Upload add_sub_sorl_v1_abs16_100K

add_sub_sorl_v1_abs16_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": 151660
+}

add_sub_sorl_v1_abs16_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_abs16_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": [
+      16.671669006347656,
+      11.76451301574707,
+      8.801431655883789,
+      7.921006202697754,
+      5.511175155639648,
+      4.23372745513916,
+      3.3669848442077637,
+      2.7362313270568848,
+      2.7798409461975098,
+      2.0342774391174316,
+      2.6778364181518555,
+      2.3960514068603516,
+      2.3930070400238037,
+      1.4117414951324463,
+      1.4301897287368774,
+      1.7312417030334473,
+      1.1918331384658813,
+      1.3515545129776,
+      1.8523873090744019,
+      1.2459018230438232,
+      0.9775326251983643,
+      -0.6364008784294128,
+      -3.243363618850708,
+      -4.095810890197754,
+      -6.960159778594971,
+      -9.253399848937988,
+      -9.103655815124512,
+      -9.563366889953613,
+      -11.231627464294434,
+      -11.219977378845215,
+      -11.866069793701172,
+      -13.595829963684082,
+      -13.131518363952637,
+      -13.202994346618652,
+      -13.279316902160645,
+      -13.075213432312012,
+      -13.784010887145996,
+      -13.952326774597168,
+      -13.966639518737793,
+      -13.274131774902344,
+      -13.492032051086426,
+      -14.071678161621094,
+      -13.921222686767578,
+      -13.53633975982666,
+      -13.5321626663208,
+      -14.607442855834961,
+      -14.159483909606934,
+      -14.47022533416748,
+      -13.999115943908691,
+      -13.745505332946777,
+      -14.885555267333984,
+      -14.397637367248535,
+      -14.321432113647461,
+      -14.085877418518066,
+      -14.414926528930664,
+      -13.892180442810059,
+      -13.484789848327637,
+      -13.856369972229004,
+      -14.260724067687988,
+      -14.607751846313477,
+      -14.285247802734375,
+      -14.1376953125,
+      -13.70485782623291,
+      -14.543573379516602,
+      -14.476613998413086,
+      -14.107405662536621,
+      -14.12656307220459,
+      -13.730513572692871,
+      -14.510255813598633,
+      -13.808155059814453,
+      -14.117059707641602,
+      -14.026846885681152,
+      -14.534915924072266,
+      -14.46925163269043,
+      -14.468484878540039,
+      -14.118741035461426,
+      -14.42296028137207,
+      -14.069829940795898,
+      -14.312148094177246,
+      -14.25886058807373,
+      -15.426573753356934,
+      -13.97634506225586,
+      -14.062787055969238,
+      -13.932819366455078,
+      -14.374216079711914,
+      -14.277177810668945,
+      -14.284172058105469,
+      -14.175317764282227,
+      -13.622069358825684,
+      -12.66834545135498,
+      -13.368535041809082,
+      -11.731734275817871,
+      -9.118104934692383,
+      -7.7215046882629395,
+      -6.636661052703857,
+      -6.065880298614502,
+      -5.550992488861084,
+      -5.093198776245117,
+      -4.752264499664307,
+      -5.129302501678467,
+      -4.841475009918213,
+      -4.835540771484375,
+      -4.206387996673584,
+      -5.119156837463379,
+      -3.9197139739990234,
+      -4.387465476989746,
+      -4.074684143066406,
+      -3.418381690979004,
+      -4.059027194976807,
+      -3.477524757385254,
+      -3.8114800453186035,
+      -3.1857621669769287,
+      -3.7550437450408936,
+      -3.1774818897247314,
+      -3.237245798110962,
+      -2.9989490509033203,
+      -3.0659162998199463,
+      -2.8278210163116455,
+      -3.4058122634887695,
+      -2.9876441955566406,
+      -2.8914389610290527,
+      -2.676650285720825,
+      -2.672682523727417,
+      -2.7980239391326904,
+      -3.3208882808685303,
+      -2.9594175815582275,
+      -2.949089765548706,
+      -2.682825803756714,
+      -2.5156426429748535,
+      -2.4986965656280518,
+      -2.7391247749328613,
+      -2.567275047302246,
+      -2.553879737854004,
+      -2.890953779220581,
+      -2.345930576324463,
+      -1.8966050148010254,
+      -2.860386371612549,
+      -2.4158926010131836,
+      -2.1681196689605713,
+      -1.8340377807617188,
+      -2.2587294578552246,
+      -1.906646490097046,
+      -2.0606415271759033,
+      -1.9160561561584473,
+      -2.48103928565979,
+      -1.9408878087997437,
+      -1.853256106376648,
+      -2.2146284580230713,
+      -2.201028347015381,
+      -1.828320026397705,
+      -2.034412384033203,
+      -1.7938942909240723,
+      -2.244307041168213,
+      -1.9889719486236572,
+      -1.6898540258407593,
+      -1.7858039140701294,
+      -2.191715955734253,
+      -1.9715027809143066,
+      -1.919372797012329,
+      -1.8914986848831177,
+      -1.8873316049575806,
+      -1.4288508892059326,
+      -2.1516754627227783,
+      -1.8888063430786133,
+      -1.8703267574310303,
+      -2.0618436336517334,
+      -1.4358248710632324,
+      -1.3266373872756958,
+      -1.5694628953933716,
+      -1.402905821800232,
+      -1.6993207931518555,
+      -1.5431238412857056,
+      -1.7176926136016846,
+      -1.5956352949142456,
+      -1.402787208557129,
+      -1.6125233173370361,
+      -1.6770639419555664,
+      -1.930527925491333,
+      -1.4489266872406006,
+      -1.4066162109375,
+      -1.8514528274536133,
+      -1.1257268190383911,
+      -1.7762130498886108,
+      -1.7130666971206665,
+      -1.2325992584228516,
+      -1.542702317237854,
+      -1.270906686782837,
+      -1.1530277729034424,
+      -1.4442468881607056,
+      -1.7895030975341797,
+      -1.8281203508377075,
+      -1.118044137954712,
+      -1.084818959236145,
+      -1.2554227113723755,
+      -1.0617523193359375,
+      -1.375596523284912,
+      -1.3179610967636108,
+      -1.1270630359649658,
+      -1.0592299699783325,
+      -1.3966926336288452,
+      -0.9684830904006958,
+      -1.1333949565887451,
+      -1.1397461891174316,
+      -1.2867859601974487,
+      -1.0852409601211548,
+      -1.3851423263549805,
+      -0.8853378295898438,
+      -0.94475919008255,
+      -0.9066417217254639,
+      -1.519598126411438,
+      -1.102278709411621,
+      -0.8912283182144165,
+      -0.7503770589828491,
+      -0.9335433840751648,
+      -0.9593464136123657,
+      -1.116734266281128,
+      -0.8759087324142456,
+      -1.2935878038406372,
+      -0.6411604881286621,
+      -0.9160860776901245,
+      -0.93536376953125,
+      -0.8380880355834961,
+      -0.7900724411010742,
+      -0.9567074179649353,
+      -0.7380226254463196,
+      -0.9075157046318054,
+      -0.8449690937995911,
+      -0.9333892464637756,
+      -0.7380180954933167,
+      -1.0785728693008423,
+      -0.8032457828521729,
+      -0.6366966962814331,
+      -0.6988163590431213,
+      -0.9169015288352966,
+      -1.214188814163208,
+      -0.6135467886924744,
+      -0.6465123891830444,
+      -0.7552400827407837,
+      -0.7670201659202576,
+      -0.8525322675704956,
+      -0.7103525400161743,
+      -0.5674810409545898,
+      -1.270815372467041,
+      -0.7840167284011841,
+      -0.5641080141067505,
+      -0.8306724429130554,
+      -0.788975179195404,
+      -0.65106600522995,
+      -0.5759876370429993,
+      -0.555364191532135,
+      -0.7142438888549805,
+      -0.602454662322998,
+      -0.7172399163246155,
+      -0.4023277461528778,
+      -0.6256492733955383,
