Upload add_sub_sorl_v1_abs100_100K

add_sub_sorl_v1_abs100_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": 151744
+}

add_sub_sorl_v1_abs100_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_abs100_100K/metrics.json

@@ -0,0 +1,3537 @@
+{
+  "history": {
+    "step": [
+      50,
+      100,
+      150,
+      200,
+      250,
+      300,
+      350,
+      400,
+      450,
+      500,
+      550,
+      600,
+      650,
+      700,
+      750,
+      800,
+      850,
+      900,
+      950,
+      1000,
+      1050,
+      1100,
+      1150,
+      1200,
+      1250,
+      1300,
+      1350,
+      1400,
+      1450,
+      1500,
+      1550,
+      1613,
+      1663,
+      1713,
+      1763,
+      1813,
+      1863,
+      1913,
+      1963,
+      2013,
+      2063,
+      2113,
+      2163,
+      2213,
+      2263,
+      2313,
+      2363,
+      2413,
+      2463,
+      2513,
+      2563,
+      2613,
+      2663,
+      2713,
+      2763,
+      2813,
+      2863,
+      2913,
+      2963,
+      3013,
+      3063,
+      3113,
+      3176,
+      3226,
+      3276,
+      3326,
+      3376,
+      3426,
+      3476,
+      3526,
+      3576,
+      3626,
+      3676,
+      3726,
+      3776,
+      3826,
+      3876,
+      3926,
+      3976,
+      4026,
+      4076,
+      4126,
+      4176,
+      4226,
+      4276,
+      4326,
+      4376,
+      4426,
+      4476,
+      4526,
+      4576,
+      4626,
+      4676,
+      4739,
+      4789,
+      4839,
+      4889,
+      4939,
+      4989,
+      5039,
+      5089,
+      5139,
+      5189,
+      5239,
+      5289,
+      5339,
+      5389,
+      5439,
+      5489,
+      5539,
+      5589,
+      5639,
+      5689,
+      5739,
+      5789,
+      5839,
+      5889,
+      5939,
+      5989,
+      6039,
+      6089,
+      6139,
+      6189,
+      6239,
+      6302,
+      6352,
+      6402,
+      6452,
+      6502,
+      6552,
+      6602,
+      6652,
+      6702,
+      6752,
+      6802,
+      6852,
+      6902,
+      6952,
+      7002,
+      7052,
+      7102,
+      7152,
+      7202,
+      7252,
+      7302,
+      7352,
+      7402,
+      7452,
+      7502,
+      7552,
+      7602,
+      7652,
+      7702,
+      7752,
+      7802,
+      7865,
+      7915,
+      7965,
+      8015,
+      8065,
+      8115,
+      8165,
+      8215,
+      8265,
+      8315,
+      8365,
+      8415,
+      8465,
+      8515,
+      8565,
+      8615,
+      8665,
+      8715,
+      8765,
+      8815,
+      8865,
+      8915,
+      8965,
+      9015,
+      9065,
+      9115,
+      9165,
+      9215,
+      9265,
+      9315,
+      9365,
+      9428,
+      9478,
+      9528,
+      9578,
+      9628,
+      9678,
+      9728,
+      9778,
+      9828,
+      9878,
+      9928,
+      9978,
+      10028,
+      10078,
+      10128,
+      10178,
+      10228,
+      10278,
+      10328,
+      10378,
+      10428,
+      10478,
+      10528,
+      10578,
+      10628,
+      10678,
+      10728,
+      10778,
+      10828,
+      10878,
+      10928,
+      10991,
+      11041,
+      11091,
+      11141,
+      11191,
+      11241,
+      11291,
+      11341,
+      11391,
+      11441,
+      11491,
+      11541,
+      11591,
+      11641,
+      11691,
+      11741,
+      11791,
+      11841,
+      11891,
+      11941,
+      11991,
+      12041,
+      12091,
+      12141,
+      12191,
+      12241,
+      12291,
+      12341,
+      12391,
+      12441,
+      12491,
+      12554,
+      12604,
+      12654,
+      12704,
+      12754,
+      12804,
+      12854,
+      12904,
+      12954,
+      13004,
+      13054,
+      13104,
+      13154,
+      13204,
+      13254,
+      13304,
+      13354,
+      13404,
+      13454,
+      13504,
+      13554,
+      13604,
+      13654,
+      13704,
+      13754,
+      13804,
+      13854,
+      13904,
+      13954,
+      14004,
+      14054,
+      14117,
+      14167,
+      14217,
+      14267,
+      14317,
+      14367,
+      14417,
+      14467,
+      14517,
+      14567,
+      14617,
+      14667,
+      14717,
+      14767,
+      14817,
+      14867,
+      14917,
+      14967,
+      15017,
+      15067,
+      15117,
+      15167,
+      15217,
+      15267,
+      15317,
+      15367,
+      15417,
+      15467,
+      15517,
+      15567,
+      15617
+    ],
+    "loss": [
+      15.550888061523438,
+      11.867366790771484,
+      9.26658821105957,
+      7.580068588256836,
+      4.967007637023926,
+      4.137729644775391,
+      2.9006338119506836,
+      2.4626848697662354,
+      2.5665512084960938,
+      2.158423900604248,
+      1.952378749847412,
+      1.7124090194702148,
+      1.3621506690979004,
+      1.1513828039169312,
+      0.8973356485366821,
+      0.23597657680511475,
+      -0.5959402322769165,
+      -2.1503584384918213,
+      -3.0244522094726562,
+      -4.247900009155273,
+      -6.408154487609863,
+      -7.557421684265137,
+      -9.527976036071777,
+      -9.702749252319336,
+      -10.156671524047852,
+      -10.3717041015625,
+      -11.726086616516113,
+      -10.757834434509277,
+      -12.286270141601562,
+      -12.50685977935791,
+      -12.508800506591797,
+      -13.805336952209473,
+      -13.685751914978027,
+      -13.519683837890625,
+      -14.172358512878418,
+      -13.008854866027832,
+      -14.175028800964355,
+      -14.324581146240234,
+      -14.473095893859863,
+      -13.220258712768555,
+      -13.878575325012207,
+      -14.405760765075684,
+      -14.200833320617676,
+      -13.725854873657227,
+      -13.388266563415527,
+      -14.710071563720703,
+      -14.153650283813477,
+      -14.6085844039917,
+      -13.753926277160645,
+      -14.155021667480469,
+      -14.645832061767578,
+      -14.0811128616333,
+      -14.055008888244629,
+      -13.960396766662598,
+      -14.022454261779785,
+      -13.710693359375,
+      -13.309112548828125,
+      -13.811192512512207,
+      -14.07404899597168,
+      -14.23593521118164,
+      -14.03106689453125,
+      -14.144307136535645,
+      -13.670321464538574,
+      -14.122215270996094,
+      -14.305556297302246,
+      -13.679603576660156,
+      -13.83180046081543,
+      -13.438374519348145,
+      -14.211256980895996,
+      -13.453189849853516,
+      -13.635964393615723,
+      -13.473335266113281,
+      -13.789403915405273,
+      -14.136305809020996,
+      -13.588306427001953,
+      -13.945990562438965,
+      -13.677070617675781,
+      -13.584909439086914,
+      -13.552124977111816,
+      -13.686911582946777,
+      -14.19415283203125,
+      -13.130370140075684,
+      -13.576313018798828,
+      -13.867148399353027,
+      -14.135629653930664,
+      -13.967987060546875,
+      -13.738463401794434,
+      -13.794072151184082,
+      -13.240439414978027,
+      -13.05858039855957,
+      -13.739139556884766,
+      -13.341608047485352,
+      -13.674799919128418,
+      -13.919906616210938,
+      -13.248772621154785,
+      -13.675357818603516,
+      -13.085427284240723,
+      -13.586677551269531,
+      -13.351914405822754,
+      -13.54724407196045,
+      -13.314041137695312,
+      -13.698709487915039,
+      -13.4844970703125,
+      -13.105769157409668,
+      -14.150449752807617,
+      -13.3515625,
+      -13.585944175720215,
+      -13.67232894897461,
+      -13.279473304748535,
+      -12.88558292388916,
+      -12.672972679138184,
+      -12.307419776916504,
+      -11.918647766113281,
+      -11.899198532104492,
+      -10.78943920135498,
+      -9.057609558105469,
+      -9.074422836303711,
+      -7.046210765838623,
+      -6.717620849609375,
+      -5.936412334442139,
+      -5.499674320220947,
+      -5.294518947601318,
+      -4.942630767822266,
+      -4.592720985412598,
+      -4.245693206787109,
+      -4.3543901443481445,
+      -4.608675956726074,
+      -4.525597095489502,
+      -4.042947769165039,
+      -3.8350584506988525,
+      -3.795506715774536,
+      -3.8952653408050537,
+      -3.7787787914276123,
+      -3.690739154815674,
+      -3.6305153369903564,
+      -3.6295809745788574,
+      -4.06860876083374,
+      -3.522972345352173,
+      -3.2608494758605957,
+      -3.4570372104644775,
+      -3.7303266525268555,
+      -3.073056221008301,
+      -3.164405345916748,
+      -2.8519835472106934,
+      -3.254319190979004,
+      -3.214355707168579,
