -0.09242932498455048,
+      -0.11965568363666534,
+      -0.18423114717006683,
+      -0.1328209638595581,
+      -0.13427221775054932,
+      -0.10924125462770462,
+      -0.11579099297523499,
+      -0.12959317862987518,
+      -0.14457730948925018,
+      -0.15940792858600616,
+      -0.12527652084827423,
+      -0.10578018426895142,
+      -0.12391418218612671,
+      -0.09362959116697311,
+      -0.08920267224311829,
+      -0.10944169014692307,
+      -0.08786164969205856,
+      -0.11908160150051117,
+      -0.11677320301532745,
+      -0.14528240263462067,
+      -0.11710207909345627,
+      -0.15048231184482574,
+      -0.12953327596187592,
+      -0.12336805462837219,
+      -0.09051825851202011,
+      -0.09682397544384003,
+      -0.09661003202199936,
+      -0.1122264415025711,
+      -0.09917972981929779,
+      -0.1620868593454361,
+      -0.09905996918678284,
+      -0.10252546519041061,
+      -0.11825735867023468,
+      -0.08984114974737167,
+      -0.16916626691818237,
+      -0.11327371746301651,
+      -0.12139031291007996,
+      -0.09524128586053848,
+      -0.10419416427612305,
+      -0.09624885767698288,
+      -0.09170279651880264,
+      -0.08392082899808884,
+      -0.08382471650838852,
+      -0.06964603811502457,
+      -0.06942091137170792,
+      -0.05918891355395317,
+      -0.09504035860300064,
+      -0.054116759449243546,
+      -0.036556001752614975,
+      -0.06687008589506149,
+      -0.049994878470897675,
+      -0.04346111789345741,
+      -0.06412678211927414,
+      -0.053715310990810394,
+      -0.0496886745095253,
+      -0.05680719017982483,
+      -0.057672612369060516,
+      -0.05488137900829315,
+      -0.04449116438627243,
+      -0.05312187969684601,
+      -0.05418844148516655,
+      -0.050388071686029434,
+      -0.0461072102189064,
+      -0.031678467988967896,
+      -0.034358035773038864,
+      -0.02909480407834053,
+      -0.0398634597659111,
+      -0.030924513936042786,
+      -0.03403429314494133,
+      -0.04238564148545265,
+      -0.035024289041757584,
+      -0.035823091864585876,
+      -0.02636440470814705,
+      -0.040887799113988876,
+      -0.033768974244594574,
+      -0.0353645496070385,
+      -0.044085800647735596,
+      -0.030717631801962852,
+      -0.026965029537677765,
+      -0.034415047615766525,
+      -0.028932679444551468,
+      -0.02953263372182846,
+      -0.035061128437519073,
+      -0.05283093452453613,
+      -0.032056089490652084,
+      -0.028361082077026367,
+      -0.022557955235242844,
+      -0.02418987825512886,
+      -0.02234337106347084,
+      -0.03744680806994438,
+      -0.026293661445379257,
+      -0.02225743979215622,
+      -0.011774775572121143,
+      -0.018087247386574745,
+      -0.012828130275011063,
+      -0.018345512449741364,
+      -0.0385664664208889,
+      -0.01708495244383812,
+      -0.025379735976457596,
+      -0.023731086403131485,
+      -0.018323229625821114,
+      -0.01817993074655533,
+      -0.018944548442959785,
+      -0.014803697355091572
+    ],
+    "abs_loss": [
+      0.6956941485404968,
+      0.6403825283050537,
+      0.4830767512321472,
+      0.36317873001098633,
+      0.22355778515338898,
+      0.06646791100502014,
+      0.02232569456100464,
+      0.008802840486168861,
+      0.004403675906360149,
+      0.002795965876430273,
+      0.0021197409369051456,
+      0.0018710936419665813,
+      0.0015149149112403393,
+      0.0012303799157962203,
+      0.001454415381886065,
+      0.0016817658906802535,
+      0.005805621854960918,
+      0.008533228188753128,
+      0.009683016687631607,
+      0.008380381390452385,
+      0.010228028520941734,
+      0.00821185577660799,
+      0.0081139225512743,
+      0.008338656276464462,
+      0.007492415141314268,
+      0.0086920615285635,
+      0.009874568320810795,
+      0.01173701323568821,
+      0.011567603796720505,
+      0.014372609555721283,
+      0.014484841376543045,
+      0.015789855271577835,
+      0.018379611894488335,
+      0.02154083363711834,
+      0.02140096388757229,
+      0.019036047160625458,
+      0.03399075195193291,
+      0.01921842247247696,
+      0.021755583584308624,
+      0.02301720529794693,
+      0.020673472434282303,
+      0.02469949796795845,