+      -0.7062656283378601,
+      -0.6178817749023438,
+      -0.4715927243232727,
+      -0.37143242359161377,
+      -0.5795446634292603,
+      -0.7000231146812439,
+      -0.6491426229476929,
+      -0.5702045559883118,
+      -0.6726967692375183,
+      -0.46789199113845825,
+      -0.649307131767273,
+      -0.612271249294281,
+      -0.45937222242355347,
+      -0.35666269063949585,
+      -0.5950183272361755,
+      -0.5590232610702515,
+      -0.5428248047828674,
+      -0.5403609275817871,
+      -0.3566666543483734,
+      -0.6202500462532043,
+      -0.4409750699996948,
+      -0.5675917863845825,
+      -0.4542946517467499,
+      -0.39824873208999634,
+      -0.3195297420024872,
+      -0.4000018239021301,
+      -0.45862290263175964,
+      -0.5217932462692261,
+      -0.39715659618377686,
+      -0.33934420347213745,
+      -0.43397337198257446,
+      -0.4806961417198181,
+      -0.3589365780353546,
+      -0.4422625005245209,
+      -0.6090811491012573,
+      -0.4298260509967804,
+      -0.3886895775794983,
+      -0.29521632194519043,
+      -0.335098534822464,
+      -0.2614021599292755,
+      -0.7061392664909363,
+      -0.39528343081474304,
+      -0.25662821531295776,
+      -0.2705487608909607,
+      -0.34987732768058777,
+      -0.26364603638648987,
+      -0.28086966276168823,
+      -0.3212970197200775,
+      -0.19448548555374146,
+      -0.28908517956733704,
+      -0.3309410810470581,
+      -0.311942994594574,
+      -0.3063529431819916,
+      -0.34509095549583435,
+      -0.39035525918006897
+    ],
+    "base_loss": [
+      10.580854415893555,
+      8.119989395141602,
+      7.3575310707092285,
+      6.388346195220947,
+      5.366873264312744,
+      3.7506906986236572,
+      2.317335367202759,
+      2.0707907676696777,
+      1.9128801822662354,
+      1.8451582193374634,
+      1.8457585573196411,
+      1.790676474571228,
+      1.8424475193023682,
+      1.8503745794296265,
+      1.8616712093353271,
+      1.8210214376449585,
+      1.791472315788269,
+      1.8120089769363403,
+      1.7665530443191528,
+      1.808519959449768,
+      1.7760982513427734,
+      1.8747520446777344,
+      1.9280900955200195,
+      1.828168511390686,
+      1.8661088943481445,
+      1.8001527786254883,
+      1.774647831916809,
+      1.6722437143325806,
+      1.6824556589126587,
+      1.7306270599365234,
+      1.6374658346176147,
+      1.7564665079116821,
+      1.6978530883789062,
+      1.6683281660079956,
+      1.670649766921997,
+      1.6164610385894775,
+      1.6997318267822266,
+      1.690144658088684,
+      1.6809651851654053,
+      1.6078029870986938,
+      1.590226173400879,
+      1.6705245971679688,
+      1.6055727005004883,
+      1.5866053104400635,
+      1.5808082818984985,
+      1.6946316957473755,
+      1.6277645826339722,
+      1.663727045059204,
+      1.633544683456421,
+      1.5950450897216797,
+      1.7065118551254272,
+      1.6284934282302856,
+      1.6226651668548584,
+      1.6058158874511719,
+      1.6366831064224243,
+      1.5840657949447632,
+      1.5265496969223022,
+      1.5768431425094604,
+      1.6252962350845337,
+      1.6619303226470947,
+      1.6097668409347534,
+      1.596845030784607,
+      1.5575586557388306,
+      1.6523736715316772,
+      1.651186227798462,
+      1.5962696075439453,
+      1.5841522216796875,
+      1.5555152893066406,
+      1.634290099143982,
+      1.561633825302124,
+      1.5917693376541138,
+      1.5894688367843628,
+      1.6380078792572021,
+      1.6202070713043213,
+      1.6248090267181396,
+      1.579480767250061,
+      1.6126420497894287,
+      1.5998783111572266,
+      1.6076043844223022,
+      1.5964406728744507,
+      1.7230976819992065,
+      1.5618696212768555,
+      1.5692088603973389,
+      1.5604798793792725,
+      1.6072237491607666,
+      1.5947141647338867,
+      1.5954008102416992,
+      1.5883949995040894,
+      1.5214720964431763,
+      1.4194573163986206,
+      1.4938545227050781,
+      1.3120161294937134,
+      1.0255168676376343,
+      0.8711405396461487,
+      0.74779212474823,
+      0.682752788066864,
+      0.6244893670082092,
+      0.5808435082435608,
+      0.5371533632278442,
+      0.5762635469436646,
+      0.5437158346176147,
+      0.5455597043037415,
+      0.47316813468933105,
+      0.5756062269210815,
+      0.44454050064086914,
+      0.49653053283691406,
+      0.45801523327827454,
+      0.38474538922309875,
+      0.4559524953365326,
+      0.39584872126579285,
+      0.42833438515663147,
+      0.3595929741859436,
+      0.4211898446083069,
+      0.3573281466960907,
+      0.3656110167503357,
+      0.3394474685192108,
+      0.34687915444374084,
+      0.31798049807548523,
+      0.38100773096084595,
+      0.33725786209106445,
+      0.32460835576057434,
+      0.30156829953193665,
+      0.3008739650249481,
+      0.31441500782966614,
+      0.3745601773262024,
+      0.33160558342933655,
+      0.33050528168678284,
+      0.3007682263851166,
+      0.283095121383667,
+      0.28178638219833374,
+      0.3084280788898468,
+      0.28848329186439514,
+      0.2892639935016632,
+      0.32449260354042053,
+      0.26323333382606506,
+      0.22359739243984222,
+      0.3220460116863251,
+      0.2926539480686188,
+      0.24356520175933838,
+      0.20664116740226746,
+      0.2545173466205597,
+      0.2144070863723755,
+      0.2385072410106659,
+      0.21697112917900085,
+      0.27931639552116394,
+      0.2182837575674057,
+      0.21005725860595703,
+      0.24993708729743958,
+      0.24655453860759735,
+      0.20667508244514465,
+      0.23139040172100067,
+      0.20187489688396454,
+      0.25210073590278625,
+      0.2255709171295166,
+      0.18960872292518616,
+      0.2023891657590866,
+      0.24529790878295898,
+      0.22237446904182434,
+      0.2161502093076706,
+      0.21211178600788116,
+      0.21355855464935303,
+      0.16337816417217255,
+      0.24225734174251556,
+      0.211660698056221,
+      0.20943006873130798,
+      0.233957439661026,
+      0.16396750509738922,
+      0.14962902665138245,
+      0.1758873611688614,
+      0.1604095846414566,
+      0.19222556054592133,
+      0.17423351109027863,
+      0.1934070587158203,
+      0.18693335354328156,
+      0.15890741348266602,
+      0.1813715547323227,
+      0.18883773684501648,
+      0.21779997646808624,
+      0.16362996399402618,
+      0.15952694416046143,
+      0.21145786345005035,
+      0.1282462626695633,
+      0.19901803135871887,
+      0.19262051582336426,
+      0.13982339203357697,
+      0.17418090999126434,
+      0.14544856548309326,
+      0.1315660923719406,
+      0.16684947907924652,
+      0.2021123468875885,
+      0.20483198761940002,
+      0.13199248909950256,
+      0.12378420680761337,
+      0.14227592945098877,
+      0.12235838919878006,
+      0.15495094656944275,
+      0.14758138358592987,
+      0.13045068085193634,
+      0.12066575139760971,
+      0.1572522073984146,
+      0.10982964187860489,
+      0.1298765391111374,
+      0.1281500905752182,
+      0.14656832814216614,
+      0.12244265526533127,
+      0.15763868391513824,
+      0.10095322877168655,
+      0.10781024396419525,
+      0.10424082726240158,
+      0.17119719088077545,