+      -3.1682052612304688,
+      -3.134302854537964,
+      -3.013920307159424,
+      -3.0451102256774902,
+      -3.121530055999756,
+      -2.7288155555725098,
+      -2.995760679244995,
+      -2.8066184520721436,
+      -2.7938923835754395,
+      -2.7591190338134766,
+      -3.200686454772949,
+      -2.7287776470184326,
+      -2.82346773147583,
+      -2.8249335289001465,
+      -2.705054759979248,
+      -2.622191905975342,
+      -2.8604440689086914,
+      -2.8222217559814453,
+      -2.4740147590637207,
+      -2.638058662414551,
+      -2.60479736328125,
+      -2.3116393089294434,
+      -2.6722817420959473,
+      -2.2449727058410645,
+      -2.4268925189971924,
+      -2.9192469120025635,
+      -2.528902530670166,
+      -2.5458292961120605,
+      -2.349612236022949,
+      -2.2004177570343018,
+      -2.4359543323516846,
+      -2.461465835571289,
+      -2.23230242729187,
+      -2.2559080123901367,
+      -2.3587558269500732,
+      -2.391399621963501,
+      -2.832726240158081,
+      -2.714812994003296,
+      -2.3990671634674072,
+      -2.195844888687134,
+      -2.1532039642333984,
+      -2.097494602203369,
+      -2.3985636234283447,
+      -2.570045232772827,
+      -2.2937211990356445,
+      -2.1569249629974365,
+      -2.2520053386688232,
+      -2.0688021183013916,
+      -1.8645415306091309,
+      -1.867919683456421,
+      -2.38214111328125,
+      -1.9624675512313843,
+      -2.0682549476623535,
+      -1.9874458312988281,
+      -2.201204776763916,
+      -1.9760252237319946,
+      -1.85667884349823,
+      -1.843688726425171,
+      -1.9453718662261963,
+      -2.0517611503601074,
+      -1.6996922492980957,
+      -2.085865020751953,
+      -1.709080457687378,
+      -1.7818503379821777,
+      -1.9570668935775757,
+      -1.7454025745391846,
+      -1.6972360610961914,
+      -1.8396483659744263,
+      -1.5027493238449097,
+      -1.907371163368225,
+      -1.7743029594421387,
+      -1.9746217727661133,
+      -1.338217854499817,
+      -1.6359403133392334,
+      -1.3770235776901245,
+      -1.7120670080184937,
+      -1.3594249486923218,
+      -1.662591576576233,
+      -1.4001332521438599,
+      -1.5567529201507568,
+      -1.8013702630996704,
+      -1.7820297479629517,
+      -1.6045730113983154,
+      -1.5203410387039185,
+      -1.5714890956878662,
+      -1.5866961479187012,
+      -1.4439531564712524,
+      -1.8579603433609009,
+      -1.571858525276184,
+      -1.3905956745147705,
+      -1.4180278778076172,
+      -1.6811977624893188,
+      -1.6187727451324463,
+      -1.5407252311706543,
+      -1.5873008966445923,
+      -1.6828949451446533,
+      -1.692954421043396,
+      -1.3257203102111816,
+      -1.4703478813171387,
+      -1.4493123292922974,
+      -1.4972003698349,
+      -1.4250011444091797,
+      -1.5240806341171265,
+      -1.230216145515442,
+      -1.4708086252212524,
+      -1.2354530096054077,
+      -1.7644381523132324,
+      -1.31303071975708,
+      -1.1198210716247559,
+      -1.3806722164154053,
+      -1.040693998336792,
+      -1.407912254333496,
+      -1.1439356803894043,
+      -1.4109939336776733,
+      -1.2327980995178223,
+      -1.1438546180725098,
+      -1.2574481964111328,
+      -1.3094208240509033,
+      -1.2854642868041992,
+      -1.417351484298706,
+      -1.161266565322876,
+      -1.1692018508911133,
+      -1.1951680183410645,
+      -1.4501426219940186,
+      -1.0620129108428955,
+      -1.0999974012374878,
+      -1.2486470937728882,
+      -0.8803353905677795,
+      -1.3792674541473389,
+      -1.3971298933029175,
+      -1.175848364830017,
+      -1.2047163248062134,
+      -1.1166435480117798,
+      -1.0748188495635986,
+      -1.260925531387329,
+      -1.0387357473373413,
+      -1.1274218559265137,
+      -1.011070966720581,
+      -0.9395217895507812,
+      -1.3620796203613281,
+      -1.258716106414795,
+      -1.041094183921814,
+      -1.2385952472686768,
+      -1.3618378639221191,
+      -1.2178571224212646,
+      -1.1151763200759888,
+      -0.8671252727508545,
+      -1.1393119096755981,
+      -1.0050714015960693,
+      -1.2483044862747192,
+      -1.0898106098175049,
+      -1.0313324928283691,
+      -0.8652176856994629,
+      -1.0769447088241577,
+      -0.8985112905502319,
+      -0.9062784314155579,
+      -0.9375243186950684,
+      -0.9505935907363892,
+      -1.046338438987732,
+      -1.0050792694091797,
+      -0.9650892615318298,
+      -0.9858003258705139,
+      -1.1620886325836182,
+      -1.1846257448196411
+    ],
+    "base_loss": [
+      10.609814643859863,
+      8.015226364135742,
+      7.298034191131592,
+      6.2845025062561035,
+      5.197947978973389,
+      3.4349818229675293,
+      2.1448423862457275,
+      2.040243148803711,
+      1.8989087343215942,
+      1.8234024047851562,
+      1.8464136123657227,
+      1.8160383701324463,
+      1.8420960903167725,
+      1.8088819980621338,
+      1.8110955953598022,
+      1.821955919265747,
+      1.8283952474594116,
+      1.920448899269104,
+      1.8867542743682861,
+      1.8831616640090942,
+      1.9312572479248047,
+      1.9106019735336304,
+      1.9592641592025757,
+      1.8384157419204712,
+      1.8146497011184692,
+      1.7597582340240479,
+      1.8155019283294678,
+      1.6613296270370483,
+      1.707415223121643,
+      1.7329599857330322,
+      1.637532114982605,
+      1.7226206064224243,
+      1.699408769607544,
+      1.6609095335006714,
+      1.7118498086929321,
+      1.5943775177001953,
+      1.6790682077407837,
+      1.6773242950439453,
+      1.6923757791519165,
+      1.6028581857681274,
+      1.60297429561615,
+      1.6737611293792725,
+      1.62065589427948,
+      1.607097864151001,
+      1.580548644065857,
+      1.6909033060073853,
+      1.60847008228302,
+      1.6662300825119019,
+      1.6017462015151978,
+      1.6096265316009521,
+      1.6725680828094482,
+      1.5998142957687378,
+      1.5998982191085815,
+      1.600284218788147,
+      1.5952374935150146,
+      1.578301191329956,
+      1.5071732997894287,
+      1.5789823532104492,
+      1.5953046083450317,
+      1.6094692945480347,
+      1.5920847654342651,
+      1.5985146760940552,
+      1.5503194332122803,
+      1.6095918416976929,
+      1.6245962381362915,
+      1.5575997829437256,
+      1.5648068189620972,
+      1.5228277444839478,
+      1.603353500366211,
+      1.5222351551055908,
+      1.544053077697754,
+      1.5283054113388062,
+      1.5700281858444214,
+      1.5998828411102295,
+      1.5484613180160522,
+      1.569875717163086,
+      1.5532366037368774,
+      1.5442272424697876,
+      1.5343350172042847,
+      1.5506829023361206,
+      1.5976637601852417,
+      1.4877575635910034,
+      1.52931809425354,
+      1.5564076900482178,
+      1.6029837131500244,
+      1.5728864669799805,
+      1.5477153062820435,
+      1.5558513402938843,
+      1.4923717975616455,
+      1.471130609512329,
+      1.5454028844833374,
+      1.5082389116287231,
+      1.5537184476852417,
+      1.561993956565857,
+      1.5067471265792847,
+      1.5380361080169678,
+      1.481860876083374,
+      1.5293447971343994,
+      1.5173250436782837,
+      1.5268032550811768,
+      1.4966020584106445,
+      1.5563942193984985,
+      1.5178576707839966,
+      1.4765022993087769,
+      1.58711838722229,
+      1.5146647691726685,
+      1.5267279148101807,
+      1.5433595180511475,
+      1.4922010898590088,
+      1.4545576572418213,
+      1.4353135824203491,
+      1.3946800231933594,
+      1.3413666486740112,
+      1.3423233032226562,
+      1.2201385498046875,
+      1.0377111434936523,
+      1.0341635942459106,
+      0.8037133812904358,
+      0.767292857170105,
+      0.6757163405418396,
+      0.6260315775871277,
+      0.6115191578865051,
+      0.5719292759895325,
+      0.525610625743866,
+      0.48804253339767456,
+      0.513914942741394,
+      0.5368272066116333,
+      0.5152927041053772,
+      0.463877409696579,
+      0.4421793818473816,
+      0.44104495644569397,
+      0.4481094181537628,
+      0.4658953845500946,
+      0.4311184287071228,
+      0.4164181649684906,
+      0.4223596155643463,
+      0.46746915578842163,
+      0.4056954085826874,
+      0.38447755575180054,
+      0.39704015851020813,
+      0.4316771626472473,
+      0.3543528616428375,
+      0.36644411087036133,
+      0.3291913568973541,