+      0.028265394270420074,
+      0.03813191130757332,
+      0.032795242965221405,
+      0.02499081939458847,
+      0.02345442585647106,
+      0.017639651894569397,
+      0.01972111128270626,
+      0.02040751278400421,
+      0.02078351005911827,
+      0.02873343415558338,
+      0.023858370259404182,
+      0.024083605036139488,
+      0.01723884418606758,
+      0.012871598824858665,
+      0.01174084097146988,
+      0.02683127671480179,
+      0.019723625853657722,
+      0.01465850230306387,
+      0.01228727214038372,
+      0.012644937261939049,
+      0.008676152676343918,
+      0.0073304432444274426,
+      0.005936686880886555,
+      0.006608534604310989,
+      0.004954442847520113,
+      0.006269386038184166,
+      0.0035136525984853506,
+      0.008629296906292439,
+      0.004004452843219042,
+      0.002656985307112336,
+      0.003876788541674614,
+      0.0022140294313430786,
+      0.0027813275810331106,
+      0.0029296008870005608,
+      0.002483564196154475,
+      0.0011703572236001492,
+      0.002269966062158346,
+      0.001995225204154849,
+      0.0017770285485312343,
+      0.001401030458509922,
+      0.0013372995890676975,
+      0.0008072683122009039,
+      0.001991731347516179,
+      0.0012733193580061197,
+      0.0004471038992051035,
+      0.001032060943543911,
+      0.0025046896189451218,
+      0.00157322664745152,
+      0.0015743678668513894,
+      0.0009261806262657046,
+      0.0016998673090711236,
+      0.0012237753253430128,
+      0.0011928051244467497,
+      0.0009904535254463553,
+      0.0010454843286424875,
+      0.0010850548278540373,
+      0.0007955672917887568,
+      0.0004715621180366725,
+      0.0010601218091323972,
+      0.0005062565905973315,
+      0.0011024527484551072,
+      0.0006975189317017794,
+      0.00041120214154943824,
+      0.0005454591591842473,
+      0.00043180410284549,
+      0.00047692624502815306,
+      0.00032515102066099644,
+      0.0005417237989604473,
+      0.0003396912943571806,
+      0.0006373580545186996,
+      0.0004139956145081669,
+      0.0002848997828550637,
+      0.0002313605509698391,
+      0.0002655877615325153,
+      0.000346259301295504,
+      0.0002491240738891065,
+      0.00022521737264469266,
+      0.0004311777011025697,
+      0.0002283934154547751,
+      0.00011914936476387084,
+      0.00012467293709050864,
+      0.00017286086222156882,
+      0.00014875028864480555,
+      0.00010279472917318344,
+      0.00013618305092677474,
+      0.00016037074965424836,
+      8.233635162469e-05,
+      0.00011739895853679627,
+      0.00015547614020761102,
+      0.00017099488468375057,
+      0.00012639487977139652,
+      0.00010321970330551267,
+      9.091859101317823e-05,
+      0.00012881489237770438,
+      0.00012048651842633262,
+      0.00010087819828186184,
+      7.654543151147664e-05,
+      6.016628321958706e-05,
+      9.542417683405802e-05,
+      0.0001161744148703292,
+      9.009813948068768e-05,
+      0.0001134482808993198,
+      0.0001473266165703535,
+      6.867557385703549e-05,
+      6.132695853011683e-05,
+      4.410174005897716e-05,
+      4.993489710614085e-05,
+      5.260844773147255e-05,
+      3.442721572355367e-05,
+      5.96915851929225e-05,
+      3.4494209103286266e-05,
+      1.4586964425689075e-05,
+      3.2134674256667495e-05,
+      7.720987196080387e-05,
+      5.327590770320967e-05,
+      5.947142926743254e-05,
+      3.0571085517294705e-05,
+      3.140626722597517e-05,
+      3.9412054320564494e-05,
+      4.637734673451632e-05,
+      3.103291237493977e-05,
+      3.9987040509004146e-05,
+      7.223954162327573e-05,
+      7.384084892692044e-05,
+      1.8627084500622004e-05,
+      1.5019706552266143e-05,
+      2.026137735811062e-05,
+      2.9476159397745505e-05,
+      2.0226163542247377e-05,
+      2.787144694593735e-05,
+      1.995943057409022e-05,
+      1.4738093341293279e-05,
+      1.4859501789032947e-05,
+      0.00012163238716311753,
+      8.50914693728555e-06,
+      2.5399749574717134e-05,
+      1.3118664355715737e-05,
+      8.922604138206225e-06,
+      1.2929611330037005e-05,
+      3.2667107007000595e-05,
+      2.8868156732642092e-05,
+      2.782613955787383e-05,