+      0.12505456805229187,
+      0.10057372599840164,
+      0.08568765968084335,
+      0.1093149334192276,
+      0.10943430662155151,
+      0.12677355110645294,
+      0.10003077983856201,
+      0.14645418524742126,
+      0.07274025678634644,
+      0.10459715873003006,
+      0.10570158064365387,
+      0.0950184240937233,
+      0.08981633931398392,
+      0.10920991003513336,
+      0.0831788033246994,
+      0.1030617505311966,
+      0.09639338403940201,
+      0.10584582388401031,
+      0.0847930759191513,
+      0.12341125309467316,
+      0.09179931879043579,
+      0.0737370103597641,
+      0.07879512757062912,
+      0.10479319095611572,
+      0.13663651049137115,
+      0.07123450189828873,
+      0.07449010759592056,
+      0.08675049990415573,
+      0.08661101013422012,
+      0.0966150164604187,
+      0.08276067674160004,
+      0.06827885657548904,
+      0.1431131213903427,
+      0.08881387859582901,
+      0.06503797322511673,
+      0.09457540512084961,
+      0.08973288536071777,
+      0.07396655529737473,
+      0.06614848226308823,
+      0.06386690586805344,
+      0.08118759840726852,
+      0.0680375024676323,
+      0.08105553686618805,
+      0.045911479741334915,
+      0.0726836770772934,
+      0.08092149347066879,
+      0.0712655708193779,
+      0.054410941898822784,
+      0.043259959667921066,
+      0.06585840880870819,
+      0.07975509017705917,
+      0.07529892027378082,
+      0.06518300622701645,
+      0.07613348215818405,
+      0.05287592485547066,
+      0.0744367465376854,
+      0.07044298201799393,
+      0.0531424805521965,
+      0.04141668602824211,
+      0.06721319258213043,
+      0.06409581750631332,
+      0.06317402422428131,
+      0.06276766210794449,
+      0.041923750191926956,
+      0.07003537565469742,
+      0.05061003938317299,
+      0.06434254348278046,
+      0.05242345854640007,
+      0.045546114444732666,
+      0.03749945014715195,
+      0.04622683301568031,
+      0.05430757999420166,
+      0.06010174751281738,
+      0.0459761805832386,
+      0.03930548578500748,
+      0.0503813736140728,
+      0.05708807706832886,
+      0.041468601673841476,
+      0.05045849829912186,
+      0.07037588953971863,
+      0.04919378086924553,
+      0.0451410673558712,
+      0.03385286405682564,
+      0.03891809657216072,
+      0.032340068370103836,
+      0.079730324447155,
+      0.045579779893159866,
+      0.031919825822114944,
+      0.03207848593592644,
+      0.04037968069314957,
+      0.03096827305853367,
+      0.032661374658346176,
+      0.037077147513628006,
+      0.023557504639029503,
+      0.035045452415943146,
+      0.038039352744817734,
+      0.03735768795013428,
+      0.03715559467673302,
+      0.04278060048818588,
+      0.04488012567162514
+    ],
+    "info_loss": [
+      -0.17996692657470703,
+      -0.3190937042236328,
+      -0.3570389747619629,
+      -0.19894742965698242,
+      -0.23011016845703125,
+      -0.1346883773803711,
+      -0.0539860725402832,
+      -0.0833503007888794,
+      -0.058913588523864746,
+      -0.12580227851867676,
+      -0.060360074043273926,
+      -0.08251023292541504,
+      -0.08766293525695801,
+      -0.18720686435699463,
+      -0.18549346923828125,
+      -0.151009202003479,
+      -0.20220160484313965,
+      -0.18667054176330566,
+      -0.12886452674865723,
+      -0.18700921535491943,
+      -0.18197119235992432,
+      -0.30183613300323486,
+      -0.5505107641220093,
+      -0.6193157434463501,
+      -0.9058982133865356,
+      -1.1273815631866455,
+      -1.1080050468444824,
+      -1.1432299613952637,
+      -1.3089624643325806,
+      -1.310119390487671,
+      -1.3641177415847778,
+      -1.548097848892212,
+      -1.4956003427505493,
+      -1.4994077682495117,
+      -1.505875587463379,
+      -1.4798970222473145,
+      -1.558509349822998,
+      -1.5736181735992432,
+      -1.5748445987701416,
+      -1.4975718259811401,
+      -1.517536997795105,
+      -1.5836838483810425,
+      -1.5615086555480957,
+      -1.5210106372833252,
+      -1.5199164152145386,
+      -1.6390318870544434,
+      -1.5876792669296265,
+      -1.6216202974319458,
+      -1.5719168186187744,
+      -1.542820930480957,
+      -1.66734778881073,
+      -1.6107299327850342,
+      -1.6022093296051025,
+      -1.5772621631622314,
+      -1.6136329174041748,
+      -1.5553803443908691,
+      -1.5086798667907715,
+      -1.551520824432373,
+      -1.5959516763687134,
+      -1.6351819038391113,
+      -1.5981864929199219,
+      -1.5813840627670288,
+      -1.5355732440948486,
+      -1.627362847328186,
+      -1.6219881772994995,
+      -1.5776844024658203,
+      -1.577966332435608,
+      -1.5382840633392334,
+      -1.6226786375045776,
+      -1.544333577156067,
+      -1.579701542854309,
+      -1.57033371925354,
+      -1.624372124671936,
+      -1.6153326034545898,
+      -1.6191116571426392,
+      -1.5754673480987549,
+      -1.6087896823883057,
+      -1.572089433670044,
+      -1.5979163646697998,
+      -1.5896975994110107,
+      -1.7210109233856201,
+      -1.559007167816162,
+      -1.5680458545684814,
+      -1.5563300848007202,
+      -1.6043293476104736,
+      -1.5922127962112427,
+      -1.5927090644836426,
+      -1.581242322921753,
+      -1.5191965103149414,
+      -1.4125746488571167,
+      -1.4907222986221313,
+      -1.3086503744125366,
+      -1.0196993350982666,
+      -0.8642669916152954,
+      -0.7444685697555542,
+      -0.6803299188613892,
+      -0.6222341060638428,
+      -0.5722401738166809,
+      -0.5327720046043396,
+      -0.5747119188308716,
+      -0.5429865717887878,
+      -0.5421769022941589,
+      -0.4714232087135315,
+      -0.5735390186309814,
+      -0.4399084150791168,
+      -0.49221763014793396,
+      -0.4561785161495209,
+      -0.3835972845554352,
+      -0.4552410840988159,
+      -0.3910060524940491,
+      -0.42728784680366516,
+      -0.3579067289829254,
+      -0.42032092809677124,
+      -0.3570646345615387,
+      -0.36265188455581665,
+      -0.3370177447795868,
+      -0.3462562561035156,
+      -0.31741368770599365,
+      -0.3804982900619507,
+      -0.33681613206863403,
+      -0.32431018352508545,
+      -0.30130457878112793,
+      -0.30021288990974426,
+      -0.3136861324310303,
+      -0.3720722198486328,
+      -0.3313670754432678,
+      -0.33003127574920654,
+      -0.3006082773208618,
+      -0.2827286720275879,
+      -0.28115639090538025,
+      -0.30827274918556213,
+      -0.2881247401237488,
+      -0.28746354579925537,
+      -0.32409948110580444,
+      -0.2629616856575012,
+      -0.21455107629299164,
+      -0.32120198011398315,
+      -0.2749188244342804,
+      -0.24328647553920746,
+      -0.20617544651031494,
+      -0.25418204069137573,
+      -0.21421004831790924,
+      -0.23501364886760712,
+      -0.21569274365901947,
+      -0.27855417132377625,
+      -0.217383474111557,
+      -0.20980972051620483,
+      -0.24906191229820251,
+      -0.24639293551445007,
+      -0.20644114911556244,
+      -0.23124420642852783,
+      -0.20180894434452057,
+      -0.2518523037433624,
+      -0.22521087527275085,
+      -0.1894122213125229,
+      -0.20228645205497742,
+      -0.2452012002468109,
+      -0.22208666801452637,