+      0.3855384290218353,
+      0.3765328824520111,
+      0.36499664187431335,
+      0.3587031960487366,
+      0.3511146903038025,
+      0.3541579246520996,
+      0.358995258808136,
+      0.3450600206851959,
+      0.3578762412071228,
+      0.3288351893424988,
+      0.34160321950912476,
+      0.3207271099090576,
+      0.36676672101020813,
+      0.3151433765888214,
+      0.3299563229084015,
+      0.32532089948654175,
+      0.31451278924942017,
+      0.3047652542591095,
+      0.3315297067165375,
+      0.3248681426048279,
+      0.28615516424179077,
+      0.30520114302635193,
+      0.2996344566345215,
+      0.2685880959033966,
+      0.30989527702331543,
+      0.2701651453971863,
+      0.2988984286785126,
+      0.3356240689754486,
+      0.2926725745201111,
+      0.29605191946029663,
+      0.27395758032798767,
+      0.25648754835128784,
+      0.2848379909992218,
+      0.28560489416122437,
+      0.2649328112602234,
+      0.26132088899612427,
+      0.2755523920059204,
+      0.27864283323287964,
+      0.32541802525520325,
+      0.3135841488838196,
+      0.2774091362953186,
+      0.2753048837184906,
+      0.25229766964912415,
+      0.24528728425502777,
+      0.290712833404541,
+      0.30390962958335876,
+      0.26389721035957336,
+      0.2526156008243561,
+      0.26138073205947876,
+      0.2451794445514679,
+      0.21623989939689636,
+      0.221551313996315,
+      0.27538686990737915,
+      0.22850997745990753,
+      0.23916366696357727,
+      0.2323302924633026,
+      0.255971223115921,
+      0.23261485993862152,
+      0.21784284710884094,
+      0.2177778035402298,
+      0.2266148030757904,
+      0.23994575440883636,
+      0.1994098275899887,
+      0.24346749484539032,
+      0.20232857763767242,
+      0.20922228693962097,
+      0.23100629448890686,
+      0.20590011775493622,
+      0.2000357061624527,
+      0.22720889747142792,
+      0.18639278411865234,
+      0.22194363176822662,
+      0.2069634348154068,
+      0.2373548299074173,
+      0.15888510644435883,
+      0.1979457139968872,
+      0.16390980780124664,
+      0.19959606230258942,
+      0.16598200798034668,
+      0.19541189074516296,
+      0.16625605523586273,
+      0.18630900979042053,
+      0.21210575103759766,
+      0.20869462192058563,
+      0.18824730813503265,
+      0.18152736127376556,
+      0.18498942255973816,
+      0.18627949059009552,
+      0.17134495079517365,
+      0.2168290913105011,
+      0.18592286109924316,
+      0.16444489359855652,
+      0.16715046763420105,
+      0.1964850276708603,
+      0.18959102034568787,
+      0.18093536794185638,
+      0.18605473637580872,
+      0.19904981553554535,
+      0.1973380595445633,
+      0.15778742730617523,
+      0.1747376173734665,
+      0.17108812928199768,
+      0.17510893940925598,
+      0.16731013357639313,
+      0.17871102690696716,
+      0.14609429240226746,
+      0.17256930470466614,
+      0.14573118090629578,
+      0.20665530860424042,
+      0.15655572712421417,
+      0.13402849435806274,
+      0.16387078166007996,
+      0.1248510554432869,
+      0.16516248881816864,
+      0.13674581050872803,
+      0.16651712357997894,
+      0.14748935401439667,
+      0.1382364183664322,
+      0.15008823573589325,
+      0.15552841126918793,
+      0.15129385888576508,
+      0.16749367117881775,
+      0.13866452872753143,
+      0.1381988376379013,
+      0.14164534211158752,
+      0.17011216282844543,
+      0.12627139687538147,
+      0.13087666034698486,
+      0.14732487499713898,
+      0.10798748582601547,
+      0.16139589250087738,
+      0.16517122089862823,
+      0.1411411613225937,
+      0.1421668827533722,
+      0.1339568793773651,
+      0.12969021499156952,
+      0.1501157432794571,
+      0.12624980509281158,
+      0.13378222286701202,
+      0.12192026525735855,
+      0.11317621916532516,
+      0.16144336760044098,
+      0.1496235579252243,
+      0.12298120558261871,
+      0.14827506244182587,
+      0.16108521819114685,
+      0.1463118940591812,
+      0.13234232366085052,
+      0.10579963773488998,
+      0.13623115420341492,
+      0.12044007331132889,
+      0.14652453362941742,
+      0.13023699820041656,
+      0.12433119863271713,
+      0.10616353899240494,
+      0.1289213001728058,
+      0.10968109220266342,
+      0.11077271401882172,
+      0.11283662170171738,
+      0.11493217945098877,
+      0.12461324781179428,
+      0.12140349298715591,
+      0.11627321690320969,
+      0.11910700052976608,
+      0.13995158672332764,
+      0.1414678394794464
+    ],
+    "info_loss": [
+      -0.11385345458984375,
+      -0.14343023300170898,
+      -0.20502328872680664,
+      -0.15923309326171875,
+      -0.2174081802368164,
+      -0.0420684814453125,
+      -0.014981746673583984,
+      -0.04172956943511963,
+      -0.015005826950073242,
+      -0.04786980152130127,
+      -0.07152450084686279,
+      -0.09154641628265381,
+      -0.12877655029296875,
+      -0.1466604471206665,
+      -0.17095613479614258,
+      -0.2396688461303711,
+      -0.3221203088760376,
+      -0.4862041473388672,
+      -0.5681549310684204,
+      -0.684140682220459,
+      -0.8983767032623291,
+      -1.0017673969268799,
+      -1.2004902362823486,
+      -1.1987392902374268,
+      -1.2378183603286743,
+      -1.2491450309753418,
+      -1.3905093669891357,
+      -1.2777379751205444,
+      -1.4317843914031982,
+      -1.4551732540130615,
+      -1.4445528984069824,
+      -1.5836052894592285,
+      -1.567819595336914,
+      -1.546617031097412,
+      -1.6144828796386719,
+      -1.4889076948165894,
+      -1.6100263595581055,
+      -1.625740647315979,
+      -1.6420310735702515,
+      -1.507124900817871,
+      -1.5704201459884644,
+      -1.6301132440567017,
+      -1.6036299467086792,
+      -1.5542329549789429,
+      -1.5175899267196655,
+      -1.6583689451217651,
+      -1.5958442687988281,
+      -1.6468164920806885,
+      -1.5525816679000854,
+      -1.5934239625930786,
+      -1.647326946258545,
+      -1.5843557119369507,
+      -1.5815900564193726,
+      -1.571997880935669,
+      -1.576888918876648,
+      -1.5461136102676392,
+      -1.497298240661621,
+      -1.5533920526504517,
+      -1.5812208652496338,
+      -1.5985703468322754,
+      -1.5774600505828857,
+      -1.5853532552719116,
+      -1.5370008945465088,
+      -1.5863351821899414,
+      -1.606804609298706,
+      -1.537649393081665,
+      -1.5525469779968262,
+      -1.5101174116134644,
+      -1.5952324867248535,
+      -1.5109450817108154,
+      -1.5317682027816772,
+      -1.5129121541976929,
+      -1.549247145652771,
+      -1.586357593536377,
+      -1.527389407157898,
+      -1.5647804737091064,
+      -1.5350285768508911,
+      -1.525258183479309,
+      -1.5214704275131226,
+      -1.5375951528549194,
+      -1.5907458066940308,
+      -1.4744808673858643,
+      -1.5221446752548218,
+      -1.5537238121032715,
+      -1.585281252861023,
+      -1.5659524202346802,
+      -1.5400159358978271,
+      -1.5463478565216064,
+      -1.484694004058838,
+      -1.46498441696167,
+      -1.5409584045410156,
+      -1.4950144290924072,
+      -1.5345021486282349,
+      -1.5598196983337402,
+      -1.4889129400253296,
+      -1.5336635112762451,
+      -1.4689878225326538,
+      -1.5244895219802856,
+      -1.4982454776763916,
+      -1.5184264183044434,
+      -1.4915274381637573,
+      -1.5377562046051025,
+      -1.5106768608093262,
+      -1.4720402956008911,
+      -1.5842334032058716,
+      -1.5001106262207031,
+      -1.5235495567321777,
+      -1.5334497690200806,
+      -1.4875925779342651,
+      -1.4477195739746094,
+      -1.4215354919433594,
+      -1.3809431791305542,
+      -1.3361492156982422,
+      -1.3366215229034424,
+      -1.2126234769821167,
+      -1.0218498706817627,
+      -1.0225090980529785,
+      -0.796761155128479,
+      -0.7610917091369629,
+      -0.6727439761161804,
+      -0.6226391792297363,
+      -0.6016292572021484,
+      -0.5633807182312012,
+      -0.5234792232513428,
+      -0.4851091206073761,
+      -0.497601717710495,
+      -0.5263352394104004,
+      -0.5140460133552551,
+      -0.4624413251876831,
+      -0.4394057095050812,
+      -0.43588510155677795,
+      -0.44685807824134827,
+      -0.43609413504600525,
+      -0.4237522780895233,
+      -0.41476190090179443,
+      -0.41749295592308044,
+      -0.4634345769882202,