+      1.045096360030584e-05,
+      1.0925508831860498e-05,
+      3.188596019754186e-05,
+      2.5854613340925425e-05,
+      7.569581612187903e-06,
+      1.5833149518584833e-05,
+      1.7612708688830025e-05,
+      1.6430862160632387e-05,
+      7.91602815297665e-06,
+      6.077641501178732e-06,
+      4.861400611844147e-06,
+      2.4933491658885032e-05,
+      3.9220751204993576e-05,
+      2.1398182070697658e-05,
+      1.5906902262941003e-05,
+      1.361541490041418e-05,
+      9.693720130599104e-06,
+      1.472571511840215e-05,
+      7.660580195079092e-06,
+      5.466765287565067e-06,
+      1.3933775335317478e-05,
+      1.1707485100487247e-05,
+      6.2591211644758005e-06,
+      3.795095381065039e-06,
+      7.233882570290007e-06,
+      7.799499144311994e-06,
+      3.169261390212341e-06,
+      5.056033842265606e-06,
+      1.7979424228542484e-05,
+      8.17397449281998e-06,
+      4.836298103327863e-06,
+      4.192570486338809e-06,
+      4.917276783089619e-06,
+      1.2816100024792831e-05,
+      6.51421032671351e-06,
+      5.48346952200518e-06,
+      4.327761871536495e-06,
+      2.0200492144795135e-05,
+      7.318658390431665e-06,
+      1.0495868082216475e-05,
+      1.3226725968706887e-05,
+      3.3489959605503827e-06,
+      3.98590782424435e-06,
+      3.988625849160599e-06,
+      3.8015793961676536e-06,
+      2.28077988140285e-06,
+      8.904497008188628e-06,
+      3.6851079130428843e-06,
+      5.232717740000226e-06,
+      3.3871547202579677e-06,
+      3.0314097330119694e-06,
+      1.3606569382318412e-06,
+      2.3692584818491014e-06,
+      6.790003681089729e-06,
+      3.873267814924475e-06,
+      4.318375431466848e-06,
+      2.1140845092304517e-06,
+      2.8637773539230693e-06,
+      2.913083790190285e-06,
+      2.412086359981913e-06,
+      2.5638828446972184e-06,
+      7.4935278462362476e-06,
+      4.210303814033978e-06,
+      2.1997677777108038e-06,
+      3.6832982459600316e-06,
+      6.044816927897045e-06,
+      1.5851035186642548e-06,
+      2.7483069970912766e-06,
+      3.7745935514976736e-06,
+      4.3389072743593715e-06,
+      2.80880317404808e-06,
+      5.580228389590047e-06,
+      5.642981705022976e-06,
+      3.0220026019378565e-06,
+      2.6318678010284202e-06,
+      2.429781488899607e-06,
+      4.930218892695848e-06,
+      2.4428300093859434e-06,
+      1.884986886580009e-06,
+      2.056345692835748e-06,
+      4.830387297261041e-06,
+      3.494228167255642e-06,
+      2.521029500712757e-06,
+      1.7527327145216987e-06,
+      3.6217859360476723e-06,
+      2.2481840460386593e-06,
+      6.567560376424808e-06,
+      2.340391802135855e-06,
+      4.115353931410937e-06,
+      1.026311906571209e-06,
+      2.5918307073879987e-06,
+      2.5536487555655185e-06,
+      3.7009060633863555e-06,
+      3.1235319966071984e-06,
+      2.0786774257430807e-06,
+      2.4688247322046664e-06,
+      2.1541113710554782e-06,
+      8.577446237723052e-07,
+      2.0712334389827447e-06,
+      3.0062208224990172e-06,
+      1.5115176665858598e-06,
+      1.4630893474532058e-06,
+      8.338574843946844e-06,
+      1.9445451471256092e-06,
+      1.1371309938112972e-06,
+      2.892598558901227e-06,
+      1.1874237770825857e-06,
+      1.3829974250256782e-06,
+      2.24999280362681e-06,
+      1.70522923781391e-06,
+      1.7257099216294591e-06,
+      2.3840793801355176e-06,
+      2.663546410985873e-06,
+      1.912902689582552e-06,
+      2.279763066326268e-06,
+      1.7657548596616834e-06,
+      2.6402431103633717e-06,
+      1.6065087038441561e-06,
+      1.0067433322547004e-06,
+      1.7247876940018614e-06,
+      1.3410799510893412e-06,
+      9.7602014648146e-07,
+      9.72294060375134e-07,
+      9.490138950241089e-07,
+      2.2453482415585313e-06,
+      1.0877777185669402e-06
+    ],
+    "zipf_loss": [
+      13.238167762756348,
+      12.025922775268555,
+      9.549297332763672,
+      7.461304187774658,
+      5.927962303161621,
+      5.057823181152344,
+      4.584174156188965,
+      4.449089050292969,
+      4.408760070800781,
+      4.396127700805664,
+      4.388035774230957,
+      4.385820388793945,
+      4.384317398071289,
+      4.356645584106445,
+      4.119325637817383,
