Delete folder add_sub_sorl_v1_abs10_K1_50K_1L3H510d with huggingface_hub

add_sub_sorl_v1_abs10_K1_50K_1L3H510d/config.json

@@ -1,36 +0,0 @@
-{
-  "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"
-  ],
-  "max_position_embeddings": 128,
-  "max_window_layers": 28,
-  "model_type": "qwen3",
-  "num_attention_heads": 3,
-  "num_hidden_layers": 1,
-  "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": 151654
-}

add_sub_sorl_v1_abs10_K1_50K_1L3H510d/generation_config.json

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

add_sub_sorl_v1_abs10_K1_50K_1L3H510d/metrics.json

@@ -1,2257 +0,0 @@
-{
-  "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
-    ],
-    "loss": [
-      8.47067642211914,
-      3.859321117401123,
-      3.0991883277893066,
-      3.3409392833709717,
-      3.3445615768432617,
-      3.071617603302002,
-      3.0536575317382812,
-      2.8200273513793945,
-      2.7634453773498535,
-      0.6405220031738281,
-      -0.9271819591522217,
-      -4.679404258728027,
-      -5.836901664733887,
-      -8.526687622070312,
-      -8.978545188903809,
-      -8.417521476745605,
-      -7.670687198638916,
-      -6.564399719238281,
-      -5.615384578704834,
-      -4.976064682006836,
-      -2.216883420944214,
-      -1.4136266708374023,
-      -1.0798640251159668,
-      -0.5408358573913574,
-      -0.22238361835479736,
-      0.16258960962295532,
-      -0.6259200572967529,
-      -0.5599202513694763,
-      0.13963162899017334,
-      0.007780313491821289,
-      -0.23957473039627075,
-      0.3784700632095337,
-      0.09624123573303223,
-      -0.2628714442253113,
-      0.1245647668838501,
-      -0.2984371781349182,
-      0.07317012548446655,
-      -0.20061078667640686,
-      0.0016877055168151855,
-      -0.08852306008338928,
-      0.007971584796905518,
-      -0.005069524049758911,
-      -0.1857169270515442,
-      0.12880273163318634,
-      -0.15553274750709534,
-      -0.31286513805389404,
-      0.004276543855667114,
-      0.043737590312957764,
-      -0.20392894744873047,
-      -0.06425321102142334,
-      0.042032390832901,
-      0.13489098846912384,
-      -0.1695605218410492,
-      -0.18983259797096252,
-      0.05254027247428894,
-      -0.17216861248016357,
-      0.42946910858154297,
-      0.03497058153152466,
-      -0.4959927201271057,
-      -0.2946988046169281,
-      -0.14054325222969055,
-      -0.1606135368347168,
-      -0.20758064091205597,
-      -0.07599818706512451,
-      -0.204355850815773,
-      -0.4362576901912689,
-      -0.42311590909957886,
-      -0.40590041875839233,
-      -0.16058385372161865,
-      -0.24916203320026398,
-      0.16023975610733032,
-      -0.2869209051132202,
-      -0.3760267496109009,
-      -0.3357478082180023,
-      -0.09387531876564026,
-      -0.7844890356063843,
-      -0.13949094712734222,
-      -0.44414588809013367,
-      -0.42745646834373474,
-      -0.3531284034252167,
-      -0.6143267154693604,
-      -0.9826579689979553,
-      -0.4692939221858978,
-      -0.3503918945789337,
-      -0.42758840322494507,
-      -0.3714824318885803,
-      -0.31439173221588135,
-      -0.6495652198791504,
-      -0.058341529220342636,
-      -0.6107015013694763,
-      -0.5336044430732727,
-      -0.4551495909690857,
-      -0.6802051067352295,
-      -0.3394930064678192,
-      -0.47552600502967834,
-      -0.01830144040286541,
-      -0.30640774965286255,
-      -0.5638172030448914,
-      -0.40761300921440125,
-      -0.3676009178161621,
-      -0.33406054973602295,
-      -0.32963287830352783,
-      -0.41975510120391846,
-      -0.4875797927379608,
-      -0.13679492473602295,
-      -0.3449482321739197,
-      -0.38658419251441956,
-      -0.5263659358024597,
-      -0.5405909419059753,
-      -0.5471492409706116,
-      -0.30939704179763794,
-      -0.6842730045318604,
-      -0.539143443107605,
-      -0.4582570791244507,
-      -0.4487195312976837,
-      -0.5894002318382263,
-      -0.07285723090171814,
-      -0.749821126461029,
-      -0.4640003442764282,
-      -0.42504963278770447,
-      -0.2539079785346985,
-      -0.4926801323890686,
-      -0.4174199104309082,
-      -0.38921570777893066,
-      -0.3867590129375458,
-      -0.3553716838359833,
-      -0.4733166992664337,
-      -0.32238879799842834,
-      -0.6644755005836487,
-      -0.5403436422348022,
-      -0.560452938079834,
-      -0.39427629113197327,
-      -0.39716440439224243,
-      -0.3601585328578949,
-      -0.5065312385559082,
-      -0.30354976654052734,
-      -0.528386652469635,
-      -0.4692227244377136,
-      -0.33632326126098633,
-      -0.4119393825531006,
-      -0.3576270639896393,
-      -0.704921543598175,
-      -0.6248269081115723,
-      -0.3194754123687744,
-      -0.3632478415966034,
-      -0.4672422707080841,
-      -0.710575520992279,
-      -0.33513766527175903,
-      -0.2928919494152069,
-      -0.4346057176589966
-    ],
-    "base_loss": [
-      6.529869079589844,
-      2.3522889614105225,
-      1.9823862314224243,
-      1.8216798305511475,
-      1.8954092264175415,
-      1.863158106803894,
-      1.8727132081985474,
-      1.8258191347122192,
-      1.8073500394821167,
-      1.8966313600540161,
-      1.8717511892318726,
-      1.9207693338394165,
-      1.8693705797195435,
-      1.9036433696746826,
-      1.8812330961227417,
-      1.6870934963226318,
-      1.6692317724227905,
-      1.4732754230499268,
-      1.365601658821106,
-      1.2771196365356445,
-      0.9189826846122742,
-      0.7958611249923706,
-      0.7418918013572693,
-      0.681869387626648,
-      0.6421297192573547,
-      0.5580056309700012,
-      0.6352177858352661,
-      0.6491341590881348,
-      0.5735772848129272,
-      0.5147913098335266,
-      0.533217191696167,
-      0.47872549295425415,
-      0.48013827204704285,
-      0.5564236640930176,
-      0.49468594789505005,
-      0.4692336618900299,
-      0.5063740611076355,
-      0.48869970440864563,
-      0.49136045575141907,