+      -0.40468814969062805,
+      -0.3737349510192871,
+      -0.396659791469574,
+      -0.4255046546459198,
+      -0.3532746434211731,
+      -0.36393654346466064,
+      -0.3283543288707733,
+      -0.3739100396633148,
+      -0.368809312582016,
+      -0.36335042119026184,
+      -0.3583308458328247,
+      -0.34817737340927124,
+      -0.35022154450416565,
+      -0.3576264977455139,
+      -0.31730690598487854,
+      -0.3450942039489746,
+      -0.3241146504878998,
+      -0.3240019679069519,
+      -0.318650484085083,
+      -0.3664001524448395,
+      -0.31477200984954834,
+      -0.3253546953201294,
+      -0.32520633935928345,
+      -0.3138948380947113,
+      -0.30230018496513367,
+      -0.32947179675102234,
+      -0.32324084639549255,
+      -0.2853688895702362,
+      -0.3047855496406555,
+      -0.29905828833580017,
+      -0.26823070645332336,
+      -0.30898529291152954,
+      -0.26060950756073,
+      -0.2832368314266205,
+      -0.3351980745792389,
+      -0.29253584146499634,
+      -0.2955663502216339,
+      -0.2738269865512848,
+      -0.2561382055282593,
+      -0.28282296657562256,
+      -0.28397995233535767,
+      -0.26278093457221985,
+      -0.2610403001308441,
+      -0.2754134237766266,
+      -0.27838048338890076,
+      -0.32534337043762207,
+      -0.31341785192489624,
+      -0.2773230969905853,
+      -0.2570829391479492,
+      -0.24993328750133514,
+      -0.24398627877235413,
+      -0.279629111289978,
+      -0.2965335547924042,
+      -0.2636924684047699,
+      -0.25213974714279175,
+      -0.2612018883228302,
+      -0.24196603894233704,
+      -0.2161180078983307,
+      -0.22004133462905884,
+      -0.27529701590538025,
+      -0.22714611887931824,
+      -0.23910602927207947,
+      -0.23192031681537628,
+      -0.2548813223838806,
+      -0.23050589859485626,
+      -0.21774822473526,
+      -0.21656401455402374,
+      -0.2265331894159317,
+      -0.2396659255027771,
+      -0.19935452938079834,
+      -0.24267248809337616,
+      -0.19900815188884735,
+      -0.20903536677360535,
+      -0.2290165275335312,
+      -0.20365144312381744,
+      -0.1999422162771225,
+      -0.21606677770614624,
+      -0.17974726855754852,
+      -0.2214144915342331,
+      -0.20677374303340912,
+      -0.23102223873138428,
+      -0.15881958603858948,
+      -0.19410057365894318,
+      -0.16373766958713531,
+      -0.1994384080171585,
+      -0.16169355809688568,
+      -0.19534216821193695,
+      -0.16603849828243256,
+      -0.1837719827890396,
+      -0.21206045150756836,
+      -0.20862361788749695,
+      -0.188018798828125,
+      -0.18006832897663116,
+      -0.18477082252502441,
+      -0.1862453669309616,
+      -0.1709631085395813,
+      -0.21677134931087494,
+      -0.18473117053508759,
+      -0.1642923802137375,
+      -0.16710221767425537,
+      -0.19642113149166107,
+      -0.18956467509269714,
+      -0.18083183467388153,
+      -0.18593645095825195,
+      -0.19899502396583557,
+      -0.1973118633031845,
+      -0.15775373578071594,
+      -0.17471984028816223,
+      -0.1710720956325531,
+      -0.1750890463590622,
+      -0.16668885946273804,
+      -0.17866910994052887,
+      -0.14603638648986816,
+      -0.17250952124595642,
+      -0.14571359753608704,
+      -0.2065526694059372,
+      -0.15653041005134583,
+      -0.1340152472257614,
+      -0.16376584768295288,
+      -0.12481392920017242,
+      -0.16514933109283447,
+      -0.13672783970832825,
+      -0.16649629175662994,
+      -0.146728515625,
+      -0.1372392326593399,
+      -0.15005478262901306,
+      -0.15551863610744476,
+      -0.1512700915336609,
+      -0.16742905974388123,
+      -0.1386333853006363,
+      -0.13817352056503296,
+      -0.14161676168441772,
+      -0.17008331418037415,
+      -0.12625375390052795,
+      -0.13071495294570923,
+      -0.1473049819469452,
+      -0.10797560960054398,
+      -0.16138042509555817,
+      -0.16511429846286774,
+      -0.1411227285861969,
+      -0.14215360581874847,
+      -0.1339317113161087,
+      -0.1294330358505249,
+      -0.1500917226076126,
+      -0.12619058787822723,
+      -0.133761465549469,
+      -0.12187721580266953,
+      -0.11312084645032883,
+      -0.16143116354942322,
+      -0.14961135387420654,
+      -0.12295876443386078,
+      -0.1482592076063156,
+      -0.1610305905342102,
+      -0.1462990939617157,
+      -0.1323220580816269,
+      -0.1057858020067215,
+      -0.13618363440036774,
+      -0.12043294310569763,
+      -0.14648698270320892,
+      -0.13023099303245544,
+      -0.12432308495044708,
+      -0.10615240782499313,
+      -0.12891289591789246,
+      -0.10966979712247849,
+      -0.11076144129037857,
+      -0.11282671242952347,
+      -0.11491944640874863,
+      -0.12460529059171677,
+      -0.12138896435499191,
+      -0.11626255512237549,
+      -0.1190897598862648,
+      -0.13994117081165314,
+      -0.1414576917886734
+    ],
+    "abs_loss": [
+      4.600222110748291,
+      4.585060119628906,
+      4.535017967224121,
+      4.411112308502197,
+      4.214651107788086,
+      3.500492572784424,
+      3.2125346660614014,
+      3.110368251800537,
+      3.050847053527832,
+      3.1303863525390625,
+      3.1481270790100098,
+      3.1907219886779785,
+      3.121077299118042,
+      3.122469902038574,
+      2.9989373683929443,
+      3.169970989227295,
+      3.1337380409240723,
+      3.1850204467773438,
+      3.1595168113708496,
+      2.9852089881896973,
+      3.003159761428833,
+      2.801137685775757,
+      2.5646042823791504,
+      2.432736396789551,
+      1.999617338180542,
+      1.8525265455245972,
+      1.9091401100158691,
+      1.9688462018966675,
+      1.8593182563781738,
+      1.6353579759597778,
+      1.7597875595092773,
+      1.9715489149093628,
+      1.6212482452392578,
+      1.7391170263290405,
+      1.5134029388427734,
+      1.6808552742004395,
+      1.5621697902679443,
+      1.6727452278137207,
+      1.6910449266433716,
+      1.6261968612670898,
+      1.3995141983032227,
+      1.3336455821990967,
+      1.5245065689086914,
+      1.3680379390716553,
+      1.4579674005508423,
+      1.1433132886886597,
+      1.4170444011688232,
+      1.4678308963775635,
+      1.2090680599212646,
+      1.0885725021362305,
+      1.0543391704559326,
+      1.1062440872192383,
+      1.0231534242630005,
+      1.0875221490859985,
+      1.1583337783813477,
+      1.048989176750183,
+      1.275425672531128,
+      0.8260887265205383,
+      1.1685587167739868,
+      0.8801293969154358,
+      0.8248441219329834,
+      0.8601986765861511,
+      0.965746283531189,
+      0.8020682334899902,
+      0.8892130255699158,
+      0.8327577114105225,
+      0.8905668258666992,
+      0.9484708905220032,
+      0.9553885459899902,
+      0.8176908493041992,
+      0.8552860617637634,
+      0.830318808555603,
+      0.7996451258659363,
+      0.666130781173706,
+      1.0367419719696045,
+      0.8224754929542542,
+      0.9066834449768066,
+      0.7016845345497131,
+      0.814628541469574,
+      0.6553899049758911,
+      0.7444800138473511,
+      0.7174248695373535,
+      0.749908447265625,
+      0.7693836688995361,
+      0.7470314502716064,
+      0.8517244458198547,
+      0.6488890051841736,
+      0.7687116861343384,
+      0.75773024559021,
+      0.7646090984344482,
+      0.7204738855361938,
+      0.5489466190338135,
+      0.8591753244400024,
+      0.8850686550140381,
+      0.7457890510559082,
+      0.8056696653366089,
+      0.8558416366577148,
+      0.866152286529541,
+      0.6059410572052002,
+      0.6956859230995178,
+      0.6744041442871094,
+      0.7838091254234314,
+      0.690622091293335,
+      0.9125334024429321,
+      0.5624383687973022,
+      0.7412213087081909,
+      0.7233027219772339,
+      0.4596283733844757,
+      0.6814550161361694,
+      0.8564143180847168,
+      0.6888761520385742,
+      0.835689127445221,
+      0.6428424715995789,
+      0.6466372013092041,
+      0.7571539878845215,
+      0.7585248947143555,
+      0.6671715974807739,
+      0.8459024429321289,
+      0.7264590263366699,
+      0.8340179920196533,
+      0.7317735552787781,
+      0.6521148681640625,
+      0.7353276014328003,
+      0.8445788621902466,
+      0.8306164741516113,
+      0.5943429470062256,
+      0.6313499808311462,
+      0.6874839067459106,
+      0.6386463046073914,