+      2.4235458374023438,
+      0.29789766669273376,
+      0.13437329232692719,
+      0.19315381348133087,
+      0.09948839247226715,
+      0.09170723706483841,
+      0.13174110651016235,
+      0.08094307780265808,
+      0.07543011754751205,
+      0.07464892417192459,
+      0.07845544070005417,
+      0.07698220014572144,
+      0.09266608953475952,
+      0.09162209182977676,
+      0.09759122133255005,
+      0.08937418460845947,
+      0.10899271070957184,
+      0.11305401474237442,
+      0.10467098653316498,
+      0.10844150930643082,
+      0.10463868826627731,
+      0.09515910595655441,
+      0.09089392423629761,
+      0.105549156665802,
+      0.08603363484144211,
+      0.08655494451522827,
+      0.09787715971469879,
+      0.08663065731525421,
+      0.09385061264038086,
+      0.09033532440662384,
+      0.08186376839876175,
+      0.0789923220872879,
+      0.08429710566997528,
+      0.08908750116825104,
+      0.07695901393890381,
+      0.08096971362829208,
+      0.07075130939483643,
+      0.07083433121442795,
+      0.07319030910730362,
+      0.06870314478874207,
+      0.07014846056699753,
+      0.07248274236917496,
+      0.07458160817623138,
+      0.07234246283769608,
+      0.06499738991260529,
+      0.06759276241064072,
+      0.06467528641223907,
+      0.06289983540773392,
+      0.06419418752193451,
+      0.06467416137456894,
+      0.06056584045290947,
+      0.06420311331748962,
+      0.06173606216907501,
+      0.06294124573469162,
+      0.05905795097351074,
+      0.05880500748753548,
+      0.0633116364479065,
+      0.05835648626089096,
+      0.05908438563346863,
+      0.05818147584795952,
+      0.05867404863238335,
+      0.05957905575633049,
+      0.058368220925331116,
+      0.05949830263853073,
+      0.05778094753623009,
+      0.05745793133974075,
+      0.05838259682059288,
+      0.058865610510110855,
+      0.05845583230257034,
+      0.057338785380125046,
+      0.0578676275908947,
+      0.05719779059290886,
+      0.057402364909648895,
+      0.05952220782637596,
+      0.05833706632256508,
+      0.05808050557971001,
+      0.058612123131752014,
+      0.05790776014328003,
+      0.05732208490371704,
+      0.05712893232703209,
+      0.0572221502661705,
+      0.05707494169473648,
+      0.05687945336103439,
+      0.057209938764572144,
+      0.057226262986660004,
+      0.05675734207034111,
+      0.056549109518527985,
+      0.05737769603729248,
+      0.056788817048072815,
+      0.056468088179826736,
+      0.056620460003614426,
+      0.05664516240358353,
+      0.05650634691119194,
+      0.0563943088054657,
+      0.056627657264471054,
+      0.05652966350317001,
+      0.056600745767354965,
+      0.056628454476594925,
+      0.056531790643930435,
+      0.05640837550163269,
+      0.056657079607248306,
+      0.05649270862340927,
+      0.056413501501083374,
+      0.056508868932724,
+      0.05646416172385216,
+      0.05665615573525429,
+      0.056431449949741364,
+      0.05654072016477585,
+      0.05633200705051422,
+      0.05644070729613304,
+      0.0563349649310112,
+      0.05633821338415146,
+      0.056363750249147415,
+      0.056320011615753174,
+      0.05634379759430885,
+      0.05629381537437439,
+      0.05636588856577873,
+      0.05636616423726082,
+      0.05642855912446976,
+      0.05629773437976837,
+      0.05672983080148697,
+      0.05631798505783081,
+      0.05624647065997124,
+      0.056270141154527664,
+      0.05622841790318489,
+      0.05625290796160698,
+      0.05636227875947952,
+      0.056298837065696716,
+      0.05634373798966408,
+      0.05641256272792816,
+      0.056264478713274,
+      0.05623895302414894,
+      0.05624010041356087,
+      0.056229181587696075,
+      0.0563385933637619,
+      0.056230440735816956,
+      0.056231409311294556,
+      0.05622752383351326,
+      0.05623721703886986,
+      0.05623391270637512,
+      0.056282367557287216,
+      0.056246183812618256,
+      0.056238412857055664,
+      0.05625399947166443,
+      0.05624382942914963,
+      0.05625015124678612,
+      0.05627382919192314,
+      0.05622268468141556,
+      0.056259721517562866,
+      0.05623580515384674,
+      0.05631982907652855,