+      -0.2159036099910736,
+      -0.2118225246667862,
+      -0.2133956402540207,
+      -0.16279751062393188,
+      -0.24200676381587982,
+      -0.2112807035446167,
+      -0.20928657054901123,
+      -0.23227792978286743,
+      -0.16352015733718872,
+      -0.14952941238880157,
+      -0.17578725516796112,
+      -0.16035915911197662,
+      -0.19215452671051025,
+      -0.1737513244152069,
+      -0.1933174580335617,
+      -0.18029972910881042,
+      -0.15852904319763184,
+      -0.18122391402721405,
+      -0.18878407776355743,
+      -0.21673789620399475,
+      -0.16333439946174622,
+      -0.15893402695655823,
+      -0.2106197327375412,
+      -0.12788090109825134,
+      -0.19891726970672607,
+      -0.19253608584403992,
+      -0.13897256553173065,
+      -0.17408844828605652,
+      -0.14527390897274017,
+      -0.13148440420627594,
+      -0.16394594311714172,
+      -0.20199790596961975,
+      -0.20452159643173218,
+      -0.12705852091312408,
+      -0.1237156018614769,
+      -0.14167487621307373,
+      -0.12115234136581421,
+      -0.15484929084777832,
+      -0.1473502814769745,
+      -0.13023601472377777,
+      -0.12042754888534546,
+      -0.15715979039669037,
+      -0.10967076569795609,
+      -0.12870651483535767,
+      -0.12807932496070862,
+      -0.14650602638721466,
+      -0.12239541858434677,
+      -0.15736448764801025,
+      -0.10068048536777496,
+      -0.1070597916841507,
+      -0.10414526611566544,
+      -0.17113865911960602,
+      -0.12484883517026901,
+      -0.10050682723522186,
+      -0.08561678975820541,
+      -0.1087525337934494,
+      -0.10938119888305664,
+      -0.12639160454273224,
+      -0.09981365501880646,
+      -0.14637455344200134,
+      -0.07250282168388367,
+      -0.104531429708004,
+      -0.10568559914827347,
+      -0.09498116374015808,
+      -0.08961202949285507,
+      -0.10917086899280548,
+      -0.08308960497379303,
+      -0.10303566604852676,
+      -0.09634371846914291,
+      -0.10571286827325821,
+      -0.08474351465702057,
+      -0.12209111452102661,
+      -0.09178029745817184,
+      -0.07351583987474442,
+      -0.0787777528166771,
+      -0.10475990176200867,
+      -0.13661758601665497,
+      -0.07112528383731842,
+      -0.07447189092636108,
+      -0.08673693239688873,
+      -0.08659523725509644,
+      -0.09658211469650269,
+      -0.08267847448587418,
+      -0.06712014973163605,
+      -0.14302897453308105,
+      -0.08876700699329376,
+      -0.06501232832670212,
+      -0.09453508257865906,
+      -0.08969758450984955,
+      -0.07390053570270538,
+      -0.0661381259560585,
+      -0.06384944915771484,
+      -0.08109883219003677,
+      -0.0680072084069252,
+      -0.08103962242603302,
+      -0.04586905613541603,
+      -0.07267068326473236,
+      -0.08090422302484512,
+      -0.07110242545604706,
+      -0.05409397929906845,
+      -0.04269455373287201,
+      -0.06583482027053833,
+      -0.07937608659267426,
+      -0.07524216175079346,
+      -0.06517204642295837,
+      -0.07612097263336182,
+      -0.052849315106868744,
+      -0.07370105385780334,
+      -0.07042257487773895,
+      -0.05312580615282059,
+      -0.04139900207519531,
+      -0.06719101965427399,
+      -0.06405308842658997,
+      -0.06315374374389648,
+      -0.06275524944067001,
+      -0.041907232254743576,
+      -0.07002557069063187,
+      -0.05053453519940376,
+      -0.06433310359716415,
+      -0.052391860634088516,
+      -0.04545298591256142,
+      -0.037478815764188766,
+      -0.046212367713451385,
+      -0.05429388955235481,
+      -0.060085806995630264,
+      -0.04596419259905815,
+      -0.03929342329502106,
+      -0.05037174001336098,
+      -0.057066384702920914,
+      -0.04145510122179985,
+      -0.050410572439432144,
+      -0.07036355882883072,
+      -0.0491868332028389,
+      -0.04512714967131615,
+      -0.0338466539978981,
+      -0.038909655064344406,
+      -0.03232991695404053,
+      -0.0797240287065506,
+      -0.04530204460024834,
+      -0.03190861642360687,
+      -0.03206954523921013,
+      -0.0403728149831295,
+      -0.030958812683820724,
+      -0.03265200927853584,
+      -0.03706395998597145,
+      -0.02355104126036167,
+      -0.035033151507377625,
+      -0.03803236037492752,
+      -0.03734046220779419,
+      -0.03714447841048241,
+      -0.04277036339044571,
+      -0.044874515384435654
+    ],
+    "abs_loss": [
+      2.777390480041504,
+      2.7542929649353027,
+      2.6910240650177,
+      2.6029257774353027,
+      2.49186372756958,
+      2.375990867614746,
+      2.3502631187438965,
+      2.308225631713867,
+      2.187129497528076,
+      2.2456979751586914,
+      2.1835029125213623,
+      2.165350914001465,
+      2.0977091789245605,
+      2.1832118034362793,
+      2.0973477363586426,
+      2.115016222000122,
+      2.146544933319092,
+      2.0843350887298584,
+      2.0572328567504883,
+      1.9873418807983398,
+      1.582409381866455,
+      1.1125017404556274,
+      0.9802367091178894,
+      0.9867104887962341,
+      0.8765336275100708,
+      0.9114257097244263,
+      0.9439901113510132,
+      0.9020074605941772,
+      0.849125862121582,
+      0.732757031917572,
+      0.6503767371177673,
+      0.6647205352783203,
+      0.626185417175293,
+      0.6453863382339478,
+      0.5542699098587036,
+      0.6062915325164795,
+      0.5217954516410828,
+      0.542735755443573,
+      0.6098823547363281,
+      0.4961710572242737,
+      0.516142725944519,
+      0.5891120433807373,
+      0.5462960004806519,
+      0.5471148490905762,
+      0.5368908643722534,
+      0.5415256023406982,
+      0.5650905966758728,
+      0.4759629964828491,
+      0.48722296953201294,
+      0.5060516595840454,
+      0.5123686790466309,
+      0.5332933664321899,
+      0.4913908541202545,
+      0.5090692043304443,
+      0.5381864309310913,
+      0.49255263805389404,
+      0.45616644620895386,
+      0.4980049431324005,
+      0.46533700823783875,
+      0.5103980302810669,
+      0.5361555218696594,
+      0.523546576499939,
+      0.5736702084541321,
+      0.5269625186920166,
+      0.47481662034988403,
+      0.5128247141838074,
+      0.45774513483047485,
+      0.47311192750930786,
+      0.5796756148338318,
+      0.4752861261367798,
+      0.4487062692642212,
+      0.3354308307170868,
+      0.31262338161468506,
+      0.3893863558769226,
+      0.24947258830070496,
+      0.32958728075027466,
+      0.2973918914794922,
+      0.2644157111644745,
+      0.32219040393829346,
+      0.27280551195144653,
+      0.22958531975746155,
+      0.2734881043434143,
+      0.282461553812027,
+      0.25209349393844604,
+      0.2485329508781433,
+      0.21366889774799347,
+      0.2146233320236206,
+      0.19329425692558289,
+      0.1748945415019989,
+      0.22702203691005707,
+      0.19403311610221863,
+      0.20346319675445557,
+      0.21548908948898315,
+      0.24604284763336182,
+      0.28499865531921387,
+      0.24369443953037262,
+      0.19881168007850647,
+      0.21112880110740662,
+      0.20992335677146912,
+      0.2196955531835556,
+      0.21074308454990387,
+      0.24281440675258636,
+      0.2148689329624176,
+      0.15726089477539062,
+      0.17713592946529388,
+      0.18005093932151794,