+      0.6640372276306152,
+      0.6259667277336121,
+      0.5979993343353271,
+      0.6901720762252808,
+      0.573563814163208,
+      0.6068687438964844,
+      0.6457435488700867,
+      0.5819556713104248,
+      0.5604647994041443,
+      0.5882904529571533,
+      0.6295926570892334,
+      0.6285600662231445,
+      0.5659893751144409,
+      0.4942191243171692,
+      0.591978907585144,
+      0.5950663685798645,
+      0.49396175146102905,
+      0.6858351230621338,
+      0.4591514766216278,
+      0.8484950661659241,
+      0.6848112940788269,
+      0.6612932085990906,
+      0.5308752059936523,
+      0.5206629037857056,
+      0.7018767595291138,
+      0.5639657974243164,
+      0.5794847011566162,
+      0.3803393542766571,
+      0.5580293536186218,
+      0.5711216926574707,
+      0.5046465396881104,
+      0.44330233335494995,
+      0.5817766189575195,
+      0.5793009400367737,
+      0.5169496536254883,
+      0.5030026435852051,
+      0.6168398261070251,
+      0.5473909974098206,
+      0.5264185667037964,
+      0.6585381031036377,
+      0.40417736768722534,
+      0.6081347465515137,
+      0.6516579389572144,
+      0.4337080121040344,
+      0.42822009325027466,
+      0.608845591545105,
+      0.4710497260093689,
+      0.5425891876220703,
+      0.7154373526573181,
+      0.5240054130554199,
+      0.6915121078491211,
+      0.438113808631897,
+      0.6105717420578003,
+      0.4793950021266937,
+      0.5409137010574341,
+      0.41575175523757935,
+      0.6154731512069702,
+      0.6129148006439209,
+      0.5004086494445801,
+      0.577346920967102,
+      0.511862576007843,
+      0.4338465631008148,
+      0.5168926119804382,
+      0.562402606010437,
+      0.49650275707244873,
+      0.431353360414505,
+      0.5206334590911865,
+      0.5558791160583496,
+      0.4398925006389618,
+      0.4828936755657196,
+      0.6310799717903137,
+      0.39086517691612244,
+      0.5998769998550415,
+      0.5554673671722412,
+      0.5007154941558838,
+      0.5605072975158691,
+      0.5763310194015503,
+      0.44387757778167725,
+      0.5221637487411499,
+      0.4436303973197937,
+      0.6459587812423706,
+      0.47150060534477234,
+      0.458065927028656,
+      0.4076268672943115,
+      0.6684603691101074,
+      0.46604329347610474,
+      0.4915527105331421,
+      0.4971369206905365,
+      0.4907045364379883,
+      0.618830144405365,
+      0.4259473383426666,
+      0.49789100885391235,
+      0.49302276968955994,
+      0.5082845687866211,
+      0.48034894466400146,
+      0.39231348037719727,
+      0.5948460698127747,
+      0.6186717748641968,
+      0.5783128142356873,
+      0.4655434489250183,
+      0.48685896396636963,
+      0.5358094573020935,
+      0.4071079194545746,
+      0.5148466229438782,
+      0.4136655628681183,
+      0.41432803869247437,
+      0.4495674967765808,
+      0.5279877185821533,
+      0.5033449530601501,
+      0.5748267769813538,
+      0.5323889255523682,
+      0.49586522579193115,
+      0.46038562059402466,
+      0.5176984667778015,
+      0.4125681519508362,
+      0.6684879064559937,
+      0.4497949182987213,
+      0.396254301071167,
+      0.4728800058364868,
+      0.4475850462913513,
+      0.3921929597854614,
+      0.4662751257419586,
+      0.4294249415397644,
+      0.4877370595932007,
+      0.5649451017379761,
+      0.4627125859260559,
+      0.6273665428161621,
+      0.4333481788635254,
+      0.39706796407699585,
+      0.45926716923713684,
+      0.5503664612770081,
+      0.4088195562362671,
+      0.49588871002197266,
+      0.449713796377182,
+      0.4749121069908142,
+      0.5020171403884888,
+      0.3331933617591858,
+      0.3349999785423279,
+      0.4275625944137573,
+      0.38952159881591797,
+      0.3819730281829834,
+      0.42513060569763184,
+      0.3870682120323181,
+      0.5191415548324585,
+      0.4761240780353546,
+      0.41521960496902466,
+      0.5030770301818848,
+      0.5881625413894653,
+      0.37058791518211365,
+      0.4363459348678589,
+      0.4512035846710205,
+      0.5564160346984863,
+      0.4164927005767822,
+      0.4175732731819153,
+      0.5126424431800842,
+      0.36320215463638306,
+      0.4501055181026459,
+      0.4496520161628723,
+      0.3972645401954651,
+      0.3614107072353363,
+      0.3568313717842102,
+      0.43815040588378906,
+      0.4537099599838257,
+      0.433208703994751,
+      0.4528942406177521,
+      0.3446529507637024,
+      0.4783862829208374,
+      0.37083593010902405,
+      0.4587956666946411,
+      0.5160912275314331,
+      0.42571330070495605,
+      0.4347543716430664,
+      0.29358673095703125,
+      0.398557186126709,
+      0.3981517553329468,
+      0.43517670035362244,
+      0.3059583902359009,
+      0.45012903213500977,
+      0.4453981816768646,
+      0.46157723665237427,
+      0.4827229380607605
+    ],
+    "zipf_loss": [
+      5.619585990905762,
+      4.827937126159668,
+      3.5652852058410645,
+      2.4467859268188477,
+      1.5216761827468872,
+      0.773383617401123,
+      0.5843554735183716,
+      0.5287005305290222,
+      0.5126160383224487,
+      0.5006808638572693,
+      0.5063974857330322,
+      0.49276259541511536,
+      0.4957123398780823,
+      0.49685826897621155,
+      0.49590766429901123,
+      0.49371200799942017,
+      0.483493834733963,
+      0.4727323055267334,
+      0.4543910026550293,
+      0.411824107170105,
+      0.34403884410858154,
+      0.2695366144180298,
+      0.26120150089263916,
+      0.20295308530330658,
+      0.20690077543258667,
+      0.17473576962947845,
+      0.17259135842323303,
+      0.16133253276348114,
+      0.13822603225708008,
+      0.14837780594825745,
+      0.1232171356678009,
+      0.1109403520822525,
+      0.13091117143630981,
+      0.11166486144065857,
+      0.10927983373403549,
+      0.11775951087474823,
+      0.08994985371828079,
+      0.08822619915008545,
+      0.08573438227176666,
+      0.08551208674907684,
+      0.08270096033811569,
+      0.0882456973195076,
+      0.0623585470020771,
+      0.07257397472858429,
+      0.061287589371204376,
+      0.06838192045688629,
+      0.05461812764406204,
+      0.04656702280044556,
+      0.04923718422651291,
+      0.06073414906859398,
+      0.049436502158641815,
+      0.052005868405103683,
+      0.058678172528743744,
+      0.0505443811416626,
+      0.035363972187042236,
+      0.06724280118942261,
+      0.02915341965854168,
+      0.061135806143283844,
+      0.025998758152127266,
+      0.0522865429520607,
+      0.06896450370550156,
+      0.02469157986342907,
+      0.052792251110076904,
+      0.05133802443742752,
+      0.048972226679325104,
+      0.056013695895671844,
+      0.039805732667446136,
+      0.045125462114810944,
+      0.042175572365522385,
+      0.05225789174437523,
+      0.05213606730103493,
+      0.044449709355831146,
+      0.053074367344379425,
+      0.06077408045530319,
+      0.03345252200961113,
+      0.049690522253513336,
+      0.02930966019630432,
+      0.05327582359313965,
+      0.04678161069750786,
+      0.07281782478094101,
+      0.04119352996349335,
+      0.054937973618507385,
+      0.040824998170137405,
+      0.03674336150288582,
+      0.03949633240699768,
+      0.03347763046622276,
+      0.04909197986125946,
+      0.036684438586235046,
+      0.038356099277734756,
+      0.04367246478796005,
+      0.05299493670463562,
+      0.04540283977985382,
+      0.030584828928112984,
+      0.027790088206529617,
+      0.05903084576129913,
+      0.042674530297517776,
+      0.037005726248025894,
+      0.042257245630025864,
+      0.052620261907577515,
+      0.04064880684018135,
+      0.03719138354063034,
+      0.044077176600694656,
+      0.035351477563381195,
+      0.04687754064798355,
+      0.04852171614766121,
+      0.06075644493103027,
+      0.05049310624599457,
+      0.07284531742334366,
+      0.03610651195049286,
+      0.05141353979706764,
+      0.03818117827177048,
+      0.02376411110162735,
+      0.03719331696629524,
+      0.06002876162528992,
+      0.04094189777970314,
+      0.04732527583837509,
+      0.04978743568062782,
+      0.03309670835733414,
+      0.05335727706551552,
+      0.03190968185663223,
+      0.027508515864610672,
+      0.0450429767370224,
+      0.04571438953280449,
+      0.03200307860970497,
+      0.034294165670871735,