+      0.05621163919568062,
+      0.05625718832015991,
+      0.05622205510735512,
+      0.05622036010026932,
+      0.05637209489941597,
+      0.05623649060726166,
+      0.056211501359939575,
+      0.056207671761512756,
+      0.05621355026960373,
+      0.05626016855239868,
+      0.05621645599603653,
+      0.05621577799320221,
+      0.056208156049251556,
+      0.05621710047125816,
+      0.056210730224847794,
+      0.05622842535376549,
+      0.05621420592069626,
+      0.056215353310108185,
+      0.05645565316081047,
+      0.05626315250992775,
+      0.05623631179332733,
+      0.05622094124555588,
+      0.056204475462436676,
+      0.056212443858385086,
+      0.05620865523815155,
+      0.05621373653411865,
+      0.05620357766747475,
+      0.056203365325927734,
+      0.056205932050943375,
+      0.05621718987822533,
+      0.056212417781353,
+      0.05621163919568062,
+      0.056214120239019394,
+      0.056283775717020035,
+      0.05621178820729256,
+      0.056223753839731216,
+      0.05620645359158516,
+      0.05620468407869339,
+      0.056295569986104965,
+      0.056240878999233246,
+      0.05620139464735985,
+      0.05620034784078598,
+      0.05619967728853226,
+      0.05620207265019417,
+      0.056205909699201584,
+      0.05620419979095459,
+      0.05620541051030159,
+      0.05623263120651245,
+      0.056202132254838943,
+      0.05620027706027031,
+      0.056214697659015656,
+      0.05621323361992836,
+      0.056206077337265015,
+      0.05620695650577545,
+      0.0562046580016613,
+      0.05619858577847481,
+      0.056205399334430695,
+      0.05621153488755226,
+      0.05623354762792587,
+      0.0562015101313591,
+      0.0561993382871151,
+      0.05620548129081726,
+      0.056200191378593445,
+      0.05620197579264641,
+      0.056204184889793396,
+      0.05621318146586418,
+      0.0561983659863472,
+      0.05619821324944496,
+      0.056206002831459045,
+      0.05619761720299721,
+      0.0561988390982151,
+      0.056211285293102264,
+      0.05620306357741356,
+      0.056201428174972534,
+      0.05620291084051132,
+      0.05619806796312332,
+      0.056197818368673325,
+      0.05620019510388374,
+      0.05620177835226059,
+      0.05620618909597397,
+      0.056202542036771774,
+      0.0562027245759964,
+      0.05620947107672691,
+      0.05621035769581795,
+      0.05620364472270012,
+      0.05621017888188362,
+      0.05620468035340309,
+      0.056214168667793274,
+      0.056197620928287506,
+      0.05620499700307846,
+      0.056196168065071106,
+      0.05619549751281738,
+      0.056199103593826294,
+      0.056197527796030045,
+      0.05620173364877701,
+      0.05619557574391365,
+      0.05619874596595764,
+      0.05620028078556061,
+      0.056205496191978455,
+      0.056197308003902435,
+      0.05620187520980835,
+      0.0561952069401741,
+      0.0561959445476532,
+      0.05619560182094574,
+      0.05619737133383751,
+      0.05619792640209198,
+      0.056197091937065125,
+      0.05619625002145767,
+      0.05619877576828003,
+      0.05619743466377258,
+      0.0561976358294487,
+      0.05619720742106438,
+      0.0561961904168129,
+      0.05619632825255394,
+      0.05619640648365021,
+      0.056195635348558426,
+      0.05619524419307709,
+      0.05620010569691658,
+      0.05619870871305466,
+      0.05619727447628975,
+      0.05619659647345543,
+      0.05619543418288231,
+      0.05619629845023155,
+      0.05619582533836365,
+      0.056196022778749466,
+      0.05619630962610245,
+      0.05619623512029648,
+      0.05624135583639145,
+      0.05620087310671806,
+      0.05619524419307709,
+      0.056197285652160645,
+      0.05619876831769943,
+      0.05619945749640465,
+      0.05619756132364273,
+      0.05619775131344795,
+      0.056195855140686035,
+      0.0561944879591465,
+      0.05619751662015915,
+      0.05619637668132782,
+      0.05619584023952484,
+      0.056196652352809906,
+      0.056194957345724106,
+      0.056196365505456924,
+      0.056197553873062134
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.6502658724784851,
+      0.2811811864376068,
+      0.2346016764640808,
+      0.21476693451404572,
+      0.18344977498054504,
+      0.10495298355817795,