-      0.44452375173568726,
-      0.4514934718608856,
-      0.45978498458862305,
-      0.39197835326194763,
-      0.4396274983882904,
-      0.42984989285469055,
-      0.4332329332828522,
-      0.4062458574771881,
-      0.37813684344291687,
-      0.3858397305011749,
-      0.3725948929786682,
-      0.4058559834957123,
-      0.2884959578514099,
-      0.3678801953792572,
-      0.336038738489151,
-      0.3166891038417816,
-      0.33307892084121704,
-      0.2845018208026886,
-      0.3227593004703522,
-      0.37250086665153503,
-      0.28964105248451233,
-      0.3191569149494171,
-      0.27697229385375977,
-      0.29193589091300964,
-      0.2866390347480774,
-      0.3261076807975769,
-      0.3937642574310303,
-      0.31407326459884644,
-      0.33535775542259216,
-      0.24889574944972992,
-      0.29406002163887024,
-      0.2409432828426361,
-      0.35092687606811523,
-      0.27360251545906067,
-      0.28033000230789185,
-      0.2864038050174713,
-      0.271339476108551,
-      0.2604587972164154,
-      0.30055779218673706,
-      0.27918002009391785,
-      0.22688111662864685,
-      0.270652711391449,
-      0.28921690583229065,
-      0.23527908325195312,
-      0.25206416845321655,
-      0.25411903858184814,
-      0.21951384842395782,
-      0.27349406480789185,
-      0.254742294549942,
-      0.2292661964893341,
-      0.1955949366092682,
-      0.24696041643619537,
-      0.22090299427509308,
-      0.22282291948795319,
-      0.20417501032352448,
-      0.2358027845621109,
-      0.19285988807678223,
-      0.1971234232187271,
-      0.22451014816761017,
-      0.18633778393268585,
-      0.22210939228534698,
-      0.23272477090358734,
-      0.19261686503887177,
-      0.21030089259147644,
-      0.1925947368144989,
-      0.22401663661003113,
-      0.18720892071723938,
-      0.17757317423820496,
-      0.18794147670269012,
-      0.22906138002872467,
-      0.21192224323749542,
-      0.16107645630836487,
-      0.22859220206737518,
-      0.20563936233520508,
-      0.21756412088871002,
-      0.1816309690475464,
-      0.18884612619876862,
-      0.14128300547599792,
-      0.2663905918598175,
-      0.19550541043281555,
-      0.17163562774658203,
-      0.17473351955413818,
-      0.1715536117553711,
-      0.17578543722629547,
-      0.16855619847774506,
-      0.18263652920722961,
-      0.17694807052612305,
-      0.13545209169387817,
-      0.1860579401254654,
-      0.20312993228435516,
-      0.19952358305454254,
-      0.21564070880413055,
-      0.17865024507045746,
-      0.22009076178073883,
-      0.15182125568389893,
-      0.17367124557495117,
-      0.17723344266414642,
-      0.159737691283226,
-      0.1377307027578354,
-      0.1133374497294426,
-      0.1590133011341095,
-      0.13635817170143127,
-      0.2107985019683838,
-      0.20057733356952667,
-      0.12804178893566132,
-      0.1539105921983719,
-      0.15954028069972992,
-      0.1959955245256424,
-      0.10784348100423813,
-      0.1844562292098999,
-      0.14684811234474182
-    ],
-    "info_loss": [
-      -0.2673468589782715,
-      -0.04440808296203613,
-      -0.07712435722351074,
-      -0.03570270538330078,
-      -0.04258620738983154,
-      -0.06644809246063232,
-      -0.06896436214447021,
-      -0.08776187896728516,
-      -0.09168505668640137,
-      -0.3129291534423828,
-      -0.46592533588409424,
-      -0.8345524072647095,
-      -0.9259849190711975,
-      -1.1986396312713623,
-      -1.2376453876495361,
-      -1.1589113473892212,
-      -1.0768587589263916,
-      -0.9370661377906799,
-      -0.8276593685150146,
-      -0.7481008768081665,
-      -0.4281192123889923,
-      -0.32412827014923096,
-      -0.2727624773979187,
-      -0.20998990535736084,
-      -0.16334781050682068,
-      -0.11577355861663818,
-      -0.19545629620552063,
-      -0.19052308797836304,
-      -0.10945683717727661,
-      -0.10963675379753113,
-      -0.13540121912956238,
-      -0.0649571418762207,
-      -0.09443607926368713,
-      -0.13581201434135437,
-      -0.09144818782806396,
-      -0.12629523873329163,
-      -0.09346094727516174,
-      -0.1159842312335968,
-      -0.09238475561141968,
-      -0.09163549542427063,
-      -0.0864732563495636,
-      -0.09146097302436829,
-      -0.1021280288696289,
-      -0.06953030824661255,
-      -0.09440380334854126,
-      -0.10704010725021362,
-      -0.07827016711235046,
-      -0.06915175914764404,
-      -0.09524726867675781,
-      -0.08362510800361633,
-      -0.07066857814788818,
-      -0.04639166593551636,
-      -0.08692899346351624,
-      -0.08005630970001221,
-      -0.05635380744934082,
-      -0.07672905921936035,
-      -0.019900262355804443,
-      -0.054761648178100586,
-      -0.11633443832397461,
-      -0.08680947124958038,
-      -0.0727926641702652,
-      -0.06547568738460541,
-      -0.07120101153850555,
-      -0.055115923285484314,
-      -0.0648864209651947,
-      -0.09370043873786926,
-      -0.08160407841205597,
-      -0.08273720741271973,
-      -0.048524439334869385,
-      -0.06075708568096161,
-      -0.015704527497291565,
-      -0.06819471716880798,
-      -0.06899033486843109,
-      -0.06627984344959259,
-      -0.042593181133270264,
-      -0.10910588502883911,
-      -0.046684086322784424,
-      -0.07880324125289917,
-      -0.07517123222351074,
-      -0.06181502342224121,
-      -0.09271721541881561,
-      -0.13140380382537842,
-      -0.07391722500324249,
-      -0.0640488713979721,
-      -0.07120409607887268,
-      -0.0627627968788147,
-      -0.06179974973201752,
-      -0.09381449222564697,
-      -0.03218841552734375,
-      -0.08402406424283981,
-      -0.0813249945640564,
-      -0.07089047133922577,
-      -0.09299984574317932,
-      -0.05727653205394745,
-      -0.07365383207798004,
-      -0.024941951036453247,
-      -0.05313184857368469,
-      -0.08307541906833649,
-      -0.06305523216724396,
-      -0.06241106986999512,