+      0.04827748239040375,
+      0.0547143816947937,
+      0.03082178719341755,
+      0.05372374877333641,
+      0.05041574686765671,
+      0.059702519327402115,
+      0.06540592759847641,
+      0.047250136733055115,
+      0.05830858647823334,
+      0.03999902307987213,
+      0.058414652943611145,
+      0.04007250815629959,
+      0.06216740608215332,
+      0.033193439245224,
+      0.0495615154504776,
+      0.030186738818883896,
+      0.048738185316324234,
+      0.05909387767314911,
+      0.04317042976617813,
+      0.0397363118827343,
+      0.04780852422118187,
+      0.03171863406896591,
+      0.04438743367791176,
+      0.03188920393586159,
+      0.03446602076292038,
+      0.029610421508550644,
+      0.046105869114398956,
+      0.0452386848628521,
+      0.03550521284341812,
+      0.04812770336866379,
+      0.04871044307947159,
+      0.058514468371868134,
+      0.04799622297286987,
+      0.04301080107688904,
+      0.051344092935323715,
+      0.07505089789628983,
+      0.037866994738578796,
+      0.044814467430114746,
+      0.03362371772527695,
+      0.04321867227554321,
+      0.0429118350148201,
+      0.031412024050951004,
+      0.0494377501308918,
+      0.041822005063295364,
+      0.050539448857307434,
+      0.04576407000422478,
+      0.03194405138492584,
+      0.0604124441742897,
+      0.0709601491689682,
+      0.05381552129983902,
+      0.05737155303359032,
+      0.053178612142801285,
+      0.021184831857681274,
+      0.07817338407039642,
+      0.02402309700846672,
+      0.07601477205753326,
+      0.052705395966768265,
+      0.04735003411769867,
+      0.05169006064534187,
+      0.05517950281500816,
+      0.03813251107931137,
+      0.03253985941410065,
+      0.04704003781080246,
+      0.04927995428442955,
+      0.040194395929574966,
+      0.03592187911272049,
+      0.0601678229868412,
+      0.04239239543676376,
+      0.05602855607867241,
+      0.03726328909397125,
+      0.05887894704937935,
+      0.039854299277067184,
+      0.036494482308626175,
+      0.03535246104001999,
+      0.03631899505853653,
+      0.052550867199897766,
+      0.03643118590116501,
+      0.04741375148296356,
+      0.05410199239850044,
+      0.03729458898305893,
+      0.047319523990154266,
+      0.05005549639463425,
+      0.045176032930612564,
+      0.0343095101416111,
+      0.034685175865888596,
+      0.0549418181180954,
+      0.03940516710281372,
+      0.06138772889971733,
+      0.026964673772454262,
+      0.061726365238428116,
+      0.035674963146448135,
+      0.03675725311040878,
+      0.04917548596858978,
+      0.02920985221862793,
+      0.06452488899230957,
+      0.04665427654981613,
+      0.033418770879507065,
+      0.04070022702217102,
+      0.047383278608322144,
+      0.05476439371705055,
+      0.03517329692840576,
+      0.04526129364967346,
+      0.03768059238791466,
+      0.04081333428621292,
+      0.05012888461351395,
+      0.03764863312244415,
+      0.04876726120710373,
+      0.042848192155361176,
+      0.05155747011303902,
+      0.048097509890794754,
+      0.04292648285627365,
+      0.03304499387741089,
+      0.03619405999779701,
+      0.029800329357385635,
+      0.0334189310669899,
+      0.03642233833670616,
+      0.061966873705387115,
+      0.03105630725622177,
+      0.052772752940654755,
+      0.035264141857624054,
+      0.045341137796640396,
+      0.038955844938755035,
+      0.0272893775254488,
+      0.03914085775613785,
+      0.044834159314632416,
+      0.03508982062339783,
+      0.03300919011235237,
+      0.045659542083740234,
+      0.03922303766012192,
+      0.04003182426095009,
+      0.030378874391317368,
+      0.03925931081175804,
+      0.03871184587478638,
+      0.04067009687423706,
+      0.03241518884897232,
+      0.04611578211188316,
+      0.040712304413318634,
+      0.04804009944200516,
+      0.04274597764015198,
+      0.02574107050895691,
+      0.05612611025571823,
+      0.05290276184678078,
+      0.031578272581100464,
+      0.040402136743068695,
+      0.04238107055425644,
+      0.03174016997218132,
+      0.037568703293800354,
+      0.025163792073726654,
+      0.043820884078741074,
+      0.031618956476449966,
+      0.03853407874703407,
+      0.03542154282331467,
+      0.03759405016899109,
+      0.04508199915289879,
+      0.044700928032398224,
+      0.03423428535461426,
+      0.05527113378047943,
+      0.034653205424547195,
+      0.03451674059033394,
+      0.04219019412994385,
+      0.04577811062335968,
+      0.042808786034584045,
+      0.025785747915506363,
+      0.05958080291748047,
+      0.05169958621263504,
+      0.05500699207186699,
+      0.03033105656504631,
+      0.04161223769187927,
+      0.041003815829753876,
+      0.044352687895298004,
+      0.022202085703611374,
+      0.04517877846956253,
+      0.04168744385242462,
+      0.03853369504213333,
+      0.04069175571203232,
+      0.04503012076020241,
+      0.06120455637574196,
+      0.03805052489042282,
+      0.04385353624820709,
+      0.03158339485526085,
+      0.056811150163412094,
+      0.03625018894672394,
+      0.041450440883636475,
+      0.05121373385190964,
+      0.040211036801338196
+    ],
+    "denoise_loss": [],
+    "ortho_loss": [
+      0.8477625250816345,
+      0.6270668506622314,
+      0.35375306010246277,
+      0.18513090908527374,
+      0.11314814537763596,
+      0.08155959099531174,
+      0.06973419338464737,
+      0.05319012328982353,
+      0.0419590063393116,
+      0.03797777369618416,
+      0.03470993414521217,
+      0.03375212103128433,
+      0.03749161586165428,
+      0.04121977835893631,
+      0.04217546805739403,
+      0.046706654131412506,
+      0.04616285860538483,
+      0.045632317662239075,
+      0.04785409942269325,
+      0.050589628517627716,
+      0.05486857146024704,
+      0.05801132321357727,
+      0.06313689053058624,
+      0.06832898408174515,
+      0.07220464199781418,
+      0.07595375180244446,
+      0.07965358346700668,
+      0.0843292623758316,
+      0.08921559900045395,
+      0.09208281338214874,
+      0.0972515195608139,
+      0.10153611749410629,
+      0.10566537082195282,
+      0.10940378159284592,
+      0.11085320264101028,
+      0.11395293474197388,
+      0.11536199599504471,
+      0.11507408320903778,
+      0.11330236494541168,
+      0.11313570290803909,
+      0.11312603950500488,
+      0.11352822929620743,
+      0.11217745393514633,
+      0.11345779895782471,
+      0.11377058178186417,
+      0.11214352399110794,
+      0.11296330392360687,
+      0.11283531785011292,
+      0.11142794042825699,
+      0.10957454890012741,
+      0.10951845347881317,
+      0.10844650119543076,
+      0.10524056851863861,
+      0.10556697845458984,
+      0.10653087496757507,
+      0.10563769191503525,
+      0.10426626354455948,
+      0.10393765568733215,
+      0.10254596173763275,
+      0.10268296301364899,
+      0.10323475301265717,
+      0.10351748019456863,
+      0.10375725477933884,
+      0.10098842531442642,
+      0.10027060657739639,
+      0.09939239919185638,
+      0.0979134738445282,
+      0.09800726175308228,
+      0.09840735793113708,
+      0.09692982584238052,
+      0.09581397473812103,
+      0.09529635310173035,
+      0.09658239781856537,
+      0.09718240797519684,
+      0.09706336259841919,
+      0.09599632769823074,
+      0.09735315293073654,
+      0.09656201303005219,
+      0.09889460355043411,
+      0.10040658712387085,
+      0.10073850303888321,
+      0.10102172195911407,
+      0.1009986400604248,
+      0.10244335234165192,
+      0.1031949371099472,
+      0.10399853438138962,
+      0.10671082139015198,
+      0.1071852445602417,
+      0.10714156180620193,
+      0.1065908893942833,
+      0.1059853732585907,
+      0.1068815365433693,
+      0.10740302503108978,
+      0.10792513191699982,
+      0.10870114713907242,
+      0.108823761343956,
+      0.10967440903186798,
+      0.11261012405157089,
+      0.11379697918891907,
+      0.11422721296548843,
+      0.11453976482152939,
+      0.11541109532117844,
+      0.11571750789880753,
+      0.11609385162591934,
+      0.11757729947566986,
+      0.11700119078159332,
+      0.11970614641904831,
+      0.11980172991752625,
+      0.12358936667442322,
+      0.12539692223072052,
+      0.12486863881349564,
+      0.1271524429321289,
+      0.12923555076122284,
+      0.1315857619047165,