+      0.08588869124650955,
+      0.018477406352758408,
+      0.0006744824349880219,
+      0.0031205001287162304,
+      0.00575696025043726,
+      0.010777078568935394,
+      0.00391048938035965,
+      0.004867402836680412,
+      0.0020913775078952312,
+      0.008061191067099571,
+      0.0064541189931333065,
+      0.011848259717226028,
+      0.014426968060433865,
+      0.02709692344069481,
+      0.02255256287753582,
+      0.02122652158141136,
+      0.032681021839380264,
+      0.03130153566598892,
+      0.036686282604932785,
+      0.045077789574861526,
+      0.050404731184244156,
+      0.06439576297998428,
+      0.07356862723827362,
+      0.07624772191047668,
+      0.0712992250919342,
+      0.08815018087625504,
+      0.10727302730083466,
+      0.12698137760162354,
+      0.13273373246192932,
+      0.13109514117240906,
+      0.1526988297700882,
+      0.168387308716774,
+      0.20321153104305267,
+      0.2037270963191986,
+      0.23864588141441345,
+      0.2625564932823181,
+      0.25588732957839966,
+      0.31165599822998047,
+      0.3387743830680847,
+      0.3405305743217468,
+      0.31628188490867615,
+      0.35578322410583496,
+      0.333683043718338,
+      0.3485804498195648,
+      0.3498348593711853,
+      0.35685494542121887,
+      0.3422837555408478,
+      0.33916112780570984,
+      0.3598865270614624,
+      0.3872186839580536,
+      0.34882667660713196,
+      0.364061564207077,
+      0.34618881344795227,
+      0.3438187837600708,
+      0.33586207032203674,
+      0.3409693241119385,
+      0.3413761556148529,
+      0.3447628617286682,
+      0.32198095321655273,
+      0.32699164748191833,
+      0.3171702027320862,
+      0.343483567237854,
+      0.34123876690864563,
+      0.34286391735076904,
+      0.35002151131629944,
+      0.35763251781463623,
+      0.35160672664642334,
+      0.35453924536705017,
+      0.35218292474746704,
+      0.3646450638771057,
+      0.37005844712257385,
+      0.3746702969074249,
+      0.3777448236942291,
+      0.38459891080856323,
+      0.3768923282623291,
+      0.35704317688941956,
+      0.35337740182876587,
+      0.3574754595756531,
+      0.3560780882835388,
+      0.37906306982040405,
+      0.3762606978416443,
+      0.37327563762664795,
+      0.38858166337013245,
+      0.3942738473415375,
+      0.41465315222740173,
+      0.42790403962135315,
+      0.4204115867614746,
+      0.44135281443595886,
+      0.45378515124320984,
+      0.45165520906448364,
+      0.4539830684661865,
+      0.48027288913726807,
+      0.47597694396972656,
+      0.47761377692222595,
+      0.48030611872673035,
+      0.48506608605384827,
+      0.4838334619998932,
+      0.4895510673522949,
+      0.49041271209716797,
+      0.48350363969802856,
+      0.48788440227508545,
+      0.5029377341270447,
+      0.505985677242279,
+      0.5064969658851624,
+      0.5154033899307251,
+      0.5173416137695312,
+      0.5221923589706421,
+      0.5203668475151062,
+      0.5313696265220642,
+      0.5403210520744324,
+      0.5236276388168335,
+      0.5321092009544373,
+      0.5334134697914124,
+      0.5349195599555969,
+      0.5409315228462219,
+      0.518031895160675,
+      0.5159770250320435,
+      0.5262935161590576,
+      0.5362329483032227,
+      0.5446694493293762,
+      0.5397369861602783,
+      0.5460827350616455,
+      0.5383400321006775,
+      0.5269708633422852,
+      0.4982045292854309,
+      0.48877066373825073,
+      0.5075152516365051,
+      0.5103316903114319,
+      0.5342748165130615,
+      0.5479627847671509,
+      0.5441040396690369,
+      0.5527230501174927,
+      0.5749324560165405,
+      0.5673231482505798,
+      0.5627517700195312,
+      0.5668912529945374,
+      0.569943368434906,
+      0.5715296864509583,
+      0.5803252458572388,
+      0.5838395357131958,
+      0.606804370880127,
+      0.6081652045249939,
+      0.6045222282409668,
+      0.606292724609375,
+      0.6084168553352356,
+      0.6058701872825623,
+      0.6053699851036072,
+      0.5967050790786743,
+      0.6303905248641968,
+      0.6074860095977783,
+      0.6097574234008789,
+      0.6100372672080994,
+      0.621992826461792,
+      0.6150597929954529,