+      0.15527662634849548,
+      0.1784200370311737,
+      0.18204009532928467,
+      0.13945381343364716,
+      0.16688433289527893,
+      0.12064576148986816,
+      0.11875905096530914,
+      0.11682071536779404,
+      0.10268653929233551,
+      0.12855763733386993,
+      0.09095726907253265,
+      0.07905097305774689,
+      0.07472017407417297,
+      0.07584404945373535,
+      0.07616132497787476,
+      0.06104925274848938,
+      0.06016627326607704,
+      0.09960777312517166,
+      0.053712934255599976,
+      0.0893075242638588,
+      0.07951012253761292,
+      0.08070825040340424,
+      0.078128382563591,
+      0.06435772776603699,
+      0.06134745851159096,
+      0.09688157588243484,
+      0.049335725605487823,
+      0.06774897128343582,
+      0.0524895079433918,
+      0.07479576766490936,
+      0.066282257437706,
+      0.08091644942760468,
+      0.057373810559511185,
+      0.05920691043138504,
+      0.03549985587596893,
+      0.034693390130996704,
+      0.1043175756931305,
+      0.046148549765348434,
+      0.0611160509288311,
+      0.06718748807907104,
+      0.06168418005108833,
+      0.046257708221673965,
+      0.042010292410850525,
+      0.03786591440439224,
+      0.04676896333694458,
+      0.04182325303554535,
+      0.054382406175136566,
+      0.06026673689484596,
+      0.02561812289059162,
+      0.03377419710159302,
+      0.03337690979242325,
+      0.02650095522403717,
+      0.027271989732980728,
+      0.028183620423078537,
+      0.031203310936689377,
+      0.06414103507995605,
+      0.03632226958870888,
+      0.042460061609745026,
+      0.048993006348609924,
+      0.038919635117053986,
+      0.028393317013978958,
+      0.01862436719238758,
+      0.03290107101202011,
+      0.025886476039886475,
+      0.031114492565393448,
+      0.04478691518306732,
+      0.024170350283384323,
+      0.02472083270549774,
+      0.04190833494067192,
+      0.02411479689180851,
+      0.01534140482544899,
+      0.024496104568243027,
+      0.03568828105926514,
+      0.02154652588069439,
+      0.011340830475091934,
+      0.017682110890746117,
+      0.014485195279121399,
+      0.01631789095699787,
+      0.013543538749217987,
+      0.013808045536279678,
+      0.02450021170079708,
+      0.03066507913172245,
+      0.020564699545502663,
+      0.015590582974255085,
+      0.017335735261440277,
+      0.023811744526028633,
+      0.0059436894953250885,
+      0.011603141203522682,
+      0.02749473974108696,
+      0.014896477572619915,
+      0.009585765190422535,
+      0.02212778851389885,
+      0.014556828886270523,
+      0.0058863405138254166,
+      0.016437437385320663,
+      0.011743432842195034,
+      0.019322680309414864,
+      0.0323767364025116,
+      0.013688751496374607,
+      0.012714856304228306,
+      0.008867435157299042,
+      0.01705610193312168,
+      0.010245328769087791,
+      0.019361324608325958,
+      0.009954091161489487,
+      0.003815968055278063,
+      0.005069333128631115,
+      0.021628838032484055,
+      0.013429581187665462,
+      0.04039314389228821,
+      0.008578795939683914,
+      0.009801072999835014,
+      0.009786183014512062,
+      0.009284639731049538,
+      0.005527643486857414,
+      0.007528270129114389,
+      0.005455834325402975,
+      0.013145661912858486,
+      0.015458343550562859,
+      0.005277408752590418,
+      0.011443623341619968,
+      0.011221352964639664,
+      0.011266663670539856,
+      0.00862051360309124,
+      0.007774342782795429,
+      0.006529485806822777,
+      0.008653920143842697,
+      0.015745297074317932,
+      0.021161241456866264,
+      0.008661377243697643,
+      0.017117857933044434,
+      0.006823975592851639,
+      0.008625506423413754,
+      0.011471963487565517,
+      0.010647810995578766,
+      0.006814572028815746,
+      0.016460467129945755,
+      0.006403671577572823,
+      0.0105741573497653,
+      0.006004702299833298,
+      0.015992794185876846,
+      0.008600767701864243,
+      0.0034296754747629166,
+      0.0049955351278185844,
+      0.004791652783751488,
+      0.003410188015550375,
+      0.007148274220526218,
+      0.007265172898769379,
+      0.0025874958373606205,
+      0.0029066037386655807,
+      0.002836958970874548,
+      0.0024897046387195587,
+      0.003422972746193409,
+      0.003125639632344246,
+      0.006093741860240698,
+      0.005198986269533634,
+      0.010119442827999592,
+      0.005335838533937931,
+      0.004960666410624981,
+      0.004177124239504337,
+      0.010487631894648075,
+      0.008371005766093731,
+      0.0095920879393816,
+      0.01057339645922184,
+      0.002690997440367937,
+      0.004933019634336233,
+      0.015536258928477764,
+      0.007761161774396896,
+      0.012462147511541843,
+      0.00429706322029233,
+      0.011079523712396622,
+      0.006766865029931068,
+      0.0027411379851400852,
+      0.005167646333575249,
+      0.007353625725954771,
+      0.004310181830078363,
+      0.012359142303466797,
+      0.006466764025390148,
+      0.004514474421739578,
+      0.006774740293622017,
+      0.00518967118114233,
+      0.015222007408738136,
+      0.0020467271097004414,
+      0.0024294336326420307,
+      0.0062983036041259766,
+      0.0029271598905324936,
+      0.0017828027484938502,
+      0.0019189594313502312,
+      0.002218303270637989,
+      0.007079057861119509,
+      0.002842569025233388,
+      0.0016404844354838133,
+      0.0045338887721300125,
+      0.00198090891353786,
+      0.00721738999709487,
+      0.01009969413280487,
+      0.002119706943631172,
+      0.0031860843300819397,
+      0.0018541456665843725,
+      0.003260502126067877,
+      0.004363999702036381,
+      0.0026570260524749756,
+      0.002687884261831641,
+      0.002293722704052925
+    ],
+    "zipf_loss": [
+      7.6127448081970215,
+      6.560031890869141,
+      4.745188236236572,
+      3.2618415355682373,
+      2.1962172985076904,
+      1.5923216342926025,
+      1.354483962059021,
+      1.2681208848953247,
+      1.2373836040496826,
+      1.2225723266601562,
+      1.2173283100128174,
+      1.213942289352417,
+      1.2174179553985596,
+      1.2151143550872803,
+      1.2137184143066406,
+      1.2088106870651245,
+      1.2077224254608154,
+      1.1978174448013306,
+      1.1687562465667725,
+      1.1087398529052734,
+      0.8629053831100464,
+      0.3959582448005676,
+      0.23563054203987122,
+      0.17050686478614807,
+      0.1450600028038025,
+      0.12911982834339142,
+      0.10734735429286957,
+      0.10648790746927261,
+      0.09062816202640533,
+      0.07731463760137558,
+      0.07260370254516602,
+      0.062209468334913254,
+      0.06401385366916656,
+      0.05821589380502701,
+      0.05336278676986694,
+      0.0466669499874115,
+      0.04917123541235924,
+      0.039436183869838715,
+      0.03985200077295303,
+      0.04416624829173088,
+      0.04149801284074783,
+      0.0357244610786438,
+      0.03366192802786827,
+      0.032449569553136826,
+      0.0325043722987175,
+      0.03409337252378464,
+      0.033035531640052795,
+      0.03465601056814194,
+      0.037785038352012634,
+      0.03705432638525963,
+      0.030172869563102722,