-      -0.06006002426147461,
-      -0.05616569519042969,
-      -0.0658143013715744,
-      -0.07124374061822891,
-      -0.03855186700820923,
-      -0.05644591152667999,
-      -0.05963750183582306,
-      -0.07424025982618332,
-      -0.07966555655002594,
-      -0.07849693298339844,
-      -0.05011218041181564,
-      -0.09463152289390564,
-      -0.07753230631351471,
-      -0.07019945979118347,
-      -0.06616665422916412,
-      -0.0811585858464241,
-      -0.02353697270154953,
-      -0.10485921800136566,
-      -0.0684717446565628,
-      -0.061824582517147064,
-      -0.045364245772361755,
-      -0.06856260448694229,
-      -0.06233055889606476,
-      -0.058004721999168396,
-      -0.05994530767202377,
-      -0.05602298676967621,
-      -0.06310290843248367,
-      -0.05322764813899994,
-      -0.08953020721673965,
-      -0.07597177475690842,
-      -0.07998061180114746,
-      -0.05991140753030777,
-      -0.06403300166130066,
-      -0.05338875204324722,
-      -0.07013767212629318,
-      -0.050085484981536865,
-      -0.07169508934020996,
-      -0.0626421645283699,
-      -0.04760565608739853,
-      -0.05929655581712723,
-      -0.051652222871780396,
-      -0.09364695101976395,
-      -0.08479366451501846,
-      -0.04709339886903763,
-      -0.053751140832901,
-      -0.06505024433135986,
-      -0.09360753744840622,
-      -0.04723042622208595,
-      -0.05039326846599579,
-      -0.06148739159107208
-    ],
-    "abs_loss": [
-      2.0777323246002197,
-      1.8372353315353394,
-      1.8596841096878052,
-      1.8447657823562622,
-      1.866260051727295,
-      1.8622106313705444,
-      1.8444541692733765,
-      1.8505001068115234,
-      1.886319637298584,
-      1.8786344528198242,
-      1.7852767705917358,
-      1.580426573753357,
-      1.4302619695663452,
-      1.4253687858581543,
-      1.3955053091049194,
-      1.238847255706787,
-      1.0613170862197876,
-      0.9975995421409607,
-      1.0350087881088257,
-      0.8302550315856934,
-      0.7335612177848816,
-      0.6067109704017639,
-      0.5130870938301086,
-      0.4377255439758301,
-      0.40763750672340393,
-      0.3429541289806366,
-      0.35336068272590637,
-      0.35311901569366455,
-      0.30836349725723267,
-      0.29843878746032715,
-      0.2701031267642975,
-      0.2620505690574646,
-      0.27157625555992126,
-      0.26693102717399597,
-      0.27656325697898865,
-      0.2799464166164398,
-      0.27106282114982605,
-      0.32633912563323975,
-      0.29606541991233826,
-      0.2582005262374878,
-      0.2775072157382965,
-      0.2881506681442261,
-      0.3060077130794525,
-      0.3220145106315613,
-      0.2445584088563919,
-      0.2725384533405304,
-      0.35818028450012207,
-      0.3085358142852783,
-      0.2886260151863098,
-      0.32749247550964355,
-      0.3170374631881714,
-      0.32663360238075256,
-      0.3656320571899414,
-      0.38276636600494385,
-      0.4017406404018402,
-      0.3261677026748657,
-      0.30932173132896423,
-      0.34228047728538513,
-      0.3847839832305908,
-      0.3398013114929199,
-      0.2953852117061615,
-      0.285983681678772,
-      0.2711770236492157,
-      0.30411458015441895,
-      0.25742125511169434,
-      0.25713059306144714,
-      0.21089105308055878,
-      0.25374868512153625,
-      0.20307719707489014,
-      0.24056212604045868,
-      0.27503547072410583,
-      0.21771258115768433,
-      0.17491371929645538,
-      0.19084621965885162,
-      0.21915553510189056,
-      0.14147557318210602,
-      0.16538940370082855,
-      0.16390661895275116,
-      0.17348627746105194,
-      0.14653095602989197,
-      0.145098015666008,
-      0.16084204614162445,
-      0.13867895305156708,
-      0.11408242583274841,
-      0.10716051608324051,
-      0.11975974589586258,
-      0.11440757662057877,
-      0.1023651733994484,
-      0.13637660443782806,
-      0.10093352198600769,
-      0.09186253696680069,
-      0.11327973753213882,
-      0.12308790534734726,
-      0.1159990057349205,
-      0.10311237722635269,
-      0.11532590538263321,
-      0.09129684418439865,
-      0.09231492877006531,
-      0.11006152629852295,
-      0.09338565915822983,
-      0.10004422813653946,
-      0.09749677032232285,
-      0.0867953896522522,
-      0.09206672757863998,
-      0.07050669193267822,
-      0.07319243252277374,
-      0.09475291520357132,
-      0.09280968457460403,
-      0.07203791290521622,
-      0.07556378096342087,
-      0.0879942774772644,
-      0.07373198121786118,
-      0.09719672799110413,
-      0.08441676944494247,
-      0.08118002861738205,
-      0.06300904601812363,
-      0.06390989571809769,
-      0.08405200392007828,
-      0.06970033794641495,
-      0.06256341934204102,
-      0.0618957094848156,
-      0.05331374332308769,
-      0.07108704745769501,
-      0.050225283950567245,
-      0.07027151435613632,
-      0.06590838730335236,
-      0.050113528966903687,
-      0.06628331542015076,
-      0.06416907906532288,
-      0.06008912995457649,
-      0.08411464840173721,
-      0.05974702164530754,
-      0.05712446570396423,
-      0.04810317978262901,
-      0.05578098073601723,
-      0.04661886766552925,
-      0.06548946350812912,
-      0.05453115701675415,
-      0.06402477622032166,
-      0.047158390283584595,
-      0.05706872418522835,
-      0.051012784242630005,
-      0.05519309639930725,
-      0.05584952235221863,
-      0.045132603496313095,
-      0.056113243103027344,
-      0.05444992706179619,
-      0.05939311161637306,
-      0.058004170656204224,
-      0.03935953974723816
-    ],
-    "zipf_loss": [
-      4.406503200531006,
-      1.7673894166946411,
-      1.7020772695541382,
-      1.691809892654419,
-      1.688388466835022,
-      1.6867194175720215,
-      1.6861424446105957,
-      1.686776876449585,
-      1.6843140125274658,
-      1.6853187084197998,
-      1.6817922592163086,
-      1.5873081684112549,