+      0.13292205333709717,
+      0.13546587526798248,
+      0.13885003328323364,
+      0.14402534067630768,
+      0.147845059633255,
+      0.15124990046024323,
+      0.15468817949295044,
+      0.1559913456439972,
+      0.1601400226354599,
+      0.16294769942760468,
+      0.16298216581344604,
+      0.16452212631702423,
+      0.1683468520641327,
+      0.16868238151073456,
+      0.17160287499427795,
+      0.17252276837825775,
+      0.17348770797252655,
+      0.17566640675067902,
+      0.17623268067836761,
+      0.17837531864643097,
+      0.17908938229084015,
+      0.18104304373264313,
+      0.18217481672763824,
+      0.18356861174106598,
+      0.18465754389762878,
+      0.18620507419109344,
+      0.18759240210056305,
+      0.18777203559875488,
+      0.19006824493408203,
+      0.19228748977184296,
+      0.19235096871852875,
+      0.1922677904367447,
+      0.19406947493553162,
+      0.19693200290203094,
+      0.19788794219493866,
+      0.19815294444561005,
+      0.1983489841222763,
+      0.20040495693683624,
+      0.20181000232696533,
+      0.2037876695394516,
+      0.20417653024196625,
+      0.20466746389865875,
+      0.20467495918273926,
+      0.20559166371822357,
+      0.20517095923423767,
+      0.2047836035490036,
+      0.20574744045734406,
+      0.2071981132030487,
+      0.2092108130455017,
+      0.20778527855873108,
+      0.20843109488487244,
+      0.21042220294475555,
+      0.21265923976898193,
+      0.2137930691242218,
+      0.21534432470798492,
+      0.21459147334098816,
+      0.21428824961185455,
+      0.21586327254772186,
+      0.21655382215976715,
+      0.21775919198989868,
+      0.21906070411205292,
+      0.2186809629201889,
+      0.2202470302581787,
+      0.22138795256614685,
+      0.22406847774982452,
+      0.22421269118785858,
+      0.22502855956554413,
+      0.22724615037441254,
+      0.22776556015014648,
+      0.22870388627052307,
+      0.2294059842824936,
+      0.23032227158546448,
+      0.23069751262664795,
+      0.23065929114818573,
+      0.23228758573532104,
+      0.23362798988819122,
+      0.2354430854320526,
+      0.2359708994626999,
+      0.2362011820077896,
+      0.23547999560832977,
+      0.236810564994812,
+      0.2378942370414734,
+      0.2370852380990982,
+      0.23747660219669342,
+      0.2379510998725891,
+      0.23851118981838226,
+      0.23896045982837677,
+      0.23781299591064453,
+      0.23904746770858765,
+      0.2397538423538208,
+      0.2394849956035614,
+      0.24049346148967743,
+      0.2408442348241806,
+      0.24402427673339844,
+      0.24560561776161194,
+      0.24606594443321228,
+      0.24576707184314728,
+      0.24569669365882874,
+      0.24569936096668243,
+      0.2463708519935608,
+      0.2471439093351364,
+      0.2489568144083023,
+      0.2494373470544815,
+      0.24941569566726685,
+      0.25020310282707214,
+      0.25147417187690735,
+      0.25302577018737793,
+      0.25658562779426575,
+      0.25757506489753723,
+      0.25599151849746704,
+      0.2575584948062897,
+      0.2580026090145111,
+      0.25835859775543213,
+      0.2620808482170105,
+      0.2628181278705597,
+      0.26342225074768066,
+      0.2643716633319855,
+      0.2652975916862488,
+      0.26612958312034607,
+      0.26860105991363525,
+      0.2692892253398895,
+      0.27039727568626404,
+      0.26999226212501526,
+      0.2700327932834625,
+      0.27080225944519043,
+      0.2718280851840973,
+      0.2721576988697052,
+      0.27287086844444275,
+      0.27328673005104065,
+      0.2742519974708557,
+      0.2743457555770874,
+      0.27543798089027405,
+      0.27694734930992126,
+      0.2783535122871399,
+      0.2793116271495819,
+      0.2808689475059509,
+      0.2817002236843109,
+      0.2811025381088257,
+      0.2815658748149872,
+      0.2821238934993744,
+      0.2833704948425293,
+      0.2831267714500427,
+      0.2842507064342499,
+      0.2866370379924774,
+      0.28746598958969116,
+      0.28841838240623474,
+      0.2894956171512604,
+      0.2908859848976135,
+      0.2916127145290375,
+      0.2926768660545349,
+      0.2931440770626068,
+      0.293007493019104,
+      0.2934287190437317,
+      0.29440173506736755,
+      0.29519739747047424,
+      0.2955481708049774,
+      0.29591187834739685,
+      0.29667699337005615,
+      0.29854315519332886,
+      0.2990916967391968,
+      0.2992062568664551,
+      0.3002597391605377,
+      0.29994961619377136,
+      0.300370454788208,
+      0.3015492856502533,
+      0.30222153663635254,
+      0.3022262156009674,
+      0.30287307500839233,
+      0.3034624457359314,
+      0.30395182967185974,
+      0.3042353689670563,
+      0.3045642673969269,
+      0.3047126531600952,
+      0.30494335293769836,
+      0.30510443449020386,
+      0.3058646023273468,
+      0.3064851760864258,
+      0.3073820173740387,
+      0.307350218296051,
+      0.3077446520328522,
+      0.30821216106414795,
+      0.30836403369903564,
+      0.30856674909591675,
+      0.30873575806617737,
+      0.30962589383125305,
+      0.3099839389324188,
+      0.3102879524230957,
+      0.31076711416244507,
+      0.31080421805381775,
+      0.31103259325027466,
+      0.3113178312778473,
+      0.31175798177719116,
+      0.31192323565483093,
+      0.31196096539497375,
+      0.3120540380477905,
+      0.3122491240501404
+    ],
+    "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": [
+      0.61,
+      0.91,
+      0.97,
+      1.0,
+      0.97,
+      1.0,
+      0.99,
+      1.0,
+      0.99,
+      1.0
+    ]
+  },
+  "final_accuracy": 0.9995833333333334,
+  "sft_eval": {
+    "config": {
+      "ops": "add_sub",
+      "K": null,
+      "mode": "sft",
+      "n_digits": 6,
+      "n_per_split": 100
+    },
+    "splits": {
+      "add_S0": {
+        "full_accuracy": 0.93,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9884297520661157,
+            "count": 605
+          },
+          "SS": {
+            "accuracy": 0.9894736842105263,
+            "count": 95
+          }
+        }
+      },
+      "add_S1": {
+        "full_accuracy": 0.84,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9803921568627451,
+            "count": 204
+          },
+          "SC": {
+            "accuracy": 0.9881656804733728,
+            "count": 169
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 31
+          },
+          "UC": {
+            "accuracy": 0.9628378378378378,
+            "count": 296
+          }
+        }
+      },
+      "add_S2": {
+        "full_accuracy": 0.63,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9754601226993865,
+            "count": 163
+          },
+          "SC": {
+            "accuracy": 0.9538461538461539,
+            "count": 130
+          },
+          "SS": {
+            "accuracy": 0.896551724137931,
+            "count": 87
+          },
+          "UC": {
+            "accuracy": 0.8719211822660099,
+            "count": 203
+          },
+          "US": {
+            "accuracy": 0.9487179487179487,
+            "count": 117
+          }
+        }
+      },
+      "add_S3": {
+        "full_accuracy": 0.46,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9834710743801653,
+            "count": 121
+          },
+          "SC": {
+            "accuracy": 0.9834710743801653,
+            "count": 121
+          },
+          "SS": {
+            "accuracy": 0.9795918367346939,
+            "count": 49
+          },
+          "UC": {
+            "accuracy": 0.7311827956989247,
+            "count": 186
+          },
+          "US": {
+            "accuracy": 0.8475336322869955,
+            "count": 223
+          }
+        }
+      },
+      "add_S4": {
+        "full_accuracy": 0.34,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9903846153846154,
+            "count": 104
+          },
+          "SC": {
+            "accuracy": 0.9716981132075472,
+            "count": 106
+          },
+          "SS": {
+            "accuracy": 1.0,
+            "count": 23
+          },
+          "UC": {
+            "accuracy": 0.65625,
+            "count": 160
+          },
+          "US": {
+            "accuracy": 0.5830618892508144,
+            "count": 307
+          }
+        }
+      },
+      "add_S5": {