+      0.02783973515033722,
+      0.028856545686721802,
+      0.030020926147699356,
+      0.030901029706001282,
+      0.028302010148763657,
+      0.02984202839434147,
+      0.032194413244724274,
+      0.026962529867887497,
+      0.031096048653125763,
+      0.03323424607515335,
+      0.02694561704993248,
+      0.03594813868403435,
+      0.024985717609524727,
+      0.04459930956363678,
+      0.02188541740179062,
+      0.023173801600933075,
+      0.04950184375047684,
+      0.024272765964269638,
+      0.026018138974905014,
+      0.04331601411104202,
+      0.05347788333892822,
+      0.03953485190868378,
+      0.024928828701376915,
+      0.07287538051605225,
+      0.023491470143198967,
+      0.02255459874868393,
+      0.024743758141994476,
+      0.027193520218133926,
+      0.014394879341125488,
+      0.03747980669140816,
+      0.02450854703783989,
+      0.020215902477502823,
+      0.044792540371418,
+      0.036999303847551346,
+      0.028868593275547028,
+      0.02605552226305008,
+      0.02938155084848404,
+      0.030934154987335205,
+      0.015242088586091995,
+      0.025429606437683105,
+      0.02240702509880066,
+      0.03182249516248703,
+      0.02542036399245262,
+      0.03173244744539261,
+      0.030296631157398224,
+      0.02697805128991604,
+      0.027246324345469475,
+      0.017310013994574547,
+      0.01958380453288555,
+      0.023600636050105095,
+      0.0163872167468071,
+      0.013189305551350117,
+      0.02490093745291233,
+      0.017116166651248932,
+      0.02017567679286003,
+      0.01355794444680214,
+      0.01500363927334547,
+      0.01922685280442238,
+      0.022741524502635002,
+      0.01637580804526806,
+      0.021647784858942032,
+      0.015100102871656418,
+      0.02415425516664982,
+      0.013393277302384377,
+      0.018925176933407784,
+      0.040671296417713165,
+      0.020430386066436768,
+      0.010690897703170776,
+      0.035674817860126495,
+      0.01943860575556755,
+      0.028722252696752548,
+      0.022555474191904068,
+      0.014461560174822807,
+      0.019902441650629044,
+      0.013716666959226131,
+      0.012766593135893345,
+      0.01441806647926569,
+      0.020736124366521835,
+      0.024645142257213593,
+      0.02903977409005165,
+      0.01580086722970009,
+      0.02655826136469841,
+      0.018773622810840607,
+      0.015203597024083138,
+      0.017828665673732758,
+      0.022959239780902863,
+      0.032549865543842316,
+      0.015442498959600925,
+      0.015154856257140636,
+      0.025023728609085083,
+      0.017577601596713066,
+      0.04055590182542801,
+      0.019285151734948158,
+      0.019074421375989914,
+      0.007944503799080849,
+      0.0286155566573143,
+      0.02142779529094696,
+      0.012145567685365677,
+      0.025629814714193344,
+      0.04196229577064514,
+      0.018138017505407333,
+      0.016676833853125572,
+      0.031539205461740494,
+      0.012097734957933426,
+      0.03129406273365021,
+      0.011660464107990265,
+      0.02433932013809681,
+      0.020786013454198837,
+      0.011796524748206139,
+      0.029945924878120422,
+      0.029332082718610764,
+      0.02250237949192524,
+      0.008093947544693947,
+      0.008209604769945145,
+      0.02308596670627594,
+      0.03256992995738983,
+      0.017165306955575943,
+      0.009232302196323872,
+      0.03768755868077278,
+      0.026887444779276848,
+      0.015677180141210556,
+      0.01965784654021263,
+      0.01795666478574276,
+      0.019405048340559006,
+      0.015932783484458923,
+      0.02040497213602066,
+      0.0166013166308403,
+      0.01721831038594246,
+      0.021042343229055405,
+      0.04215270280838013,
+      0.023067686706781387,
+      0.012493138201534748,
+      0.01804201304912567,
+      0.015948686748743057,
+      0.02262052521109581,
+      0.03393382579088211,
+      0.0271836519241333,
+      0.026306526735424995,
+      0.026804517954587936,
+      0.010530227795243263,
+      0.01816735416650772,
+      0.027958549559116364,
+      0.01788981258869171,
+      0.024663256481289864,
+      0.01645573601126671,
+      0.007001818157732487,
+      0.04263361543416977,
+      0.022924071177840233,
+      0.017064398154616356,
+      0.016751211136579514,
+      0.0226193405687809,
+      0.010964600369334221,
+      0.028468243777751923,
+      0.014901716262102127,
+      0.029592378064990044,
+      0.019627047702670097,
+      0.016322851181030273,
+      0.02954556792974472,
+      0.018655192106962204,
+      0.020159626379609108,
+      0.012884648516774178,
+      0.01959625631570816,
+      0.042504213750362396,
+      0.02368829771876335,
+      0.01636890508234501,
+      0.02133912965655327,
+      0.022723492234945297,
+      0.010148895904421806,
+      0.02370261587202549,
+      0.015237944200634956,
+      0.015952318906784058,
+      0.015685973688960075,
+      0.024476774036884308,
+      0.008148754015564919,
+      0.019251490011811256,
+      0.02093036286532879,
+      0.0167714674025774,
+      0.023497289046645164,
+      0.018064919859170914,
+      0.021980443969368935,
+      0.024071721360087395,
+      0.009300675243139267,
+      0.02432986907660961,
+      0.01323446910828352,
+      0.025605402886867523,
+      0.022004632279276848,
+      0.024696262553334236,
+      0.011458626948297024,
+      0.015526635572314262,
+      0.032606691122055054,
+      0.03476012125611305,
+      0.014715206809341908,
+      0.014199046418070793,
+      0.019919952377676964,
+      0.01950252801179886,
+      0.01666855439543724,
+      0.013112653978168964,
+      0.018902208656072617,
+      0.01876382902264595,
+      0.015077623538672924,
+      0.009238878265023232,
+      0.01138590183109045,
+      0.009724806994199753,
+      0.028115125373005867,
+      0.02156449854373932,
+      0.021593233570456505,
+      0.014687166549265385,
+      0.011910850182175636,
+      0.012632545083761215,
+      0.013373291119933128,
+      0.027460210025310516,
+      0.015320975333452225,
+      0.011845862492918968,
+      0.00722916703671217,
+      0.012848971411585808,
+      0.02046278491616249,
+      0.01790623366832733,
+      0.014951449818909168,
+      0.008621307089924812,
+      0.01714271306991577,
+      0.025045320391654968,
+      0.022870264947414398,
+      0.019705796614289284,
+      0.008724011480808258,
+      0.013330519199371338,
+      0.010288799181580544,
+      0.016523845493793488,
+      0.010460881516337395,
+      0.0172421894967556,
+      0.015159675851464272,
+      0.029577387496829033,
+      0.017727192491292953,
+      0.015862464904785156,
+      0.013833096250891685,
+      0.01868520677089691,
+      0.032360661774873734,
+      0.012623627670109272,
+      0.011180063709616661,
+      0.023935578763484955,
+      0.01221864391118288,
+      0.017148122191429138,
+      0.00921904481947422,
+      0.01488802395761013,
+      0.02933511696755886,
+      0.01066274382174015,
+      0.011872964911162853,
+      0.030374063178896904,
+      0.017614852637052536,
+      0.013273060321807861,
+      0.014252067543566227,
+      0.0119790593162179,
+      0.012053417041897774,
+      0.01714882254600525,
+      0.026015477254986763,
+      0.01101711206138134,