+      0.6282275915145874,
+      0.6282679438591003,
+      0.6219140887260437,
+      0.6237756609916687,
+      0.6266230940818787,
+      0.6319355368614197,
+      0.6140732169151306,
+      0.6123427748680115,
+      0.5977766513824463,
+      0.5882521867752075,
+      0.6038938760757446,
+      0.6030605435371399,
+      0.5987586975097656,
+      0.5784411430358887,
+      0.5886924266815186,
+      0.5950058102607727,
+      0.6019317507743835,
+      0.6207942366600037,
+      0.6202986836433411,
+      0.6206125020980835,
+      0.6213878393173218,
+      0.6198179721832275,
+      0.6370460987091064,
+      0.6362895369529724,
+      0.631364643573761,
+      0.6306073665618896,
+      0.6332530975341797,
+      0.6333451271057129,
+      0.6305362582206726,
+      0.6334693431854248,
+      0.626459538936615,
+      0.6242359280586243,
+      0.6261436939239502,
+      0.6296249628067017,
+      0.6298355460166931,
+      0.6314758658409119,
+      0.6265407800674438,
+      0.6235003471374512,
+      0.6169945001602173,
+      0.6198533773422241,
+      0.6230413913726807,
+      0.6242404580116272,
+      0.6321998238563538,
+      0.6288551092147827,
+      0.6274933218955994,
+      0.6307497620582581,
+      0.6272670030593872,
+      0.6341568827629089,
+      0.6304832100868225,
+      0.6272768378257751,
+      0.6211402416229248,
+      0.6236705780029297,
+      0.6292702555656433,
+      0.6233572363853455,
+      0.6246389150619507,
+      0.6268891096115112,
+      0.6349132657051086,
+      0.605574905872345,
+      0.6070116758346558,
+      0.6122633814811707,
+      0.6229851245880127,
+      0.6252979636192322,
+      0.6241100430488586,
+      0.6217414736747742,
+      0.627311646938324,
+      0.6314742565155029,
+      0.6266555786132812,
+      0.6285045146942139,
+      0.6282162070274353,
+      0.6257549524307251,
+      0.6248719692230225,
+      0.6256452798843384,
+      0.6277334094047546,
+      0.6294560432434082,
+      0.6282111406326294,
+      0.6301722526550293,
+      0.6323681473731995,
+      0.6334131360054016,
+      0.631639838218689,
+      0.6324410438537598,
+      0.6289135217666626,
+      0.6276329159736633,
+      0.6267264485359192,
+      0.6281037330627441,
+      0.6255887150764465,
+      0.6275599002838135,
+      0.6262612342834473,
+      0.6259809732437134,
+      0.6265996098518372,
+      0.6338443160057068,
+      0.6319806575775146,
+      0.635006844997406,
+      0.6332499980926514,
+      0.634567141532898,
+      0.6330824494361877,
+      0.6282050013542175,
+      0.6263772249221802,
+      0.6274346113204956,
+      0.626335859298706,
+      0.6266767978668213,
+      0.6246297955513,
+      0.6254687905311584,
+      0.6249617338180542,
+      0.6263309717178345,
+      0.6262295246124268,
+      0.6264058947563171,
+      0.6264864802360535,
+      0.6223540306091309,
+      0.6259410977363586,
+      0.6367257833480835,
+      0.6359933614730835,
+      0.6366913318634033,
+      0.636050283908844,
+      0.6368836760520935,
+      0.63709557056427,
+      0.6359667181968689,
+      0.6363386511802673,
+      0.6391788721084595,
+      0.6415848731994629,
+      0.6400963664054871,
+      0.6401890516281128,
+      0.6399085521697998,
+      0.6401390433311462,
+      0.6399528384208679,
+      0.6384453177452087,
+      0.6376702189445496,
+      0.636957049369812,
+      0.6382962465286255,
+      0.6389310359954834,
+      0.6387096047401428,
+      0.6374481320381165,
+      0.6395060420036316,
+      0.6384611129760742,
+      0.638874888420105,
+      0.638683021068573,
+      0.6387286186218262,
+      0.639114260673523,
+      0.6400926113128662,
+      0.6399978399276733,
+      0.6397255659103394,
+      0.6396890878677368,
+      0.6395076513290405,
+      0.639596700668335,
+      0.6389221549034119,
+      0.6387419104576111,
+      0.6388429999351501,
+      0.6392461657524109,
+      0.6398524045944214,
+      0.6402888298034668,
+      0.6405676603317261
+    ],
+    "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": [
+      0.04,
+      0.79,
+      0.99,
+      1.0,
+      0.99,
+      1.0,
+      0.99,
+      0.99,
+      0.99,
+      1.0
+    ]
+  },