-      1.4105513095855713,
-      1.413528323173523,
-      1.3771247863769531,
-      1.3606131076812744,
-      1.3225369453430176,
-      1.2332267761230469,
-      1.1921072006225586,
-      1.1447988748550415,
-      1.071969985961914,
-      0.9711239337921143,
-      0.8545601963996887,
-      0.8334212303161621,
-      0.7282010912895203,
-      0.7280241250991821,
-      0.6580890417098999,
-      0.660864531993866,
-      0.6297863721847534,
-      0.559512734413147,
-      0.5542100071907043,
-      0.5231109261512756,
-      0.5333061218261719,
-      0.5121319890022278,
-      0.5167043805122375,
-      0.4672868847846985,
-      0.47429925203323364,
-      0.4378978908061981,
-      0.40456825494766235,
-      0.35748809576034546,
-      0.3934599459171295,
-      0.4209401607513428,
-      0.4129841923713684,
-      0.35227686166763306,
-      0.33419957756996155,
-      0.2970491647720337,
-      0.3449143171310425,
-      0.32626476883888245,
-      0.3338414430618286,
-      0.3666537404060364,
-      0.3111584484577179,
-      0.27764832973480225,
-      0.29528602957725525,
-      0.23641511797904968,
-      0.25921517610549927,
-      0.2294262945652008,
-      0.3130377531051636,
-      0.22559970617294312,
-      0.2563723921775818,
-      0.2497747242450714,
-      0.2386879324913025,
-      0.1885727047920227,
-      0.18537591397762299,
-      0.1581105887889862,
-      0.09265856444835663,
-      0.08126936107873917,
-      0.05776248127222061,
-      0.06073905527591705,
-      0.055457063019275665,
-      0.04029260575771332,
-      0.04883820563554764,
-      0.02232813462615013,
-      0.022782735526561737,
-      0.027635956183075905,
-      0.023737134411931038,
-      0.02108277752995491,
-      0.05035217106342316,
-      0.02693808451294899,
-      0.02772720716893673,
-      0.023487640544772148,
-      0.02768293395638466,
-      0.026078997179865837,
-      0.02073132060468197,
-      0.02662445418536663,
-      0.019617483019828796,
-      0.024655727669596672,
-      0.01867099106311798,
-      0.023600950837135315,
-      0.02063876762986183,
-      0.02385086938738823,
-      0.02349885366857052,
-      0.021524202078580856,
-      0.014661639928817749,
-      0.01749744638800621,
-      0.014898284338414669,
-      0.026725592091679573,
-      0.01865765079855919,
-      0.03319530561566353,
-      0.025595368817448616,
-      0.025061847642064095,
-      0.02381046488881111,
-      0.029657501727342606,
-      0.019407516345381737,
-      0.02305617742240429,
-      0.017656439915299416,
-      0.02498270571231842,
-      0.02274232730269432,
-      0.018814217299222946,
-      0.019799407571554184,
-      0.018341470509767532,
-      0.0218488872051239,
-      0.026076773181557655,
-      0.020820580422878265,
-      0.017731718719005585,
-      0.023198017850518227,
-      0.02703857235610485,
-      0.014838501811027527,
-      0.023975318297743797,
-      0.018241601064801216,
-      0.01530423853546381,
-      0.018811386078596115,
-      0.01606089621782303,
-      0.022991515696048737,
-      0.017252791672945023,
-      0.02303040400147438,
-      0.0213193129748106,
-      0.017248917371034622,
-      0.01720144785940647,
-      0.02127968519926071,
-      0.013841651380062103,
-      0.015300977975130081,
-      0.020212795585393906,
-      0.01736241951584816,
-      0.017097417265176773,
-      0.015596166253089905,
-      0.01540975272655487,
-      0.02227761410176754,
-      0.014015071094036102,
-      0.0199933722615242,
-      0.01729702018201351,
-      0.016830110922455788,
-      0.01564820483326912,
-      0.0170130655169487,
-      0.01783180795609951,
-      0.015839725732803345,
-      0.018108535557985306,
-      0.024059370160102844,
-      0.023383792489767075,
-      0.020784102380275726,
-      0.029484093189239502
-    ],
-    "denoise_loss": [],
-    "ortho_loss": [
-      0.39830857515335083,
-      0.2433786690235138,
-      0.20988474786281586,
-      0.19384649395942688,
-      0.19618169963359833,
-      0.21874834597110748,
-      0.20772071182727814,
-      0.18882158398628235,
-      0.2314612716436386,
-      0.25052326917648315,
-      0.2812398374080658,
-      0.3191724717617035,
-      0.3201376497745514,
-      0.3000963032245636,
-      0.26885080337524414,
-      0.268261581659317,
-      0.2745248079299927,
-      0.2682482898235321,
-      0.27149900794029236,
-      0.2707764804363251,
-      0.27085885405540466,
-      0.2695680856704712,
-      0.2618236243724823,
-      0.2597876489162445,
-      0.2581340968608856,
-      0.2508399486541748,
-      0.24459116160869598,
-      0.24822774529457092,
-      0.2757481336593628,
-      0.2812578082084656,
-      0.28060707449913025,
-      0.27701568603515625,
-      0.2774236798286438,
-      0.27818116545677185,
-      0.27627792954444885,
-      0.2726292312145233,
-      0.2671484351158142,
-      0.26871195435523987,
-      0.2720329463481903,
-      0.27176108956336975,
-      0.27438652515411377,
-      0.2732361853122711,
-      0.2705092430114746,
-      0.271069198846817,
-      0.27583789825439453,
-      0.2766655385494232,
-      0.27713289856910706,
-      0.27564698457717896,
-      0.2780587077140808,
-      0.278008371591568,
-      0.2802888751029968,
-      0.28178492188453674,
-      0.2823565602302551,
-      0.28541451692581177,
-      0.29519784450531006,
-      0.2944939136505127,
-      0.2898300886154175,
-      0.2950134873390198,
-      0.2994725704193115,
-      0.3095267415046692,
-      0.3123369514942169,
-      0.3123491704463959,
-      0.31109243631362915,
-      0.3140035569667816,
-      0.3158407211303711,
-      0.3229905664920807,
-      0.3247292637825012,
-      0.32620954513549805,
-      0.32591450214385986,
-      0.32379913330078125,
-      0.3245965540409088,
-      0.32910656929016113,
-      0.3264192044734955,
-      0.3264644145965576,
-      0.32559576630592346,
-      0.3237476944923401,