+      0.02366754040122032,
+      0.027670545503497124,
+      0.03956326097249985,
+      0.013280421495437622
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8031302690505981,
+      0.651446521282196,
+      0.47225630283355713,
+      0.31398698687553406,
+      0.2344164401292801,
+      0.20158666372299194,
+      0.1811789870262146,
+      0.11740656942129135,
+      0.09180840104818344,
+      0.06979136914014816,
+      0.05881809443235397,
+      0.05668292194604874,
+      0.05968532711267471,
+      0.06590636074542999,
+      0.07760761678218842,
+      0.07676749676465988,
+      0.08586236834526062,
+      0.08619170635938644,
+      0.08521468937397003,
+      0.09206206351518631,
+      0.0947783812880516,
+      0.10129841417074203,
+      0.10299093276262283,
+      0.1065870001912117,
+      0.10834246873855591,
+      0.11157213151454926,
+      0.11360572278499603,
+      0.11754754185676575,
+      0.12079672515392303,
+      0.12221942842006683,
+      0.12430651485919952,
+      0.12570339441299438,
+      0.12773926556110382,
+      0.13032616674900055,
+      0.1337440013885498,
+      0.13685867190361023,
+      0.1394176036119461,
+      0.14119425415992737,
+      0.143585667014122,
+      0.1459008753299713,
+      0.14823469519615173,
+      0.1483970731496811,
+      0.14824025332927704,
+      0.1483682096004486,
+      0.14987003803253174,
+      0.1468847692012787,
+      0.14886698126792908,
+      0.149628683924675,
+      0.14818134903907776,
+      0.1488112509250641,
+      0.1479089856147766,
+      0.14857356250286102,
+      0.14692968130111694,
+      0.14757348597049713,
+      0.15208891034126282,
+      0.15442827343940735,
+      0.15509766340255737,
+      0.1590951681137085,
+      0.1531846821308136,
+      0.15168388187885284,
+      0.15258321166038513,
+      0.15024863183498383,
+      0.15156753361225128,
+      0.15119343996047974,
+      0.15037044882774353,
+      0.15048056840896606,
+      0.14808736741542816,
+      0.14513351023197174,
+      0.1465653032064438,
+      0.14518895745277405,
+      0.1469978243112564,
+      0.15150853991508484,
+      0.1535106748342514,
+      0.15811696648597717,
+      0.15943248569965363,
+      0.15731456875801086,
+      0.1617055982351303,
+      0.16405713558197021,
+      0.16909068822860718,
+      0.17723920941352844,
+      0.17886748909950256,
+      0.17061243951320648,
+      0.17493830621242523,
+      0.17122139036655426,
+      0.16704106330871582,
+      0.1666063666343689,
+      0.1669788807630539,
+      0.16966800391674042,
+      0.17728915810585022,
+      0.17715327441692352,
+      0.17924952507019043,
+      0.1768399178981781,
+      0.1799253523349762,
+      0.18971866369247437,
+      0.21061910688877106,
+      0.22413702309131622,
+      0.23895341157913208,
+      0.25049567222595215,
+      0.2607962489128113,
+      0.2692263424396515,
+      0.28197380900382996,
+      0.28492170572280884,
+      0.2932630479335785,
+      0.29921630024909973,
+      0.30437731742858887,
+      0.3101743161678314,
+      0.3136626183986664,
+      0.31526586413383484,
+      0.32267218828201294,
+      0.33414366841316223,
+      0.3383929133415222,
+      0.343045175075531,
+      0.3495849072933197,
+      0.35210493206977844,
+      0.3652360439300537,
+      0.36629679799079895,
+      0.3676069676876068,
+      0.3763442933559418,
+      0.38191697001457214,
+      0.38559967279434204,
+      0.39005047082901,
+      0.39169827103614807,
+      0.3972143232822418,
+      0.3990357220172882,
+      0.40587127208709717,
+      0.4073086678981781,
+      0.4106364846229553,
+      0.4139386713504791,
+      0.4195164144039154,
+      0.4218643307685852,
+      0.43074914813041687,
+      0.43609270453453064,
+      0.43693414330482483,
+      0.43645963072776794,
+      0.43526560068130493,
+      0.43738865852355957,
+      0.43702515959739685,
+      0.43787628412246704,
+      0.4392624795436859,
+      0.4444485902786255,
+      0.4473108649253845,
+      0.4466269016265869,
+      0.44968798756599426,
+      0.4528096616268158,
+      0.4538198411464691,
+      0.46537908911705017,
+      0.46137139201164246,
+      0.46286407113075256,
+      0.4660347104072571,
+      0.4669730067253113,
+      0.4688158929347992,
+      0.4694839417934418,
+      0.47653937339782715,
+      0.47856423258781433,
+      0.47925111651420593,
+      0.4842240810394287,
+      0.4890192151069641,
+      0.4899299144744873,
+      0.49078568816185,
+      0.48639532923698425,
+      0.48752135038375854,
+      0.4861641228199005,
+      0.48675820231437683,
+      0.4877900779247284,
+      0.49160829186439514,
+      0.492164671421051,
+      0.48940518498420715,
+      0.48711976408958435,
+      0.4914919137954712,
+      0.4937744438648224,
+      0.49668481945991516,
+      0.4990174472332001,
+      0.49750903248786926,
+      0.5013089776039124,
+      0.5032296180725098,
+      0.5059557557106018,
+      0.5062442421913147,
+      0.5080885887145996,
+      0.5070073008537292,
+      0.5054982900619507,
+      0.5073270201683044,
+      0.5032066702842712,
+      0.5031166076660156,
+      0.5061224102973938,
+      0.5110673308372498,
+      0.5123271346092224,
+      0.5108535289764404,
+      0.5125726461410522,
+      0.5130029916763306,
+      0.5180069804191589,
+      0.5202614068984985,
+      0.5197465419769287,
+      0.521242082118988,
+      0.5211664438247681,
+      0.5246459245681763,
+      0.5352731347084045,
+      0.5375922322273254,
+      0.5366690158843994,
+      0.538623571395874,
+      0.538221538066864,
+      0.5386557579040527,
+      0.5386679172515869,
+      0.5378957390785217,
+      0.5390077829360962,
+      0.5402559041976929,
+      0.5461403727531433,
+      0.5466548800468445,
+      0.5490771532058716,
+      0.5516842007637024,
+      0.551599383354187,
+      0.5516263842582703,
+      0.5547515749931335,
+      0.5528894066810608,
+      0.5557861328125,
+      0.5564415454864502,
+      0.5575445294380188,
+      0.5567478537559509,
+      0.5571569204330444,
+      0.5580320358276367,
+      0.5580832362174988,
+      0.5589568018913269,
+      0.5598948001861572,
+      0.5620438456535339,
+      0.5638320446014404,
+      0.5632127523422241,
+      0.5648950338363647,
+      0.5680603384971619,
+      0.5679644346237183,
+      0.5684031248092651,
+      0.5670228004455566,
+      0.5679007768630981,
+      0.5674344897270203,
+      0.5691047310829163,
+      0.5682027339935303,
+      0.5701022744178772,
+      0.5724801421165466,
+      0.5727242827415466,
+      0.5743642449378967,
+      0.5754973292350769,
+      0.5763208866119385,
+      0.5771183967590332,
+      0.5775476694107056,
+      0.5774152874946594,
+      0.5786145925521851,
+      0.5791900753974915,
+      0.5807589292526245,
+      0.5780942440032959,
+      0.5789430141448975,
+      0.5790625810623169,
+      0.580161988735199,
+      0.5809789299964905,
+      0.583286702632904,
+      0.5834190845489502,
+      0.5843989253044128,
+      0.5873239040374756,
+      0.5864567160606384,
+      0.5904589891433716,
+      0.5906545519828796,
+      0.5907052755355835,
+      0.5911244750022888,
+      0.5919567942619324,