+        "full_accuracy": 0.4,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 0.96,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.56,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.575,
+            "count": 400
+          }
+        }
+      },
+      "add_S6": {
+        "full_accuracy": 0.58,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.64,
+            "count": 100
+          },
+          "US": {
+            "accuracy": 0.674,
+            "count": 500
+          }
+        }
+      },
+      "add_random": {
+        "full_accuracy": 0.825,
+        "n_examples": 200,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.9798657718120806,
+            "count": 447
+          },
+          "SC": {
+            "accuracy": 0.984375,
+            "count": 320
+          },
+          "SS": {
+            "accuracy": 0.9642857142857143,
+            "count": 56
+          },
+          "UC": {
+            "accuracy": 0.9565217391304348,
+            "count": 529
+          },
+          "US": {
+            "accuracy": 0.9166666666666666,
+            "count": 48
+          }
+        }
+      },
+      "add_C3": {
+        "full_accuracy": 0.54,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.96,
+            "count": 300
+          },
+          "SC": {
+            "accuracy": 0.98,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.7979274611398963,
+            "count": 193
+          },
+          "US": {
+            "accuracy": 0.794392523364486,
+            "count": 107
+          }
+        }
+      },
+      "add_C4": {
+        "full_accuracy": 0.5,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 0.99,
+            "count": 200
+          },
+          "SC": {
+            "accuracy": 0.95,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.83203125,
+            "count": 256
+          },
+          "US": {
+            "accuracy": 0.75,
+            "count": 144
+          }
+        }
+      },
+      "add_C5": {
+        "full_accuracy": 0.45,
+        "n_examples": 100,
+        "per_subtask": {
+          "SA": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "SC": {
+            "accuracy": 0.93,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8202614379084967,
+            "count": 306
+          },
+          "US": {
+            "accuracy": 0.7835051546391752,
+            "count": 194
+          }
+        }
+      },
+      "add_C6": {
+        "full_accuracy": 0.48,
+        "n_examples": 100,
+        "per_subtask": {
+          "SC": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UC": {
+            "accuracy": 0.8524590163934426,
+            "count": 366
+          },
+          "US": {
+            "accuracy": 0.8632478632478633,
+            "count": 234
+          }
+        }
+      },
+      "sub_M0": {
+        "full_accuracy": 0.82,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9700499168053245,
+            "count": 601
+          },
+          "ME": {
+            "accuracy": 0.9292929292929293,
+            "count": 99
+          }
+        }
+      },
+      "sub_M1": {
+        "full_accuracy": 0.86,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.982078853046595,
+            "count": 279
+          },
+          "MB": {
+            "accuracy": 0.9724137931034482,
+            "count": 145
+          },
+          "ME": {
+            "accuracy": 0.9166666666666666,
+            "count": 24
+          },
+          "UB": {
+            "accuracy": 0.9841269841269841,
+            "count": 252
+          }
+        }
+      },
+      "sub_M2": {
+        "full_accuracy": 0.7,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9906103286384976,
+            "count": 213
+          },
+          "MB": {
+            "accuracy": 0.9823008849557522,
+            "count": 113
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 85
+          },
+          "UB": {
+            "accuracy": 0.850828729281768,
+            "count": 181
+          },
+          "UD": {
+            "accuracy": 0.9814814814814815,
+            "count": 108
+          }
+        }
+      },
+      "sub_M3": {
+        "full_accuracy": 0.4,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 0.970873786407767,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 0.9821428571428571,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.697986577181208,
+            "count": 149
+          },
+          "UD": {
+            "accuracy": 0.7981220657276995,
+            "count": 213
+          }
+        }
+      },
+      "sub_M4": {
+        "full_accuracy": 0.37,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 0.98,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.59,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.6866666666666666,
+            "count": 300
+          }
+        }
+      },
+      "sub_M5": {
+        "full_accuracy": 0.55,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.63,
+            "count": 100
+          },
+          "UD": {
+            "accuracy": 0.67,
+            "count": 400
+          }
+        }
+      },
+      "sub_random": {
+        "full_accuracy": 0.81,
+        "n_examples": 200,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.9833333333333333,
+            "count": 600
+          },
+          "MB": {
+            "accuracy": 0.9700374531835206,
+            "count": 267
+          },
+          "ME": {
+            "accuracy": 0.9622641509433962,
+            "count": 53
+          },
+          "UB": {
+            "accuracy": 0.9567198177676538,
+            "count": 439
+          },
+          "UD": {
+            "accuracy": 0.9512195121951219,
+            "count": 41
+          }
+        }
+      },
+      "sub_B3": {
+        "full_accuracy": 0.71,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 0.99,
+            "count": 300
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8934010152284264,
+            "count": 197
+          },
+          "UD": {
+            "accuracy": 0.7864077669902912,
+            "count": 103
+          }
+        }
+      },
+      "sub_B4": {
+        "full_accuracy": 0.58,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 200
+          },
+          "MB": {
+            "accuracy": 0.97,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8825910931174089,
+            "count": 247
+          },
+          "UD": {
+            "accuracy": 0.7450980392156863,
+            "count": 153
+          }
+        }
+      },
+      "sub_B5": {
+        "full_accuracy": 0.57,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 100
+          },
+          "UB": {
+            "accuracy": 0.8590604026845637,
+            "count": 298
+          },
+          "UD": {
+            "accuracy": 0.8415841584158416,
+            "count": 202
+          }
+        }
+      }
+    },
+    "summary": {
+      "overall_accuracy": 0.6229166666666667,
+      "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": 0.99,
+        "n_examples": 100,
+        "per_subtask": {
+          "MD": {
+            "accuracy": 1.0,
+            "count": 179
+          },
+          "MB": {
+            "accuracy": 1.0,
+            "count": 103
+          },
+          "ME": {
+            "accuracy": 1.0,
+            "count": 56
+          },
+          "UB": {
+            "accuracy": 0.9932885906040269,
+            "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": 0.9995833333333334,
+      "total_examples": 2400,
+      "n_splits": 22
+    }
+  },
+  "sorl_overall_accuracy": 0.9995833333333334,
+  "sft_overall_accuracy": 0.6229166666666667
+}

add_sub_sorl_v1_abs100_100K/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e4b24a426ca15853e23168908b5f1d34be3e3a3625865c28031bb24ca1a1faf
+size 650671040

add_sub_sorl_v1_abs100_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_abs100_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": 100,
+  "dataset_size": 100000,
+  "mode": "sorl",
+  "device": "cuda",
+  "push_to_hub": true,
+  "no_wandb": false,
+  "n_params": 162591062,
+  "run_name": "add_sub_sorl_v1_abs100_100K",
+  "git_commit": "f447da529caceac8c7d256cbb2cd185cbc50feac",
+  "timestamp": "2026-04-12T15:51:04.267579+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": "e484ym4g",
+  "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/e484ym4g",
+  "final_accuracy": 0.9995833333333334,
+  "sft_accuracy": 0.6229166666666667,
+  "eval_method": "ArithmeticEvaluator"
+}