-      0.3267005980014801,
-      0.33368119597435,
-      0.33319106698036194,
-      0.3357834815979004,
-      0.33304426074028015,
-      0.32880592346191406,
-      0.3296840190887451,
-      0.3254052698612213,
-      0.31745028495788574,
-      0.3167537748813629,
-      0.3174743354320526,
-      0.3163991868495941,
-      0.3160884976387024,
-      0.3137410283088684,
-      0.31337714195251465,
-      0.31452980637550354,
-      0.31249645352363586,
-      0.3148419260978699,
-      0.31133177876472473,
-      0.30948013067245483,
-      0.31456074118614197,
-      0.31482842564582825,
-      0.31413134932518005,
-      0.3161856532096863,
-      0.31791284680366516,
-      0.3175433874130249,
-      0.3201428949832916,
-      0.31951627135276794,
-      0.3191130757331848,
-      0.32220661640167236,
-      0.32215243577957153,
-      0.32431453466415405,
-      0.3254556953907013,
-      0.3261330723762512,
-      0.33019202947616577,
-      0.33232834935188293,
-      0.3316073417663574,
-      0.3316960632801056,
-      0.3329588770866394,
-      0.33336374163627625,
-      0.3353458046913147,
-      0.3332311511039734,
-      0.33092865347862244,
-      0.33475038409233093,
-      0.33227968215942383,
-      0.3314722180366516,
-      0.33147549629211426,
-      0.32883259654045105,
-      0.3315986394882202,
-      0.33407390117645264,
-      0.33586472272872925,
-      0.33549898862838745,
-      0.3368147015571594,
-      0.3380911648273468,
-      0.33757898211479187,
-      0.33975356817245483,
-      0.34040141105651855,
-      0.34014108777046204,
-      0.34206295013427734,
-      0.34369945526123047,
-      0.34482741355895996,
-      0.3449490964412689,
-      0.3461775779724121,
-      0.3472226858139038,
-      0.3477751910686493,
-      0.3483361303806305,
-      0.3491489887237549,
-      0.3505137860774994,
-      0.35050302743911743,
-      0.350710391998291,
-      0.35074424743652344,
-      0.35159772634506226,
-      0.35225003957748413,
-      0.35208776593208313
-    ],
-    "lr": [
-      7.840000000000001e-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,
-      7.932818532818534e-05,
-      7.816988416988418e-05,
-      7.701158301158302e-05,
-      7.585328185328185e-05,
-      7.469498069498071e-05,
-      7.353667953667954e-05,
-      7.237837837837838e-05,
-      7.122007722007721e-05,
-      7.006177606177606e-05,
-      6.890347490347492e-05,
-      6.774517374517375e-05,
-      6.65868725868726e-05,
-      6.542857142857144e-05,
-      6.427027027027027e-05,
-      6.311196911196911e-05,
-      6.121235521235521e-05,
-      6.0054054054054064e-05,
-      5.8895752895752895e-05,
-      5.773745173745175e-05,
-      5.6579150579150584e-05,
-      5.542084942084943e-05,
-      5.426254826254825e-05,
-      5.310424710424711e-05,
-      5.194594594594594e-05,
-      5.0787644787644786e-05,
-      4.9629343629343644e-05,
-      4.8471042471042475e-05,
-      4.7312741312741326e-05,
-      4.615444015444014e-05,
-      4.4996138996139e-05,
-      4.309652509652511e-05,
-      4.1938223938223946e-05,
-      4.07799227799228e-05,
-      3.962162162162162e-05,
-      3.846332046332047e-05,
-      3.73050193050193e-05,
-      3.6146718146718155e-05,
-      3.4988416988416986e-05,
-      3.383011583011584e-05,
-      3.267181467181467e-05,
-      3.151351351351352e-05,
-      3.0355212355212367e-05,
-      2.9196911196911198e-05,
-      2.8038610038610046e-05,
-      2.6880308880308876e-05,
-      2.4980694980694983e-05,
-      2.3822393822393838e-05,
-      2.266409266409267e-05,
-      2.1505791505791517e-05,
-      2.0347490347490348e-05,
-      1.9189189189189195e-05,
-      1.8030888030888026e-05,
-      1.6872586872586878e-05,
-      1.571428571428571e-05,
-      1.455598455598456e-05,
-      1.3397683397683389e-05,
-      1.223938223938224e-05,
-      1.1081081081081092e-05,
-      9.92277992277992e-06,
-      8.764478764478772e-06
-    ],
-    "emb_lr": [],
-    "eval_step": [
-      750,
-      1532,
-      2314,
-      3096,
-      3878,
-      4660,
-      5442,
-      6224,
-      7006,
-      7788
-    ],
-    "eval_accuracy": [
-      0.01,
-      0.0,
-      0.0,
-      0.0,
-      0.0,
-      0.0,
-      0.0,
-      0.0,
-      0.0,
-      0.0
-    ]
-  },
-  "final_accuracy": 0.6078571428571429,
-  "sft_eval": {
-    "config": {
-      "ops": "add_sub",
-      "K": null,
-      "mode": "sft",
-      "n_digits": 6,
-      "n_per_split": 50
-    },
-    "splits": {
-      "add_S0": {
-        "full_accuracy": 0.88,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9796610169491525,
-            "count": 295
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 55
-          }
-        }
-      },
-      "add_S1": {
-        "full_accuracy": 0.76,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9761904761904762,
-            "count": 126
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 79
-          },
-          "SS": {
-            "accuracy": 0.9523809523809523,
-            "count": 21
-          },
-          "UC": {
-            "accuracy": 0.9354838709677419,
-            "count": 124
-          }
-        }
-      },
-      "add_S2": {
-        "full_accuracy": 0.4,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9866666666666667,
-            "count": 75
-          },
-          "SC": {
-            "accuracy": 0.9193548387096774,
-            "count": 62
-          },
-          "SS": {
-            "accuracy": 0.9230769230769231,
-            "count": 39
-          },
-          "UC": {
-            "accuracy": 0.7747747747747747,
-            "count": 111
-          },
-          "US": {
-            "accuracy": 0.9523809523809523,
-            "count": 63
-          }
-        }
-      },
-      "add_S3": {
-        "full_accuracy": 0.3,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9666666666666667,
-            "count": 60
-          },
-          "SC": {
-            "accuracy": 0.9473684210526315,