+  "final_accuracy": 0.99875,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9983471074380166,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9940828402366864,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9923076923076923,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.91,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 0.9795918367346939,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.956989247311828,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.75,
+        "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": 0.86875,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.9706840390879479,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.57,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.69,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.86,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.6,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.62,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.83,
+            "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": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9792746113989638,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.92,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.99,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.98046875,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.9930555555555556,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.9,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9705882352941176,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.979381443298969,
+            "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": 0.9914529914529915,
+            "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": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.996415770609319,
+            "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": 0.93,
+        "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": 0.9530201342281879,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.62,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.7,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.9533333333333334,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.49,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.875,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.99,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9983333333333333,
+            "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": 0.9,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.99,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9644670050761421,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.9,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9595141700404858,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.9934640522875817,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.89,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9697986577181208,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.9900990099009901,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.8833333333333333,
+      "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": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.99,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.9966666666666667,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.99,
+        "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": 0.9975,
+            "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": 0.99875,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 0.99875,
+  "sft_overall_accuracy": 0.8833333333333333
+}

add_sub_sorl_v1_abs2_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3139180f867078da4cdae7e87fa4be93f25a3704718ec0ff77e4996d232a489
+size 650271004

add_sub_sorl_v1_abs2_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_abs2_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": 2,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162491102,
+  "run_name": "add_sub_sorl_v1_abs2_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T17:28:05.950563+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": "9339oiey",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/9339oiey",
+  "final_accuracy": 0.99875,
+  "sft_accuracy": 0.8833333333333333,
+  "eval_method": "ArithmeticEvaluator"
+}