+      0.591834545135498,
+      0.5913709402084351,
+      0.5923407077789307,
+      0.5932527184486389,
+      0.5939968228340149,
+      0.5941485166549683,
+      0.5949186086654663,
+      0.5945107936859131,
+      0.5946792960166931,
+      0.5944833159446716,
+      0.5948854088783264,
+      0.5956472158432007,
+      0.5962466597557068,
+      0.5968529582023621,
+      0.5970098972320557,
+      0.5972112417221069,
+      0.5973919630050659,
+      0.5978026390075684,
+      0.5975343585014343,
+      0.5981355309486389,
+      0.5980138182640076,
+      0.598423182964325,
+      0.598626971244812,
+      0.5991927981376648,
+      0.5995078086853027,
+      0.6002234220504761,
+      0.6004067063331604,
+      0.6008868217468262,
+      0.6006950736045837,
+      0.6007959246635437,
+      0.60096275806427,
+      0.6009566187858582,
+      0.601643979549408,
+      0.6017308831214905,
+      0.6019169688224792,
+      0.6020408868789673,
+      0.6028541922569275,
+      0.6031910181045532,
+      0.6032954454421997,
+      0.6035951375961304,
+      0.6039335131645203,
+      0.6041980981826782,
+      0.6043713688850403,
+      0.6045557260513306,
+      0.6045573353767395,
+      0.6046279072761536,
+      0.6046212911605835,
+      0.6046490669250488,
+      0.6049583554267883
+    ],
+    "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": [
+      0.53,
+      0.92,
+      1.0,
+      0.99,
+      0.99,
+      1.0,
+      1.0,
+      1.0,
+      1.0,
+      1.0
+    ]
+  },
+  "final_accuracy": 1.0,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.97,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9950413223140496,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9950980392156863,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9966216216216216,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9923076923076923,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 0.9885057471264368,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.9852216748768473,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.81,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9917355371900827,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.9032258064516129,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.9955156950672646,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.43,
+        "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": 0.675,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.8957654723127035,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.2,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.47,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.67,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.3,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.34,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.636,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.96,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9955257270693513,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.99375,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 0.9821428571428571,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.994328922495274,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 0.9791666666666666,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.83,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9966666666666667,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.917098445595855,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.9532710280373832,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.83,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9375,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.9652777777777778,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.78,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9313725490196079,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.9536082474226805,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.87,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.9672131147540983,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.9786324786324786,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.98,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9966722129783694,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.97,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.989247311827957,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.95,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9906103286384976,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.9834254143646409,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.81,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.9902912621359223,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.8791946308724832,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.32,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.51,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.8833333333333333,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.19,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.59,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.7125,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.965,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9916666666666667,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9954441913439636,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.8,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9866666666666667,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9137055837563451,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.78,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.995,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9311740890688259,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.954248366013072,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.88,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.9664429530201343,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.9801980198019802,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.77,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": 4,
+      "mode": "sorl",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 1.0,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 1.0,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 1.0,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 1.0,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 1.0,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 1.0,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 1.0,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 1.0,
+  "sft_overall_accuracy": 0.77
+}

add_sub_sorl_v1_abs16_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8af91a334202c57fddfa057167168e0fee38a83b32cdc8a0f5a722fbcd79a8a5
+size 650328152

add_sub_sorl_v1_abs16_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_abs16_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": 16,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162505382,
+  "run_name": "add_sub_sorl_v1_abs16_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T18:46:45.514385+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": "nhhjduxx",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/nhhjduxx",
+  "final_accuracy": 1.0,
+  "sft_accuracy": 0.77,
+  "eval_method": "ArithmeticEvaluator"
+}