-            "count": 57
-          },
-          "SS": {
-            "accuracy": 0.9473684210526315,
-            "count": 19
-          },
-          "UC": {
-            "accuracy": 0.7019230769230769,
-            "count": 104
-          },
-          "US": {
-            "accuracy": 0.7909090909090909,
-            "count": 110
-          }
-        }
-      },
-      "add_S4": {
-        "full_accuracy": 0.28,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 48
-          },
-          "SC": {
-            "accuracy": 0.9807692307692307,
-            "count": 52
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 7
-          },
-          "UC": {
-            "accuracy": 0.6179775280898876,
-            "count": 89
-          },
-          "US": {
-            "accuracy": 0.577922077922078,
-            "count": 154
-          }
-        }
-      },
-      "add_S5": {
-        "full_accuracy": 0.08,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.26,
-            "count": 50
-          },
-          "US": {
-            "accuracy": 0.255,
-            "count": 200
-          }
-        }
-      },
-      "add_S6": {
-        "full_accuracy": 0.14,
-        "n_examples": 50,
-        "per_subtask": {
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.32,
-            "count": 50
-          },
-          "US": {
-            "accuracy": 0.328,
-            "count": 250
-          }
-        }
-      },
-      "add_random": {
-        "full_accuracy": 0.805,
-        "n_examples": 200,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9837587006960556,
-            "count": 431
-          },
-          "SC": {
-            "accuracy": 0.9841772151898734,
-            "count": 316
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 39
-          },
-          "UC": {
-            "accuracy": 0.9464285714285714,
-            "count": 560
-          },
-          "US": {
-            "accuracy": 0.9259259259259259,
-            "count": 54
-          }
-        }
-      },
-      "add_C3": {
-        "full_accuracy": 0.58,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 150
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8173076923076923,
-            "count": 104
-          },
-          "US": {
-            "accuracy": 0.8043478260869565,
-            "count": 46
-          }
-        }
-      },
-      "add_C4": {
-        "full_accuracy": 0.4,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 100
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.7560975609756098,
-            "count": 123
-          },
-          "US": {
-            "accuracy": 0.6363636363636364,
-            "count": 77
-          }
-        }
-      },
-      "add_C5": {
-        "full_accuracy": 0.42,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.7792207792207793,
-            "count": 154
-          },
-          "US": {
-            "accuracy": 0.8229166666666666,
-            "count": 96
-          }
-        }
-      },
-      "add_C6": {
-        "full_accuracy": 0.42,
-        "n_examples": 50,
-        "per_subtask": {
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8186813186813187,
-            "count": 182
-          },
-          "US": {
-            "accuracy": 0.8559322033898306,
-            "count": 118
-          }
-        }
-      },
-      "sub_M0": {
-        "full_accuracy": 0.84,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9727891156462585,
-            "count": 294
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 56
-          }
-        }
-      },
-      "sub_M1": {
-        "full_accuracy": 0.74,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9790209790209791,
-            "count": 143
-          },
-          "MB": {
-            "accuracy": 0.9565217391304348,
-            "count": 69
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 15
-          },
-          "UB": {
-            "accuracy": 0.9349593495934959,
-            "count": 123
-          }
-        }
-      },
-      "sub_M2": {
-        "full_accuracy": 0.48,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9907407407407407,
-            "count": 108
-          },
-          "MB": {
-            "accuracy": 0.9615384615384616,
-            "count": 52
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 52
-          },
-          "UB": {
-            "accuracy": 0.7471264367816092,
-            "count": 87
-          },
-          "UD": {
-            "accuracy": 0.9215686274509803,
-            "count": 51
-          }
-        }
-      },
-      "sub_M3": {
-        "full_accuracy": 0.16,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 94
-          },
-          "MB": {
-            "accuracy": 0.9607843137254902,
-            "count": 51
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 25
-          },
-          "UB": {
-            "accuracy": 0.5256410256410257,
-            "count": 78
-          },
-          "UD": {
-            "accuracy": 0.6470588235294118,
-            "count": 102
-          }
-        }
-      },
-      "sub_M4": {
-        "full_accuracy": 0.02,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 100
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.26,
-            "count": 50
-          },
-          "UD": {
-            "accuracy": 0.4066666666666667,
-            "count": 150
-          }
-        }
-      },
-      "sub_M5": {
-        "full_accuracy": 0.04,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.42,
-            "count": 50
-          },
-          "UD": {
-            "accuracy": 0.285,
-            "count": 200
-          }
-        }
-      },
-      "sub_random": {
-        "full_accuracy": 0.67,
-        "n_examples": 200,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9778911564625851,
-            "count": 588
-          },
-          "MB": {
-            "accuracy": 0.9701492537313433,
-            "count": 268
-          },
-          "ME": {
-            "accuracy": 0.9833333333333333,
-            "count": 60
-          },
-          "UB": {
-            "accuracy": 0.8993288590604027,
-            "count": 447
-          },
-          "UD": {
-            "accuracy": 0.8918918918918919,
-            "count": 37
-          }
-        }
-      },
-      "sub_B3": {
-        "full_accuracy": 0.46,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 150
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.7570093457943925,
-            "count": 107
-          },
-          "UD": {
-            "accuracy": 0.813953488372093,
-            "count": 43
-          }
-        }
-      },
-      "sub_B4": {
-        "full_accuracy": 0.3,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 100
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.6929824561403509,
-            "count": 114
-          },
-          "UD": {
-            "accuracy": 0.6976744186046512,
-            "count": 86
-          }
-        }
-      },
-      "sub_B5": {
-        "full_accuracy": 0.28,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.6928104575163399,
-            "count": 153
-          },
-          "UD": {
-            "accuracy": 0.5773195876288659,
-            "count": 97
-          }
-        }
-      }
-    },
-    "summary": {
-      "overall_accuracy": 0.495,
-      "total_examples": 1400,
-      "n_splits": 22
-    }
-  },
-  "sorl_eval": {
-    "config": {
-      "ops": "add_sub",
-      "K": 1,
-      "mode": "sorl",
-      "n_digits": 6,
-      "n_per_split": 50
-    },
-    "splits": {
-      "add_S0": {
-        "full_accuracy": 0.9,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9830508474576272,
-            "count": 295
-          },
-          "SS": {
-            "accuracy": 0.9636363636363636,
-            "count": 55
-          }
-        }
-      },
-      "add_S1": {
-        "full_accuracy": 0.88,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9841269841269841,
-            "count": 126
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 79
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 21
-          },
-          "UC": {
-            "accuracy": 0.9596774193548387,
-            "count": 124
-          }
-        }
-      },
-      "add_S2": {
-        "full_accuracy": 0.56,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9733333333333334,
-            "count": 75
-          },
-          "SC": {
-            "accuracy": 0.9838709677419355,
-            "count": 62
-          },
-          "SS": {
-            "accuracy": 0.9743589743589743,
-            "count": 39
-          },
-          "UC": {
-            "accuracy": 0.8288288288288288,
-            "count": 111
-          },
-          "US": {
-            "accuracy": 0.9841269841269841,
-            "count": 63
-          }
-        }
-      },
-      "add_S3": {
-        "full_accuracy": 0.48,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 60
-          },
-          "SC": {
-            "accuracy": 0.9649122807017544,
-            "count": 57
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 19
-          },
-          "UC": {
-            "accuracy": 0.7884615384615384,
-            "count": 104
-          },
-          "US": {
-            "accuracy": 0.9090909090909091,
-            "count": 110
-          }
-        }
-      },
-      "add_S4": {
-        "full_accuracy": 0.28,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 48
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 52
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 7
-          },
-          "UC": {
-            "accuracy": 0.6067415730337079,
-            "count": 89
-          },
-          "US": {
-            "accuracy": 0.7792207792207793,
-            "count": 154
-          }
-        }
-      },
-      "add_S5": {
-        "full_accuracy": 0.16,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.36,
-            "count": 50
-          },
-          "US": {
-            "accuracy": 0.48,
-            "count": 200
-          }
-        }
-      },
-      "add_S6": {
-        "full_accuracy": 0.22,
-        "n_examples": 50,
-        "per_subtask": {
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.38,
-            "count": 50
-          },
-          "US": {
-            "accuracy": 0.54,
-            "count": 250
-          }
-        }
-      },
-      "add_random": {
-        "full_accuracy": 0.85,
-        "n_examples": 200,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9837587006960556,
-            "count": 431
-          },
-          "SC": {
-            "accuracy": 0.9968354430379747,
-            "count": 316
-          },
-          "SS": {
-            "accuracy": 1.0,
-            "count": 39
-          },
-          "UC": {
-            "accuracy": 0.9571428571428572,
-            "count": 560
-          },
-          "US": {
-            "accuracy": 0.9814814814814815,
-            "count": 54
-          }
-        }
-      },
-      "add_C3": {
-        "full_accuracy": 0.66,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.9933333333333333,
-            "count": 150
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8653846153846154,
-            "count": 104
-          },
-          "US": {
-            "accuracy": 0.8260869565217391,
-            "count": 46
-          }
-        }
-      },
-      "add_C4": {
-        "full_accuracy": 0.52,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 0.99,
-            "count": 100
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8130081300813008,
-            "count": 123
-          },
-          "US": {
-            "accuracy": 0.7272727272727273,
-            "count": 77
-          }
-        }
-      },
-      "add_C5": {
-        "full_accuracy": 0.48,
-        "n_examples": 50,
-        "per_subtask": {
-          "SA": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8051948051948052,
-            "count": 154
-          },
-          "US": {
-            "accuracy": 0.8645833333333334,
-            "count": 96
-          }
-        }
-      },
-      "add_C6": {
-        "full_accuracy": 0.42,
-        "n_examples": 50,
-        "per_subtask": {
-          "SC": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UC": {
-            "accuracy": 0.8296703296703297,
-            "count": 182
-          },
-          "US": {
-            "accuracy": 0.8305084745762712,
-            "count": 118
-          }
-        }
-      },
-      "sub_M0": {
-        "full_accuracy": 0.96,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9931972789115646,
-            "count": 294
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 56
-          }
-        }
-      },
-      "sub_M1": {
-        "full_accuracy": 0.86,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.986013986013986,
-            "count": 143
-          },
-          "MB": {
-            "accuracy": 0.9855072463768116,
-            "count": 69
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 15
-          },
-          "UB": {
-            "accuracy": 0.967479674796748,
-            "count": 123
-          }
-        }
-      },
-      "sub_M2": {
-        "full_accuracy": 0.7,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9629629629629629,
-            "count": 108
-          },
-          "MB": {
-            "accuracy": 0.9807692307692307,
-            "count": 52
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 52
-          },
-          "UB": {
-            "accuracy": 0.8735632183908046,
-            "count": 87
-          },
-          "UD": {
-            "accuracy": 0.9607843137254902,
-            "count": 51
-          }
-        }
-      },
-      "sub_M3": {
-        "full_accuracy": 0.34,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 94
-          },
-          "MB": {
-            "accuracy": 0.9803921568627451,
-            "count": 51
-          },
-          "ME": {
-            "accuracy": 1.0,
-            "count": 25
-          },
-          "UB": {
-            "accuracy": 0.6282051282051282,
-            "count": 78
-          },
-          "UD": {
-            "accuracy": 0.7843137254901961,
-            "count": 102
-          }
-        }
-      },
-      "sub_M4": {
-        "full_accuracy": 0.2,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 100
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.38,
-            "count": 50
-          },
-          "UD": {
-            "accuracy": 0.6,
-            "count": 150
-          }
-        }
-      },
-      "sub_M5": {
-        "full_accuracy": 0.12,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.38,
-            "count": 50
-          },
-          "UD": {
-            "accuracy": 0.42,
-            "count": 200
-          }
-        }
-      },
-      "sub_random": {
-        "full_accuracy": 0.795,
-        "n_examples": 200,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 0.9863945578231292,
-            "count": 588
-          },
-          "MB": {
-            "accuracy": 0.9813432835820896,
-            "count": 268
-          },
-          "ME": {
-            "accuracy": 0.9833333333333333,
-            "count": 60
-          },
-          "UB": {
-            "accuracy": 0.9373601789709173,
-            "count": 447
-          },
-          "UD": {
-            "accuracy": 0.9459459459459459,
-            "count": 37
-          }
-        }
-      },
-      "sub_B3": {
-        "full_accuracy": 0.68,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 150
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.8598130841121495,
-            "count": 107
-          },
-          "UD": {
-            "accuracy": 0.8837209302325582,
-            "count": 43
-          }
-        }
-      },
-      "sub_B4": {
-        "full_accuracy": 0.56,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 100
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.8157894736842105,
-            "count": 114
-          },
-          "UD": {
-            "accuracy": 0.8488372093023255,
-            "count": 86
-          }
-        }
-      },
-      "sub_B5": {
-        "full_accuracy": 0.46,
-        "n_examples": 50,
-        "per_subtask": {
-          "MD": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "MB": {
-            "accuracy": 1.0,
-            "count": 50
-          },
-          "UB": {
-            "accuracy": 0.8366013071895425,
-            "count": 153
-          },
-          "UD": {
-            "accuracy": 0.711340206185567,
-            "count": 97
-          }
-        }
-      }
-    },
-    "summary": {
-      "overall_accuracy": 0.6078571428571429,
-      "total_examples": 1400,
-      "n_splits": 22
-    }
-  },
-  "sorl_overall_accuracy": 0.6078571428571429,
-  "sft_overall_accuracy": 0.495
-}

add_sub_sorl_v1_abs10_K1_50K_1L3H510d/model.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d83775b1d2ac97ad283814b3a9af8f2969068a1ff887cab77d2f8f9d3b7db9a1
-size 634679036

add_sub_sorl_v1_abs10_K1_50K_1L3H510d/train_config.json

@@ -1,35 +0,0 @@
-{
-  "mode": "sorl",
-  "ops": "add_sub",
-  "n_digits": 6,
-  "n_layer": 1,
-  "n_head": 3,
-  "n_embd": 510,
-  "abs_vocab": 10,
-  "K": 1,
-  "alpha_info_gain": 10.0,
-  "alpha_abs": 0.1,
-  "alpha_soft_zipf": 1.0,
-  "batch_size": 64,
-  "num_epochs": 10,
-  "dataset_size": 50000,
-  "lr": 8e-05,
-  "output_dir": "ckpt/sweep/as_sorl_abs10_K1_50K_1L3H510d",
-  "device": "cuda",
-  "push_to_hub": true,
-  "no_wandb": false,
-  "n_params": 158593426,
-  "run_name": "add_sub_sorl_v1_abs10_K1_50K_1L3H510d",
-  "git_commit": "8b149cd78544f600d46584ae2f143c0261d8eeb8",
-  "timestamp": "2026-04-12T04:55:37.572226+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": "tw2j1378",
-  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/tw2j1378",
-  "final_accuracy": 0.6078571428571429,
-  "sft_accuracy": 0.495,
-  "eval_method": "ArithmeticEvaluator"
-}